Innervated Myotendinous Cylinders in Human Extraocular Muscles

Abstract

purpose. To analyze palisade endings and their end organs, the so-called innervated myotendinous cylinders (IMCs), of human extraocular muscle (EOM) in more detail and to clarify with the help of double-fluorescent labeling and electron microscopy whether terminals in IMCs are sensory, serving proprioception.

methods. EOMs obtained from a donated cadaver (66 years) and distal parts of EOMs from multiorgan donors (35, 53 years) were processed for double-fluorescent labeling. Antibodies against the protein gene product 9.5 and α-bungarotoxin labeling were used on cryostat sections of distal myotendons. EOMs from multiorgan donors (2, 17 years) were prepared for electron microscopy.

results. Palisade endings investing muscle fiber tips established contacts with tendon fibrils and the muscle fiber attached. α-Bungarotoxin bound to myoneural contacts but not to axonal varicosities in the tendon compartment. Ultrastructural analysis revealed that palisade endings form IMCs, which were associated exclusively with multiply innervated global layer muscle fibers. IMCs consisted of a muscle fiber tendon junction, tightly enclosed by fibrocytes, and a supplying axon with preterminals and terminals. Terminals contained mitochondria, few neurotubuli, few neurofilaments, and accumulations of clear vesicles of uniform size. A basal lamina always intervened between axolemma and tendon fibrils as well as between axolemma and muscle fiber cell membrane.

conclusions. Palisade endings of human EOM form IMCs as in cat, monkey, and sheep. In contrast to animals, myoneural contacts in human IMCs are almost certainly motor, whereas terminals contacting tendon fibrils are arguably sensory. Thus, IMCs might be best described as“ propriocept-effectors.”

Increasing importance is attributed to the proprioceptive innervation of extraocular muscles (EOMs), in particular for the development of binocular vision. ^{1 2 3 4} Electrophysiological and neuronal tracing studies in animals demonstrated EOM proprioceptive input to most central nervous system (CNS) regions involved in vision and oculomotor control. ³ This clearly suggests activity of proprioceptors in EOMs, ^{1 2 3 4 5 6 7} but their exact role is still far from clear, necessitating more detailed analysis. 

Compared with other skeletal muscles, the proprioceptive complement of EOM is unique and exhibits striking interspecies differences. Previously, we demonstrated that human EOM is richly supplied with specifically structured muscle spindles. ^{5 6 7} Golgi tendon organs (GTOs) were reported to be an exceptional element in human EOM. ⁸ So far, the occurrence of GTOs has been reported only in rhesus monkey and sheep EOM. ^{9 10 11} GTOs in EOM differ morphologically from their counterparts in other skeletal muscles. ^{9 10 11} Another putative receptor usually ascribed to the proprioceptive complement of EOM in humans, monkey, and cat, are the so-called “palisade endings.” First described in silver-stained material in the early years of the last century ¹² and located in more or less high numbers at EOM myotendon junctions, palisade endings typically consist of a dense, caplike ramification of an axon branch investing the tip of a single muscle fiber. The existence of these EOM-specific nerve endings was later confirmed by several authors in animals ^{13 14 15} and more recently by Richmond et al. ¹⁶ in human EOM. 

The encapsulated nervous end organ–containing palisade endings were, in their fine structure, first described in the cat, ¹⁴ and, in an almost coincident study on monkey EOM, ¹⁵ were called “innervated myotendinous cylinder” (IMC). IMCs, ^{14 15} recently also discovered and described in ultrastructure in sheep EOM, ¹⁷ differ significantly from classical GTOs ¹⁸ as well as from GTOs of monkey ⁹ and sheep EOM. ^{10 11} IMCs are enclosed by a loose capsule of flat connective tissue cells and contain a single muscle fiber tendon junction. They are innervated by the unidirectional palisade-like terminal arborization of the branch of a small myelinated nerve fiber. Within IMCs, terminals establish synaptic contacts to both the muscle fiber surface and to collagen fibrils. In contrast, GTOs are in toto buried in tendon. ¹⁸ They are enclosed by perineurial capsules of several cell layers all invested by basal lamina. ^{9 10 11 18} Within a subcapsular fluid space, GTOs contain tendon fibrils that are attached to the tips of more or less muscle fibers outside the capsule. ^{9 10 11 18} The perineurium of the entering myelinated nerve fiber is continuous with the capsule. The nerve fiber arborizes bidirectionally into a number of terminals that exclusively contact collagen fibrils. ^{9 10 11 18}  

