Molecular Characteristics Suggest an Effector Function of Palisade Endings in Extraocular Muscles

Abstract

purpose. To analyze palisade endings in cat extraocular muscles (EOMs) and to clarify whether these EOM-specific organs are sensory or motor.

methods. Twelve cats aged between 1 and 16 years were analyzed. Whole EOM tendons were immunostained using four different combinations of triple fluorescence labeling. Triple labeling included antibodies against choline acetyltransferase (ChAT), neurofilament, synaptophysin, and α-bungarotoxin. Preparations were examined by confocal laser scanning microscopy. ChAT-labeled EOMs were also analyzed by immunoelectron microscopy. Three-dimensional reconstructions were made of palisade endings.

results. Palisade endings were found in the distal and proximal myotendinous regions of cat EOMs. These endings arose from thin nerve fibers coming from the muscle and extending into the tendon. There, the nerve fibers turned back 180° to divide into terminal branches around the muscle fiber tips. Terminal branches established numerous contacts with the tendon attached to the muscle fiber tip and only a few contacts with the muscle fiber. Often, nerve fibers forming palisade endings on muscle fiber tips were observed to establish multiple motor contacts on muscle fibers outside palisade endings. Three-dimensional reconstructions depicted the complex morphology of the palisade endings. All nerve fibers supplying palisade endings stained positively for ChAT and neurofilament. All nerve terminals in palisade endings were ChAT and synaptophysin positive. Only neuromuscular contacts in palisade endings were positive for α-bungarotoxin, as well.

conclusions. This study provides evidence that palisade endings in cat EOMs have effector function. The findings may be of significance for strabismus surgery because palisade endings are also found in human EOMs.

Proprioceptive input from extraocular muscles (EOMs) is supposed to provide important information about the eye position in the orbit. ^{1 2} There is evidence that afferent signals from EOM proprioceptors play a role in the development of normal binocular vision, in orienting behavior, and in depth perception of mammals. ^{1 3 4 5 6}  

Classic proprioceptors in limb muscles of mammals are muscle spindles and Golgi tendon organs. In mammalian EOMs the endowment with these proprioceptors shows striking interspecies variations. Muscle spindles and Golgi tendon organs are numerous in the EOMs of even-toed ungulates. ^{7 8 9 10 11 12 13 14} In human EOMs, specifically structured muscle spindles are a regular feature, whereas Golgi tendon organs were reported to be absent. ^{15 16 17 18 19} In EOMs of cats, rats, guinea pigs, and rabbits, both muscle spindles and Golgi tendon organs were not found. ⁸  

Another nervous end organ in the EOMs of mammals is the so-called palisade ending which is unique to EOMs. Palisade endings were found at the myotendinous junction in the EOMs of several mammals, including cats, ^{20 21 22 23} dogs, ²⁴ horses, ²⁴ camels, ²⁵ rhesus monkeys, ^{21 26} sheep, ²⁷ rabbits, ^{21 28} and humans. ^{29 30 31}  

Palisade endings are dense ramifications of preterminal axons and nerve terminals around muscle fiber tips. The nerve fiber supplying a palisade ending comes from the muscle and extends into the tendon. Within the tendon, the nerve fiber turns back 180° to approach the tip of a muscle fiber. The fine structure of palisade endings was investigated primarily in rhesus monkeys ²⁶ and cats ²² and later in sheep, ²⁷ rabbits, ²⁸ and humans. ³⁰ The nervous end organ containing the palisade ending is ensheathed by a connective tissue capsule and consists of the terminal portion of a muscle fiber and its attached tendon. Ruskell ²⁶ termed such a formation an “innervated myotendinous cylinder.” Palisade endings establish nerve terminals with the collagen fibrils of the tendon and with the muscle fiber. Palisade endings in rabbit EOMs are an exception, because exclusively neuromuscular contacts were observed. ²⁸ Palisade endings are always associated with a particular type of muscle fiber that has several motor contacts along its length (multiple innervated muscle fiber). 