The situation toward IMCs in humans is unclear. Because of their localization at EOM tendon junctions, IMCs are of particular interest for strabismus surgery. However, their exact morphology has attracted little scientific interest. Sodi et al. ¹⁹ presented the ultrastructure of nerve endings lying close to muscle fibers at their myotendinous junction. Occasionally, terminals were partly enclosed by a fibrous capsule. ¹⁹ In the myotendinous region, Bruenech and Ruskell ²⁰ and Ruskell ²¹ noted a few small colonies of unencapsulated terminals in adult EOMs; however, they did not find any terminals in postmortem myotendons of children. So far, it has remained unclear whether palisade endings form IMCs in human EOMs. In our laboratory, myotendinous cylinders were studied in EOMs obtained from several species, including humans, cat, monkey, sheep, ¹⁷ and rabbit (Blumer R, Wasicky R, Hoetzenecker W, Mayr R, Lukas JR, unpublished results), to assess interspecies variation. IMCs are generally regarded as proprioceptors, ^{14 15 17 21} although their sensory function has also been doubted. ^{14 22 23} Until now, significant functional evidence is still missing. It was the aim of the present study to investigate palisade endings of human EOM in more detail and to clarify, with the help of double-fluorescent staining and electron microscopy, whether these nerve endings form IMCs in humans and whether their terminal contacts exhibit morphologic characteristics of sensory and/or motor nerve endings. 

Materials and Methods

The material used in this study consisted of all EOMs dissected at full length 5 to 24 hours postmortem from each of four body donors to the Institutes of Anatomy of Vienna (67, 72, 84 years of age) and of Erlangen (66 years of age). Further, distal parts of human EOMs were removed from the globes of four multiorgan donors (2, 17, 35, 53 years of age) a few hours postmortem. Methods for securing human tissues were humane, included proper consent and approval, and complied with the tenets of the Declaration of Helsinki and the Austrian federal transplantation law. 

All 12 EOMs of a female body donor (66 years) were collected for double-fluorescent staining of nerve fibers and motor nerve endings. Five hours postmortem, the skull was opened, and after removal of the brain, both orbits were fixed by perfusion through the carotid arteries with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) and remained in the same fixative for 4 hours. EOMs dissected in toto were measured and divided into five parts, each approximately 10 mm long. Thereafter, all parts were frozen in isopentane cooled in dry ice. Proximal and distal parts were processed for double-fluorescent staining. Likewise, distal parts of EOMs, including myotendons from the 35- and 53-year-old multiorgan donors, were immediately frozen as described above and used for double-fluorescent labeling. 

Complete transverse serial sections of all EOMs of three body donors (67, 72, 83 years of age), used earlier to analyze muscles spindles, ⁵ served to study the course and distribution of nervous structures in the myotendinous region. Detailed protocols of sectioning and staining (Mowry staining; silver impregnation ²⁴ ) were published previously. ⁵  

EOM pieces of the 2- and 17-year-old donors were prepared for electron microscopy. Material of the 2-year-old donor had already been used to study the ultrastructure of human EOM spindles in infancy. ²⁵  

Double-Fluorescent Staining

Longitudinal cryostat serial sections of human EOM tissue blocks containing the proximal or distal myotendon were cut at 10 μm and mounted on silane-coated slides. Sections were rinsed in Tris-buffered saline (TBS) two times (5 minutes each) and transferred into a 1% solution of swine serum in TBS. Commercially available antiserum (polyclonal rabbit anti-human Ig) against the protein gene product (PGP) 9.5 (Ultraclone, Isle of Wight, UK, and Biogenesis, Poole, Dorset, UK) was reconstituted with 50 μl distilled water. The diluted (1:500) primary antiserum was applied on to the slide, and sections were incubated in a moist chamber under conditions of near darkness at 21°C overnight. Sections were rinsed four times in TBS for 60 minutes and transferred to a 1:40 dilution of FITC-linked swine anti-rabbit second antibody (Dako, Glostrup, Denmark), followed by rinsing three times in TBS (5 minutes each). Rhodamine-labeled,α -bungarotoxin (α-BT, tetramethylrhodamine-α-bungarotoxin; Molecular Probes, Eugene, OR) was reconstituted with 0.5 ml distilled water and diluted in TBS (1:100). Sections were incubated under the above-described conditions for 5 minutes. After rinsing two times in TBS, slides were mounted in Aquatex (Merck, Munich, Germany) or glycerol/TBS (1:1, pH 8.6). In controls, the primary antibody was omitted, and FITC-linked second antibody was used alone. Sections were immediately viewed and photodocumented using an epifluorescence microscope (Axioskop; Zeiss, Oberkochen, Germany) equipped with a rhodamine filter (filter set 09; Zeiss) and a FITC filter set (filter set 15; Zeiss), using oil immersion objectives (40×, 100×) with high aperture (1.3). The additional use of an interference red barrier filter that extinguished all red light enabled an unambiguous differentiation between rhodamine and FITC fluorescence. ²⁶ To assess the validity of the double-fluorescent staining used for the identification of motor nerve endings, all types of myoneural synapses of human EOM were also studied by confocal laser scanning microscopy (Bio-Rad MRC 1000, Munich, Germany). 

Electron Microsopy

Distal parts of human EOMs were cut longitudinally into four stripes, each 6 mm in length and 3 mm in diameter. Tissue blocks were immersion fixed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer at pH 7.4 for several days. After rinsing in the same buffer, specimens were postfixed in 1% osmium tetroxide in cacodylate buffer, dehydrated in graded solutions of ethanol, and embedded in epon. 