Several groups ^{6 16 21 22 23 26 32 33 34} have suggested that palisade endings could be the source of afferent signals, but there are still conflicting reports on the functional nature of palisade endings, whether they are sensory ^{6 16 21 22 23 26 32 33 34} or motor ^{28 35} structures, or both. ³⁰  

Fine structural analyses on palisade endings in cats, rhesus monkeys, and sheep have indicated a sensory function. ^{22 26 27} Nerve terminals, which are intermingled with collagen fibrils, are most probably sensory, and the nerve terminals contacting the muscle fiber resemble sensory neuromuscular contacts. In human EOMs, palisade endings establish sensory-like nerve terminals among the collagen fibrils. ³⁰ Nerve terminals on the muscle fiber tip exhibit morphologic features of motor terminals which is confirmed by α-bungarotoxin staining. ³⁰ Lukas et al. ³⁰ proposed that palisade endings in human might combine sensory and motor functions. Palisade endings in rabbit EOMs have a motor function. In fine structural investigations and after labeling with α-bungarotoxin, exclusively motor neuromuscular contacts were seen. ²⁸ It is important to note, however, that palisade endings in rabbits have no neurotendinous contacts that are typical of palisade endings in other species, including humans. 

The question of functional properties of palisade endings is complicated by contradicting results regarding the location of the perikarya of the nerve fibers forming the palisade endings. Sas and Scháb ³⁵ provided evidence that the cell bodies of the palisade endings are located in the oculomotor nucleus and suggested a motor function of these nervous end organs. A recent study showed that the cell bodies of the palisade endings are in the trigeminal ganglion, which implies a sensory function. ²³ In summary, the functional properties of palisade endings are, at present, not fully understood but a sensory role is favored. 

It was the goal of the present study to clarify whether palisade endings are sensory or motor. Palisade endings in cat EOMs were labeled with a antibody against choline acetyltransferase (ChAT) and with α-bungarotoxin and analyzed by confocal laser scanning microscopy and by immunoelectron microscopy. The antibody against ChAT is a specific marker for cholinergic neurons in the central and peripheral nervous systems. ^{36 37 38 39} α-Bungarotoxin binds with high affinity to postsynaptic nicotinic cholinergic receptors and is widely used to detect motor terminals in vertebrate skeletal muscles. ⁴⁰ Cats were chosen in our study, because palisade endings in this species have nerve terminals in their tendon and are comparable with those in humans. Due to their localization at the muscle tendon junction, palisade endings are crucially important for strabismus surgery. 

Material and Methods

When our collaborating veterinaries euthanatized a cat, we were supplied with the cadaver. Twelve cats of either sex, their ages varying between one and 16 years, were analyzed. Animals were used in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

From each cat, the ascending aorta was cannulated and the circulatory system was rinsed with Ringer solution followed by perfusion fixation with 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4). The orbits of the animals were dissected, and the four rectus EOMs and the two oblique EOMs were removed in full length. The proximal and distal tendons including the myotendinous regions were cut from the EOMs and further processed. For fluorescence immunohistochemistry and immunoelectron microscopy, the specimens were immersion fixed for an additional 2 hours in 4% paraformaldehyde in PB. For conventional transmission electron microscopy, the tissue was immersion fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in PB. 

Ninety-six EOMs were processed for confocal laser scanning microscopy, 24 for immunoelectron microscopy, and 24 for conventional electron microscopy. Three-dimensional (3-D) reconstructions were made of the immunolabeled palisade endings. 

Triple Fluorescence Labeling for Confocal Laser Scanning Microscopy

Proximal and distal EOM myotendons were divided into four groups, and different combinations of triple labeling were performed on the wholemounts as follows. 

Phalloidin and Antibodies against Neurofilament and Synaptophysin.