Proceeding from tendon to muscle, each block was scanned for the presence of myelinated nerve fibers, light microscopically analyzing step series of semithin cross sections (1 μm) stained with toluidine blue. As soon as a nerve fiber or a myotendinous cylinder appeared within the tendon, serial semithin sections were examined. In case a nerve fiber entered a myotendinous cylinder, ultrathin sections alternating with semithin ones were cut at appropriate intervals, mounted on dioxane formvar-coated copper grids, immersed in an aqueous solution of 2% uranyl acetate followed by a solution of 0.4%lead citrate in 0.1 M sodium hydroxide, and examined under a Zeiss EM10 transmission electron microscope. IMCs were investigated in complete serial sections from their distal to their proximal ends. In further transverse serial semithin sections, the course of their supplying nerve fibers was traced back as far as possible toward the EOM belly. 

Results

Light Microscopy

Identification of Motor Nerve Endings by Double-Fluorescent Labeling.

PGP 9.5–positive nerve fibers were visible within all parts of the specimens investigated and exhibited a clear green fluorescence viewed with the FITC filter set. α-BT–stained structures fluoresced red viewed with the rhodamine filter and orange-green using the FITC filter set. As unambiguously confirmed in the confocal laser scanning microscope, α-BT staining was confined to the postsynaptic membranes of motor nerve endings of both “en plaque” and “en grappe” types (Fig. 1) . The double labeling applied enabled a clear differentiation between sensory and motor axon terminals at the level of the light microscope, simply by switching between filter positions. In the laser scanning microscope, this methodological approach allowed discrimination of fine structural details of “en plaque” and “en grappe” motor nerve endings and studying of the terminal course of the supplying motor nerve fibers (Fig. 1) . The decisive criterion for the identification of a multiply innervated muscle fiber (MIF) in longitudinal sections (Fig. 1A) was the demonstration of a single PGP 9.5—positive nerve fiber accompanying a muscle fiber and establishing multiple smallα -BT–stained myoneural contacts up to its tendon junction. The diameter of nerve fibers supplying MIFs varied between 2.5 and 4.5μ m. Motor end plates of singly innervated muscle fibers (SIFs) of human EOM were supplied by a PGP 9.5–positive nerve fiber branch of larger size. It terminated in a dense array of synaptic knobs at a well-developed sole plate of a SIF, which exhibited α-BT–stained synaptic troughs (Figs. 1B 1C) . 

Palisade Endings.

In double-fluorescently labeled longitudinal sections of human EOM, single PGP 9.5–immunoreactive axons were observed to penetrate the tendon and, after running forward in the tendon, to turn back to the muscle tendon junction. At this site, each axon ramified into several branches directed toward the muscle, mainly in parallel with its longitudinal axis, thereby surrounding a single muscle fiber tip (Fig. 2) . Alternatively, an axon ramified at the muscle fiber tip without previously turning back from the tendon. Axon branches exhibited multiple varicosities within the tendon compartment, obviously establishing contacts with tendon fibrils (Fig. 2) . Axon terminals also were found in contact with the muscle fiber. In many instances, these myoneural contacts exhibited α-BT staining (Fig. 2) . Even under high-power oil immersion objectives, we were not able to detect anyα -BT staining within the tendon (Fig. 2) . 

Palisade endings were sheeted by thin capsules to constitute IMCs. These capsules were difficult to identify in immunohistochemically stained, longitudinal sections. Within the tendon compartment of each IMC, the arborization of only one single axon was observed. It is important to note that myoneural synapses exhibiting the morphology of“ en grappe” motor nerve endings were observed at the same muscle fiber in close proximity to the palisade ending but evidently outside its encapsulation. A relatively close affinity of palisade endings to MIFs seemed evident, but, admittedly, a precise identification of this type of muscle fiber was not possible in each palisade ending. 

Although less conspicuous and less numerous than in cat EOMs, palisade endings were observed to be regularly present both in the proximal and in the distal myotendons of human EOMs. Complete transverse serial sections of all EOMs of three aged persons, used in an earlier study on human EOM spindles, ⁵ had been alternately treated by silver impregnation ²⁴ and ferric oxide staining (after Mowry ⁵ ) to study nervous structures at the myotendinous junction. Both myotendinous junctions of EOM are shaped like arcs with convexities directed toward the tendons, as shown in schematic drawings presented in Figure 4 of an earlier publication. ⁵ At the level of the distal myotendinous junction, mostly single nerve fibers were observed, whereas in each EOM three to four small nerves containing up to seven nerve fibers were regularly visible at more proximal levels. It is suggested that these nerves are the source of axons supplying palisade endings of IMCs. As there was no evidence from our material that these nerve fibers might branch to supply nerve endings of several IMCs, one could assume that, roughly estimated, between 20 and 30 IMCs would be present in the distal myotendon of a human EOM. Nevertheless, they were found to be not evenly distributed in different EOMs and might exhibit considerable interindividual differences in frequency. 

Ferric oxide staining (after Mowry ⁵ ) was used to demonstrate acidic mucopolysaccharides within encapsulated receptors. In Mowry-stained cross sections, we were not able to identify IMCs because of the obvious absence of mucopolysaccharides and because of their rather inconspicuous capsules. Despite a thorough and complete investigation of five proximal and a large number of distal EOM tendons in silver-impregnated and immunohistochemically stained sections, we were not able to identify any GTO. 

Electron Microscopy

Myotendinous Junction.

The tips of MIFs often thicken toward their distal ends and exhibit, as usual, deep invaginations of their sarcolemma where bundles of collagen fibrils are anchored by reticular fibrils within the glycocalix, securing the force transduction at the muscle fiber tendon junction. In many cases, the muscle fiber tendon junction of MIFs is ensheathed by connective tissue, forming a myotendinous cylinder. When such a cylinder is supplied by a nerve fiber arborizing inside, Ruskell ¹⁵ called it an “innervated myotendinous cylinder“ (Figs. 3 4) . All muscle fibers in the most distal parts of the myotendinous region exhibited the same ultrastructure. In particular, these muscle fibers showed rare mitochondria and a poorly developed sarcoplasmatic reticulum, indicating that they were MIFs of the global layer (Fig. 4) . 

IMC Capsule.

The capsule of each IMC consisted of one to three layers of flat processes of fibrocytes without any basal lamina investment. This capsule was discontinuous with the perineurium of the nerve fiber entering the IMC (Fig. 4) . Each capsule had the shape of a cylinder tightly enclosing a single muscle fiber tendon junction. In contrast to the perineurial capsules of muscle spindles, GTOs, and many other complex receptors, IMC connective tissue cell capsules were devoid of any conspicuous subcapsular fluid space. In the 2-year-old infant, IMCs exhibited complete capsules, whereas in the 17-year-old human, capsules were partly incomplete. 

Nerve Fibers Supplying IMCs.

A myelinated nerve fiber approached the tendon through muscle. Thereby, the nerve fiber was observed approximately 20 μm away from its target IMC. Finally, after looping in the tendon, the nerve fiber entered an IMC. Occasionally, the nerve fibers penetrated the cylinder capsule more directly, without previously turning back from the tendon. The caliber of myelinated nerve fibers supplying IMCs varied between 2.5 and 3.5 μm. After entering the cylinder, the nerve fiber lost its myelin sheath and ramified into preterminals exhibiting diameters of 1 to 2 μm. Preterminals intermingled with tendinous fibrils and approached the muscle fiber tip. Preterminals were completely surrounded by a Schwann cell and bore mitochondria, neurotubules, and neurofilaments (Fig. 4) . A basal lamina separated the Schwann cell investment from adjacent collagenous fibrils. Closer to the muscle fiber tip, preterminals partly lost their Schwann cell investment and became terminals. 

Frequently, terminals that were only partly surrounded or free from a Schwann cell investment established contacts with either connective tissue or the encapsulated part of the muscle fiber (Fig. 5) . Terminals intermingled with collagenous fibrils or were found within the finger-like intrusions of the muscle fiber tendon junction and also at the external sarcolemma surface. A basal lamina was always interposed between the axolemma devoid of Schwann cell investment and the collagen fibrils or the muscle fiber plasmalemma. These myoneural contacts were confined to the encapsulated cylinder. Neither the supplying nerve nor branches of it established myoneural contacts outside the IMC capsule. 

Terminals contacting collagen fibrils contained few neurofilaments, few neurotubuli, small mitochondria, and accumulations of clear vesicles of uniform size (Fig. 5) . In many instances, the terminal was almost completely filled with these vesicles. Close contacts between axolemma and collagen, always separated by a basal lamina, were frequently present. 

Terminals contacting the sarcolemma conformed morphologically with those contacting connective tissue (Fig. 5) . A basal lamina was present in each myoneural synaptic cleft, which measured 50 to 70 nm. In some terminals contacting the sarcolemma, accumulations of clear vesicles were associated with a presynaptic membrane thickening, but postsynaptic membranes were smooth. The particular morphology of these terminals resembled motor terminals of nearby MIFs, and even the diameter of their clear vesicles was comparable to those observed in myoneural synapses in focally innervated muscle fibers of the same EOM (Figs. 5C 5D) . Dense cored vesicles were not observed in terminals, neither in the tendon nor in the muscular IMC part. 

Discussion

Palisade endings of human EOM were demonstrated in this study to form the constituent nervous elements of encapsulated tissue cylinders, each containing a single muscle fiber tendon junction and the terminal arborization of the supplying myelinated nerve fiber. Ruskell ¹⁵ called these particular encapsulated nervous end organs of monkey EOM innervated myotendinous cylinders. In the present article, the occurrence and fine structure of IMCs of human EOM were described for the first time. Double-fluorescent labeling and electron microscopy gave convincing evidence that myoneural contacts within human IMCs are almost certainly motor. Axonal varicosities and terminals intermingled between collagen bundles in the tendinous part of IMCs are arguably sensory. In accordance with previous animal studies, ^{14 15 17} IMCs in humans were found to be associated with one particular muscle fiber type, the MIFs of the global layer. Similarities shown between human IMCs and those of other animals encourage hope that studies in other species might allow tentative conclusions about IMC function in humans. 