After immersion fixation, wholemounts were rinsed in 0.1 M phosphate-buffered saline (PBS; pH 7.4) containing 0.05% Tween 20. Specimens were immersed for 1 hour in prechilled acetone at 4°C, rinsed in PBS with 0.05% Tween, and subsequently transferred into a blocking solution (PBS containing 10% normal goat serum) for 1 hour at room temperature. Preparations were incubated in a mixture of primary antibodies against neurofilament (1:200, rabbit anti-neurofilament, catalog no, AB1978; Chemicon, Temecula, CA) and synaptophysin (1: 200, mouse anti-synaptophysin, catalog no. MAB 329; Chemicon) for 48 hours at 15°C. Then, preparations were incubated with the Alexa-linked secondary antibody (1:500, AlexaFluor 488 goat anti-mouse; Molecular Probes, Eugene, OR) and, after sections were rinsed, with the rhodamine-conjugated secondary antibody (1:200, goat anti-rabbit; Chemicon), each for 4 hours at room temperature. Finally specimens were labeled for 30 minutes with AlexaFluor 633–conjugated phalloidin (1:80, catalog no. A22284; Molecular Probes) at room temperature. Between the incubation steps, specimens were extensively rinsed (seven times, each 15 minutes) in PBS containing 0.05% Tween. After the staining procedure, specimens were rinsed in PBS and mounted (Citifluor; Agar, Stansted, UK). 

Phalloidin and Antibodies against Neurofilament and Choline Acetyltransferase.

Wholemounts were processed by the method just described. PBS containing 10% donkey serum and 3% bovine serum albumin was used as blocking solution. Specimens were incubated in a mixture of primary antibodies against neurofilament and ChAT (1:100, goat anti-ChAT, catalog no. AB 144P; Chemicon). The wholemounts were then stained with the rhodamine-conjugated secondary antibody (1:200, donkey anti-rabbit; Chemicon) and with the Alexa-linked secondary antibody (1:500, AlexaFluor 488 rabbit anti-goat; Molecular Probes). Finally, specimens were labeled with AlexaFluor-633–conjugated phalloidin. 

Phalloidin and Antibodies against Choline Acetyltransferase and Synaptophysin.

PBS containing 10% rabbit serum was used as a blocking solution. Preparations were incubated in the primary antibodies against ChAT and synaptophysin and then in the Alexa-linked secondary antibody (1:500, AlexaFluor 488 rabbit anti-goat; Molecular Probes). After the reaction was blocked in PBS with 10% goat serum, the preparations were stained with the rhodamine-conjugated secondary antibody (1:200, goat anti-mouse; Chemicon). Thereafter, specimens were labeled with AlexaFluor-633–conjugated phalloidin. 

α-Bungarotoxin and Antibodies against Choline Acetyltransferase and Synaptophysin.

Specimens were incubated in the primary antibodies against ChAT and synaptophysin, and then the wholemounts were labeled with the Alexa-linked secondary antibodies (1:500, AlexaFluor 488 rabbit anti-goat and 1:100 AlexaFluor 633 goat anti-mouse; Molecular Probes). Finally specimens were labeled with rhodamine-conjugated α-bungarotoxin (1:1000, tetramethylrhodamine α-bungarotoxin, catalog no. T1175; Molecular Probes) for 20 minutes. 

Wholemounts were examined under a confocal laser scanning microscope (CLSM; model LSM 410; Carl Zeiss Meditec, Oberkochen, Germany). When the palisade endings were identified, a series of longitudinal virtual CLSM sections of 0.4- to 1-μm thickness were cut through the specimens. Each section was photodocumented and subsequently 3-D projections were calculated with the microscope software (LSM Image Examiner; Carl Zeiss Meditec). 

Besides ChAT, antibodies against vesicular acetylcholine transporter (VAChT) are used to mark cholinergic neurons in the nervous system. ^{36 37} The VAChT antisera that are currently available react with the VAChT protein in human, rat, and mouse. We tested two anti-VAChT antibodies and labeled motor nerve terminals in rat skeletal muscle. We unfortunately had to realize that the VAChT antisera did not cross react with the cat VAChT protein. On request, we were told that an antibody against VAChT that would react in the feline species is generally not available. 

3-D Reconstruction

3-D reconstructions were made of palisade endings that were labeled with phalloidin and antibodies against ChAT and synaptophysin. Wholemounts were analyzed by CLSM and virtual longitudinal sections of 0.4- to 1-μm thickness were cut through the palisade endings. Image data were stored from each virtual section. The image sets of each immunostaining were transformed into binary images and finally transferred to a computer workstation (Silicon Graphics, Mountain View, CA). A reconstruction software package (Velocity Image 3 LCC; Silicon Graphics) converted the image sets into 3-D objects. 

Conventional Electron Microscopy

After immersion fixation was completed, EOM tendons were cut into small pieces. Specimens were postfixed in 1% osmium tetroxide, dehydrated in ethanol, and embedded in Epon. Semithin cross sections (1 μm) were cut from tissue blocks and analyzed in the light microscope. When palisade endings were identified, ultrathin sections were cut at appropriate intervals. Sections were mounted on dioxane polyvinyl formal–coated (Formvar; SPI, West Chester, PA) copper grids, stained in a solution of 2% uranyl acetate followed by 0.4% lead citrate, and examined with an electron microscope (model EM10; Carl Zeiss Meditec). 

Immunoelectron Microscopy with Pre-embedding Method

After immersion fixation, the wholemounts were rinsed in PBS containing 0.05% Tween 20. Then specimens were immersed for 1 hour in prechilled acetone at 4°C and rinsed in PBS with 0.05% Tween. To inhibit the endogenous peroxidase activity, we incubated wholemounts in PBS containing 0.05% phenylhydrazine and then in PBS containing 10% rabbit serum to block the nonspecific binding sites, each for 1 hour. Subsequently, wholemounts were transferred to the primary antibody against ChAT (1:100) for 48 hours at 15°C. Specimens were incubated in the biotinylated rabbit anti goat antibody (1:500 catalog no. AP106B; Chemicon) overnight and then in avidin-biotin-complex (ABC, catalog no. K0355; Dako, Glostrup, Denmark) for 6 hours, each at room temperature. The peroxidase reaction was developed in a chromogen solution containing 0.05% diaminobenzidine (DAB) and 0.01% H₂O₂. Between the incubation steps, specimens were extensively rinsed (seven times, each 15 minutes) in PBS containing 0.05% Tween. 

Specimens were cut into small pieces, postfixed in 1% osmium tetroxide, dehydrated in ethanol, and embedded in Epon. Semithin cross sections (1 μm) were cut, and when palisade endings were identified in the light microscope, ultrathin sections were obtained. Sections were mounted on dioxane polyvinyl formal–coated copper grids, stained in a solution of 2% uranyl acetate followed by 0.4% lead citrate, and examined with an electron microscope (model EM10; Carl Zeiss Meditec). 

Control Experiments

In control experiments, the primary antibodies were omitted, and the secondary antibodies were used alone. In all cases, the omission of the primary antibodies resulted in a complete lack of immunostaining. 

Other control experiments were performed to confirm the specificity of the antibody against ChAT used in the present study. Cryostat sections of cat limb muscles were double labeled with anti-ChAT and α-bungarotoxin; or, alternatively, sections were double labeled with anti-ChAT and anti-neurofilament. The former experiment showed that the motor endplates on skeletal muscle fibers were positive for ChAT and α-bungarotoxin. In the latter experiment, sensory nerve fibers innervating muscle spindles and Golgi tendon organs stained only positively for neurofilament, whereas motor nerve fibers stained positively for neurofilament as well as for ChAT. 

Results

Number and Distribution of Palisade Endings

Palisade endings were observed in all EOMs in the proximal and distal myotendinous regions, their number being higher in the rectus EOMs than in the oblique EOMs. In two medial rectus EOMs, we counted 81 and 94 palisade endings at the distal myotendinous region and a similar number was estimated for the other rectus EOMs. Palisade endings (n = 30 and 34) were counted in two inferior oblique EOMs, comparable to the number estimated in the superior oblique EOMs. In all EOMs, the number of palisade endings in the distal myotendinous region exceeded that in the proximal region. 

Morphology of Palisade Endings

Confocal Laser Scanning Microscopy and 3-D Reconstructions.