Since their description in silver-impregnated sections in humans ¹² in the early years of the last century, the morphology of human palisade endings has received little further attention. After the fine structural description of cat ¹⁴ and monkey IMCs, ¹⁵ a renewed interest in extraocular proprioception led to a new account on palisade endings, confirming their presence in human EOM by en bloc silver staining and silver impregnation of serial sections. ¹⁶ Succeeding reports on the fine structure of nerve endings at the myotendinous junction of human EOMs, ¹⁹ however, cannot be equated with those on IMC structure in animals. ^{14 15 17} Recently, Bruenech and Ruskell ²⁰ reported that in distal myotendons “obtained from infants aged 2 weeks to 4 years none contained nerve terminals whereas in adult material a few colonies of unencapsulated terminals were present—most areas of the tendons had none.” Thus, Ruskell ²¹ in his recent review denied the existence of IMCs in human EOM. He claimed that “myotendinous receptors may not be present at birth in man,” and on the basis of his observation that in mature muscle, “apart from a palisade form in some of the nerve endings, and the occasional partial fibrous enclosure, the terminals have an irregular form freely associated with the tendon fibers,” he also denied their role in proprioception. 

The present study casts very grave doubts on whether Ruskell ^{20 21} is correct. Nervous end organs at muscle fiber tendon junctions of human EOM were found in this study to conform in their substantial ultrastructural characteristics with those in animals ^{14 15 17} and are therefore to be classified as IMCs. As in these animals, human IMCs were enveloped by a connective tissue capsule and consisted of one single muscle fiber tendon junction and the supplying axon with its terminal arborization. In the present article, the demonstration of the multiple innervation of muscle fibers attached to IMCs and the ultrastructural particularities of these muscle fibers indicate that they are MIFs of the global layer. ^{27 28 29} This cannot completely compensate for the need of histochemical ^{27 29} and immunohistochemical evidence ²⁹ for EOM fiber classification. Rough estimations of the number of IMCs in human EOM also stress their importance for functional considerations in humans, even though they are less numerous than in animals. With respect to infancy, this study gave clear evidence for the presence of IMCs in distal myotendons of a 2-year-old human and that these IMCs were as well developed in structure as in the adult. In conclusion, the data of the present study help to resolve prior points of conflict between animal and human studies. 

Because, in any species, no one has yet recorded from afferents that can be assigned to IMCs, the nature of the signals provided by IMCs must be entirely speculative. The major argument to classify IMCs subjected to ultrastructural investigation, as sensory receptors was the morphology of their nerve terminals. Thus, the particular morphology of IMC axon terminals demands greater attention. In cat, ¹⁴ in monkey, ¹⁵ and recently discovered in sheep ¹⁷ IMCs, terminals frequently established close myoneural contacts lacking a basal lamina in their synaptic cleft. Likewise indicating the sensory nature of these terminals, attachment plaques were demonstrated in monkey IMC. ¹⁵ In cat ¹⁴ and sheep ¹⁷ IMCs, most terminals established contacts exclusively with tendon fibrils, in general with interposition of a basal lamina between axolemma and collagen. Terminals bore mitochondria, dense cored vesicles, and clear vesicles. Clusters of clear vesicles were often arranged close to membrane thickenings of the axolemma, “resembling active zones in chemical synapses.” ¹⁴ In both cat ¹⁴ and sheep, ¹⁷ such active zones were observed only in terminals contacting connective tissue, but never at the narrow myoneural synaptic clefts. Sodi et al. ¹⁹ described different nerve terminals in myotendons of human EOMs. Interestingly, these authors reported the rare presence of terminals with typical features of motor nerve endings. ¹⁹ In rabbit EOMs, which are richly endowed with IMCs, axon terminals were found to form almost exclusively myoneural synapses with typical fine structural features of motor terminals, among them a 50-nm synaptic cleft containing a distinct layer of basal lamina (Blumer et al., unpublished results). The present article demonstrated that, according to the tissue component contacted, human IMCs showed axon terminals in two fundamentally different locations. It is evident that mammalian and human IMCs, although exhibiting a common principle of their tissue composition, showed considerable species variations with regard to their proportional number of neurotendinous and myoneural terminals as well as of ultrastructurally motor- and/or sensory-like myoneural nerve endings. 