In wholemounts, nerve fibers were labeled with antibody against neurofilament (a marker for sensory and motor nerve fibers) and nerve terminals with antibody against synaptophysin (a marker for all nerve terminals, sensory, and motor). Muscle fibers were stained with phalloidin. 3-D reconstructions were made of palisade endings, in which nerve fibers were stained with anti-ChAT, nerve terminals with anti-synaptophysin, and muscle fibers with phalloidin. 

At the site of the junction of muscle fibers to the tendon, muscle fibers had longitudinal finger-shaped processes that attached the muscle fibers to the tendon. Thin neurofilament-positive axons (2–3 μm in diameter) were observed that came from the muscle to penetrate the tendon. The nerve fibers ran either separately or in bundles of up to four axons and extended for variable distances into the tendon and then looped back to the myotendinous region. One or two axons approached a single muscle fiber tip where each axon ramified into several, approximately parallel branches (Figs. 1 2) . Alternatively, nerve fibers were found that ramified at the muscle fiber tip without previously turning back from the tendon (see Fig. 5D ). Several axon branches surrounded the extremity of the muscle fiber. Other axon branches ran between the longitudinal processes of the muscle fiber. 

Throughout their course, axon branches repeatedly exhibited varicosities that established synaptophysin-positive nerve terminals. Nerve terminals laid within the tendon compartment, and others were observed between the muscle fiber processes or around the muscle fiber surface (Figs. 1 2) . 

In many cases, we observed that the same nerve fiber supplying a palisade ending at the muscle fiber tip established multiple motor contacts on the muscle fiber before forming the palisade complex (see Fig. 5D ). In other cases, we could not find out whether the same nerve fiber supplying a palisade ending at the muscle fiber tip also formed motor contacts outside the palisade ending, because, in the myotendinous region, the axon intermingled with others and could not be traced further. 

3-D reconstructions of palisade endings depicted the overall morphology of this complex formation. The course and branching pattern of the nerve fibers, the distribution of nerve terminals in the tendon compartment between the muscle fiber processes and around the muscle fiber tip was clearly shown in 3-D images. Moreover, the longitudinal finger-shaped processes of the muscle fiber tip at its attachment to the tendon became apparent (Fig. 2) . 

Conventional Electron Microscopy.

Each palisade ending was sheathed by a thin capsule consisting of two to four cell layers. The capsule cells had the appearance of fibrocytes. The capsule had the shape of a cylinder and enclosed the terminal portion of a muscle fiber and its attached tendon (Fig. 3) . The myelinated nerve fiber supplying a palisade ending penetrated the capsule in the tendon compartment and was surrounded by perineural cells, which had a basal lamina. Inside the capsule, the nerve fiber lost its perineural envelope. 

Approaching the muscle fiber, the nerve fiber shed its myelin sheath and divided into preterminal axons of 1- to 2-μm thickness. The preterminal axons intermingled with collagen fibrils and extended toward the muscle fiber tip. Preterminal axons contained neurofilaments, neurotubules, and mitochondria and were completely surrounded by Schwann cells, which were coated by a basal lamina (Figs. 3 4A) . Closer to the muscle fiber tip, segments of the preterminal axons repeatedly partially lost their Schwann cell investment and formed nerve terminals. 

Nerve terminals contacting collagen fibrils were observed in the tendon compartment and were only partly surrounded by Schwann cells. A basal lamina was always interposed between the axolemma devoid of a Schwann cell and the neighboring collagen fibrils. Nerve terminals contacting collagen were also found around the entire muscle fiber tip and between the longitudinal muscle fiber processes. A thin layer of collagen material separated the nerve terminals from the muscle fiber surface. In some palisade endings, we observed a few nerve terminals that established contacts with the muscle fiber. In these neuromuscular contacts, the synaptic cleft was free from a basal lamina. Nerve terminals contacting collagen fibrils as well as neuromuscular contacts contained mitochondria and were filled with small, clear vesicles (Figs. 4B 4C 4D) . 

Molecular Properties of Palisade Endings

Confocal Laser Scanning Microscopy.