No previous article has presented a sensory-motor differentiation of palisade endings by histochemistry or fluorescently labeled acetylcholine receptors. Alvarado-Mallart and Pincon-Raimond ¹⁴ claimed that relevant information on palisade endings would be added when acetylcholinesterase staining and silver staining were combined. However, these authors did not succeed in establishing a combination of these staining methods. In the present article, the combination of immunofluorescence with α-BT staining was helpful to resolve the ambiguity inherent in light microscopy. PGP 9.5 immunofluorescence is generally regarded as highly sensitive to detect nerve tissue. ^{30 31} This cytoplasmatic neuronal protein was identified as an ubiquitin carboxyl-terminal hydrolase. ³² α-BT binds exclusively and with high affinity to postsynaptic nicotinic cholinergic receptors of muscle fibers. ³³ It is widely used specifically to detect and to analyze all types of motor nerve endings in vertebrate skeletal muscle. ³³ In conclusion, the double-fluorescent labeling used enabled a reliable discrimination of sensory and motor terminals, thereby establishing the dualistic functional concept of IMC nerve endings. Light microscopic results were in accordance with ultrastructural observations. Confocal laser scanning microscopy confirmed the validity of the double-fluorescent labeling used. 

In contrast with previous animal studies, ^{14 15 17} myoneural junctions of human IMCs exhibited a 50-nm synaptic cleft containing a continuous layer of basal lamina. The presence of a basal lamina within a 50-nm synaptic cleft is a characteristic feature of motor end plates, ^{13 28 34} whereas a 20-nm cleft free of basal lamina is typical of sensory myoneural contacts. ^{5 6 7 8 10 13 14 15 17 21 25 35} In sensory myoneural contacts, the terminal axon is covered by the basal lamina of the muscle fiber. ¹³ Presynaptic accumulations of clear vesicles and presynaptic membrane thickenings, also observed in IMCs of human EOMs, are further characteristics of motor end plates. ^{28 34} Attachment plaques have been found only in sensory endings. ³⁵ Although present in monkey IMCs, ¹⁵ attachment plaques were absent from myoneural contacts in human IMCs. The ultrastructure of myoneural junctions in human IMCs conformed with that of motor terminals contacting MIFs outside IMCs, including accumulations of clear vesicles of equal size in both axon terminals. However, it is important to note that, in contrast to GTOs, ^{9 11 18} terminals of neurotendinous contacts of IMCs exhibited similar accumulations of clear vesicles.α -BT staining provided direct and persuasive evidence for the presence of acetylcholinergic receptors in myoneural contacts in IMCs of human EOM. Applying these morphologic criteria, the majority of terminals contacting the tips of MIFs at their outer surface or in the depth of their sarcolemmal infoldings were classified as motor terminals. Contacts with collagen fibrils were likely to be sensory in nature, providing proprioceptive information. In conclusion, based on their morphology, IMCs are supposed to combine proprioceptor and effector qualities, and the hypothesis is put forward that they might function as “propriocept-effectors.” For clarification, physiological studies are warranted. 

Important support for a possible motor function of palisade endings is available from the literature. After intracranial section of the III, IV, and VI cranial nerves, Tozer and Sherington ³⁶ noted the degeneration of practically all nerves within monkey EOMs, including tendon endings that, represented in a drawing, look very similar to palisades. In a similar experiment in cats, which lack spindles and GTOs in their EOMs, Sas and Scháb ²² described the degeneration of all palisade endings, and after small stereotactic lesions in the oculomotor nuclei, they observed degenerated palisades in those muscles in which motor end plates also were degenerated. Therefore, they concluded that neurons innervating palisades must be motoneurons situated within the oculomotor nuclei and suggested that palisades are motor instead of receptor endings. That these authors, expecting twitches, failed to elicit such contractions after stimulation of an intramuscular nerve branch of cat inferior oblique known to consist solely of afferents to palisade endings cannot rule out a motor role of palisades. As stated above, palisades were found to innervate exclusively tips of global layer MIFs. MIFs resemble in structure slow tonic fibers of amphibian skeletal muscle ³⁷ and have been shown repeatedly to occur in mammalian and human EOM ^{27 28 29} since 1961. ³⁸ Physiological studies demonstrated the presence of non-twitch motor units in the global layer of rat and cat EOM. ^{39 40 41} Along these lines, it is suggested that stimulation of motor type palisade endings would result in slow, long-lasting local contractions confined to the ends of global layer MIFs, rendering extremely unimportant contributions to muscle pull, which remained undetected by Sas and Scháb. ²²  

The prevailing opinion that palisade endings are sensory is supported by a recent experimental study of Billig et al. ⁴² in the cat. After application of neuronal tracers into the trigeminal ganglion, four different types of nerve endings were labeled within EOMs, indicating their origin from trigeminal sensory neurons, one type conforming to palisade endings at the myotendon junction and three other types of previously unknown free nerve endings within the muscle belly. 

GTOs both in skeletal muscle and EOM differ principally from IMCs of EOM. In particular, the former are ensheathed by perineurium. ¹⁸ GTO terminals contacting tendon fibrils contain few mitochondria, neurotubules, neurofilaments, and few, if any, clear vesicles. ^{9 11 18} A basal lamina intervenes between GTO terminals and collagen fibrils. ¹⁸ Two conflicting theories are generally discussed in the literature focusing on IMCs. The particular morphology of IMCs might be a sign of structures retained in development ²² or could be the result of ongoing maturation. ¹⁵ During embryonic development, myoneural contacts are also present on muscle fiber tips in other skeletal muscles, but these contacts diminish shortly after birth. ⁴³ These formations are regarded as precursors of GTOs. Transient neuronal contacts with myotubes conformed morphologically with myoneural contacts in mature monkey IMCs. ^{15 43} Because no data are available on the special morphology of possible myoneural contacts in perinatal human EOMs, this important item awaits further clarification. No indications for ongoing maturation during life were observed in the present study when we compared IMCs in infant EOMs to those in EOMs obtained from older individuals. 

Functional Considerations

In a physiological study of extraocular afferent fibers, Cooper and Fillenz ⁴⁴ isolated a grouping of nonspontaneous, rapidly adapting responses with higher tension thresholds and ascribed these responses to unidentified stretch receptors in the tendon. Because GTOs are absent from cat EOM, palisade endings were supposed to be the source of these responses. A possible explanation for the unique occurrence of palisade endings in EOMs was that these stretch receptors are associated with slow fibers, the functional relationship being similar to that between GTOs and twitch muscles fibers. ¹³ Morphologic differences between IMCs and GTOs also indicate different functional properties. GTOs ¹⁸ like muscle spindles exhibit a more or less wide subcapsular fluid space under a multilayered outer perineurial capsule. ¹³ In contrast, human IMCs showed no obvious subcapsular space, indicating that the connective tissue capsule of IMCs is not as tightly closed. Thus, we would like to argue that the IMC capsule is likely not serving as a semipermeable diffusion barrier as muscle spindle and GTO capsules do. ¹³ Because the muscular IMC components are surrounded by other MIFs, IMCs are expected to behave differently on passive stretch versus active contraction of the muscle fiber attached. In particular, MIFs “are thought to be secure from direct collagen movement and probably from passive stretch, but deformation would inevitably follow cylinder muscle fiber contraction.” ¹⁵ It was therefore concluded that IMC would be contraction-sensitive rather than tension-sensitive. ^{15 16} Collagen movement may, however, excite those terminals that we frequently observed within IMC tendons. The present study added the puzzle of why IMC terminals exhibit a dualistic morphology, why synaptic axon terminals both at tendon and muscle fibers contained considerable accumulations of clear synaptic vesicles, and which transmitter they may contain. 

The close association of IMCs with MIFs indicates a crucial importance of MIF-IMC units for the functional organization of EOMs. A better knowledge of the functional role of MIFs would help to understand the function of IMCs. With respect to their physiological behavior, Chiarandini ³⁹ demonstrated two classes of EOM fibers, fast twitch and slow tonic fibers, the latter mainly consisting of global layer MIFs. Concluding from fatigue resistance and fiber classification experiments, Dean ⁴⁵ argued that in EOMs the motor units recruited first consist of MIFs. MIFs are thought to provide ripple-free control of eye position at low firing rates of oculomotor nerves. ⁴⁵  

In studies of the architecture of cat EOM, ⁴⁶ global layer MIFs were found to run the whole length of the layer, whereas global layer SIFs were much shorter and found to be connected by myomyous junctions to each other and end-to-side with MIFs. A number of SIFs may be associated to one or more MIFs to form functional units with an intrinsic capacity to regulate muscle tone or muscle contraction properties of cat EOM. Short SIFs connected in series could contribute to shorten twitch contraction time and to elevate fusion frequency, because α-motor neurons conduct action potentials faster than muscle fibers and excitation contraction coupling will be completed faster in shorter twitch fibers connected in series than in a single long one. Twitching SIFs with end-to-side connection to MIFs could excite sensory IMC terminals by pulling terminal parts of MIFs toward the muscle belly. Excited IMCs might, however, elicit long-lasting local contractions of terminal parts of MIFs to compensate for this pull, thereby possibly dampening the effect of EOM twitch contraction on the globe. 

Afferent signals from EOMs are widely spread within the CNS, indicating proprioceptor activity in EOMs. Extraocular input was traced in the superior colliculi, lateral geniculate body, pulvinar thalami, tegmentum, gigantocellular nucleus, vestibular nuclei, and prepositus hypoglossi nucleus as well as the cerebellum, the Brodmann areas 17 and 18 of the cortex, the Clare Bishop area, and the frontal cortex. ³ In many structures involved in vision and/or oculomotor control, afferent input from EOM interacts with input from the vestibular apparatus. ³ Further, EOM afferents were described to significantly influence the development of binocularity of cortical visual neurons and those of the brain stem in cats. ³ In human EOM, only two possible sources for proprioceptive input to the CNS have been described. First, terminals in the tendinous compartment of IMCs may serve proprioception. Second, human EOMs are richly supplied with muscle spindles of particular structure. ^{5 6 7 8 25} Notably, their intrafusal muscle fibers receive nerve endings that are unambiguously sensory and highly resemble sensory endings in skeletal muscle spindles. ^{5 6 7 8 25} The presence of both, IMCs (herein) and spindles, ²⁵ also was demonstrated in infancy. 