In wholemounts, three combinations of triple labeling were performed: Wholemounts were labeled with phalloidin, anti-ChAT, and anti-neurofilament; with phalloidin, anti-ChAT, and anti-synaptophysin; and with α-bungarotoxin, anti-ChAT, and anti-synaptophysin. 

Observations by CLSM showed that ChAT-positive nerve fibers formed typical palisade endings at the myotendinous junction. Bundles of nerve fibers or single nerve fibers extended from the muscle into the tendon and turned back to form a caplike ramification on the tip of a single muscle fiber. Other nerve fibers reached the muscle fiber tip more directly, without previously looping in the tendon (Fig. 5D) . In the wholemounts that were labeled with phalloidin, anti-ChAT, and anti-neurofilament, all nerve fibers forming palisade endings stained positively for both ChAT and neurofilament. No neurofilament-positive, ChAT-negative nerve fibers were detected in the palisade complex (Fig. 5) . 

ChAT-positive axon branches approaching the muscle fiber tip exhibited varicosities forming nerve terminals. In wholemounts stained with phalloidin, anti-ChAT, and anti-synaptophysin, all nerve terminals in palisade endings were positive for both ChAT and synaptophysin. No synaptophysin-positive, ChAT-negative terminals were observed (Fig. 6) . 

In wholemounts labeled with α-bungarotoxin, anti-ChAT, and anti-synaptophysin, we observed palisade endings in which nerve terminals stained positively for ChAT and synaptophysin but were negative for α-bungarotoxin. Such nerve terminals obviously established contacts with collagen fibrils. In a few palisade endings we found, in addition to ChAT-synaptophysin–immunoreactive nerve terminals, a small number of nerve terminals that were also positive for α-bungarotoxin. Such triple-positive nerve terminals obviously established contacts with the muscle fiber (Fig. 7) . 

Immunoelectron Microscopy.

With the transmission electron microscope (TEM) ChAT immunoreactivity was identified by the electron-dense deposit of the DAB reaction product. 

ChAT-immunoreactive myelinated axons surrounded by perineural cells were observed to penetrate the fibrocyte-capsule of the palisade endings in the tendon compartment. Inside the capsule, the ChAT-positive axon lost its myelin sheath and branched into preterminal axons. ChAT positive preterminal axons were entirely covered by Schwann cells (Figs. 8A 8B 8C) . 

In their course toward the muscle fiber tip, segments of the preterminal axons repeatedly lost their Schwann cell investment, thereby establishing nerve terminals. The axolemma of the nerve terminals, free of a Schwann cell, was covered by a basal lamina. Nerve terminals were surrounded by collagen fibrils and were observed in the tendon compartment, between the muscle fiber processes, and around the muscle fiber tip. All these nerve terminals exhibited ChAT immunoreactivity. The DAB reaction product in the nerve terminals was so abundant that mitochondria and vesicles were masked (Figs. 8D 8E 8F) . In those palisade endings that were analyzed, we did not find neuromuscular contacts. 

Morphology and Innervation of Muscle Fibers Associated with Palisade Endings

Observations by TEM showed that muscle fibers associated with palisade endings had densely packed myofibrils, few mitochondria, and a poorly developed sarcoplasmic reticulum. 

On muscle fibers associated with palisade endings, we found motor terminals outside the encapsulation of the palisade endings. Motor nerve terminals contained mitochondria and were full of small, clear vesicles. Motor contacts had a basal lamina in the synaptic cleft and stained positively for ChAT. The subsynaptic membrane exhibited shallow folding (Figs. 9A 9B) . 

By CLSM, the muscle fibers associated with palisade endings were traced along their length. We found that all muscle fibers with palisade endings received several ChAT-positive motor contacts distributed along their length. Motor contacts also stained positively for α-bungarotoxin (Figs. 5D 9C 9D) . 

The morphologic characteristics and the molecular properties of nerve terminals in palisade endings and of motor terminals on multiple innervated muscle fibers are summarized in Table 1 . 