Dengis et al. ⁴⁷ reported that in humans with strabismus, botulinum toxin injected in EOMs alters proprioception from eye muscles only over the long term. These authors suggested that the toxin affected proprioceptive feedback from palisade endings. Their study strengthens the concept that proprioception may contribute to a long-term recalibration of the oculomotor system ⁴⁸ and to a long-term regulation of ocular alignment and eye movement conjugacy. ⁴⁹ A better comprehension of extraocular proprioception, however, may also help to gain new insights into strabismus and amblyopia. 

 Submitted for publication December 28, 1999; revised March 14, 2000; accepted March 22, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Laser scanning photomicrographs of longitudinal sections of adult human lateral rectus. PGP 9.5 immunoreactivity (PGP 9.5–IR) fluoresces green viewed with the FITC filter, α-bungarotoxin (α-BT) staining appears red (rhodamine filter). (A) “En grappe” motor ending of a multiply innervated extraocular muscle fiber. Small arrows: acetylcholinergic nicotinic receptors stained by α-BT (red); large arrow: supplying axon (green). Magnification, ×355. (B, C) “En plaque” motor ending of a focally innervated extraocular muscle fiber demonstrated by PGP 9.5–IR andα -BT, viewed with two different filter positions. (B) FITC filter; (C) arrow: muscle fiber (rhodamine filter). Magnification, ×290.

Figure 2.

 

Palisade endings in distal myotendons of adult human EOM. Axons labeled by anti-PGP 9.5 fluorescing green, and α-bungarotoxin staining of postsynaptic acetylcholinergic receptors fluorescing orange-green viewed with the FITC filter. MF, muscle fiber. (A) Terminal axon (arrow) investing the muscle fiber tip. FITC and interference red barrier filter. Magnification, ×290. (B) PGP 9.5—positive axon (large arrow) with varicosities approaching the muscle fiber tip from the tendon, at the contact with the muscle fiber (small arrow) in addition exhibiting α-bungarotoxin staining. Asterisk: “en grappe” motor ending in proximity to palisade ending. FITC filter. Magnification, ×190. (C) Teased palisade ending from adult human EOM. Muscle fiber tip and PGP 9.5–immunoreactive axon varicosities (arrowhead) within the continuing tendon compartment (T). Varicosities in the muscular compartment exhibit α-bungarotoxin–positive fluorescence in addition to PGP 9.5–IR, fluorescing orange (double arrowhead) in a double-exposure photomicrograph. FITC and rhodamine filter. Magnification, ×290. (D) Preparation of palisade ending shown in (C) viewed with different filter position: FITC and interference red barrier filter. Arrowheads: varicosities in the tendinous compartment; double arrowhead: a varicosity in the muscular compartment. Magnification, ×290.

Figure 3.

 

Schematic drawing of an innervated myotendinous cylinder, nerve fiber (N) enters an encapsulated (C) myotendinous junction and ramifies. Terminals (T) establish several contacts with collagenous fibrils (COL) and sarcolemma. Myoneural contacts are observed within the fingerlike intrusions of the sarcolemma and outside at the muscle fiber (MF). Fibrocyte in the tendinous compartment is marked by F.

Figure 4.

 

Ultrathin cross section through an innervated myotendinous cylinder (IMC), two-year-old human. (A) Tendinous compartment, the IMC is encircled by a capsule (C) of fibrocytes. Inside, collagen fibrils, fibrocytes (F) and two preterminal axons (arrow) are present. Scale bar, 10 μm. Right upper corner: enlargement of the two preterminal axons. Each axon is sheathed by a Schwann cell (S). Scale bar, 1 μm. (B) Muscular compartment, typical multiply innervated fiber ensheathed by a capsule (C). Scale bar, 10 μm.

Figure 5.

 

(A) Two-year-old human. Ultrathin cross section through an innervated myotendinous cylinder (IMC) terminal (T) lying among collagen fibrils. The terminal contains mitochondria and clear vesicles. Vesicles are concentrated in that part of the axolemma which is devoid of a Schwann cell (S). Basal lamina (arrow). Scale bar, 1 μm. (B) Two-year-old human. Ultrathin cross section through an IMC at its myotendinous junction. A nerve terminal (T) establishes contact with a muscle fiber protrusion. The nerve terminal contains mitochondria and densely packed clear vesicles. The synaptic cleft is filled with a basal lamina (arrow). Scale bar, 1μm. (C) Seventeen-year-old human. Ultrathin cross section of a nerve terminal (T) establishing contact with an IMC muscle fiber. Basal lamina (arrow). Scale bar, 1 μm. (D) Two-year-old human. Ultrathin cross section through a motor terminal (MT) outside the IMC. The ultrastructure of motor terminal conforms with myoneural IMC terminals as demonstrated in (C). Even the size of clear vesicles was concordant in both types of terminals. The subsynaptic membrane is smooth and the synaptic cleft is filled with a basal lamina (arrow). Scale bar, 1 μm.

The authors thank Maria Lipowec and Christiane Krivanek for their valuable technical help.