Discussion

Up to now there have been contradicting opinions about the function of palisade endings; but, all evidence considered, a sensory role is favored. In the present study, we analyzed the palisade endings of cat EOMs by CLSM and TEM by using antibodies against ChAT and α-bungarotoxin. ChAT synthesizes acetylcholine, and anti-ChAT defines cholinergic neurons in the central and peripheral nervous system. ^{36 37 38 39} α-Bungarotoxin labels postsynaptic acetylcholinergic receptors. ⁴⁰ Our goal was to clarify whether palisade endings in cat EOMs are sensory or motor. 

In accordance with the fine structural investigation of Alvarado-Mallart and Pinçon-Raymond, ²² our results confirmed the presence of palisade endings in cat EOMs. Myelinated nerve fibers forming palisade endings extended from the muscle into the tendon and turned back to form a terminal arborization on the tip of a single muscle fiber. As a new and surprising finding, our results showed that many nerve fibers forming palisade endings on muscle fiber tips established multiple motor contacts on muscle fibers outside the palisade endings. Such a situation was not observable in each palisade ending, because nerve fibers were abundant at the myotendinous junction, and tracing of individual axons forming a palisade complex was sometimes very difficult. A further novelty was that nerve fibers supplying palisade endings were ChAT immunoreactive. Labeling with anti-neurofilament (a general marker for nerve fibers) and anti-ChAT demonstrated that all nerve fibers supplying palisade endings stained positively for ChAT. An intermixing of cholinergic and noncholinergic nerve fibers at the palisade organs therefore can almost certainly be excluded. The 3-D reconstruction of palisade endings was a novel approach in this study and contributed to the understanding of the complex spatial morphology of this formation. 

Concordantly with previous studies in rhesus monkeys ²⁶ cat, ²² and sheep ²⁷ we found that most of the nerve terminals in palisade endings were surrounded by collagen fibrils and that only a few of them established neuromuscular contacts with no interposition of a basal lamina. Both neurotendinous and neuromuscular contacts were full of small, clear vesicles. All nerve terminals in palisade endings, those contacting the collagen fibrils and those establishing neuromuscular contacts, were positive for synaptophysin (a general marker for nerve terminals). It is important to note that all synaptophysin positive nerve terminals were also ChAT positive. Exclusively, the neuromuscular contacts stained positively for α-bungarotoxin, as well. Likewise, by TEM we observed ChAT-positive neurotendinous contacts. We did not find neuromuscular contacts in the ChAT-labeled palisade endings analyzed by TEM, which may have been because neuromuscular contacts were absent in many palisade endings and, if present, they were few. 

In summary, the results of the present study showed that cholinergic nerve fibers supply palisade endings that established cholinergic neurotendinous and cholinergic/motor neuromuscular contacts. In principle, these results allow two interpretations with respect to the function of palisade endings: Either palisade endings represent cholinergic sensory organs, or they are effector organs involving collagen fibrils. 

Palisade Endings as Sensory Organs

In our fine structural analyses we found that nerve terminals in palisade endings formed intimate contacts with the collagen fibrils, which is analogous to nerve terminals in Golgi tendon organs. ^{10 12 13 14 41 42 43} It is most probable that nerve terminals contacting collagen fibrils as target structures are sensory. Nerve terminals contacting the muscle fiber in palisade endings of cat, lacked a basal lamina in the synaptic cleft, and thus resembled sensory nerve terminals on intrafusal muscle fibers in muscle spindles. ^{9 14 15 16 17 18 19} Former fine structural studies on palisade endings in rhesus monkey, ²⁶ cat, ²² and sheep ²⁷ EOMs were in line with our results. Based on these fine structural characteristics, palisade endings were classified as sensory organs. ^{22 26 27}  

Further arguments for a sensory role of palisade endings came from tracing experiments. In cats, neuronal tracers were injected into the EOMs. ³² By retrograde transport of the tracer, perikarya in the ipsilateral trigeminal ganglion were labeled. The trigeminal ganglion is thought to contain the perikarya of neurons innervating EOM proprioceptors. ^{6 32 44 45} Because classic proprioceptors such as muscle spindles and Golgi tendon organs are absent in cat EOMs, Porter and Spencer ³² concluded that the labeled neurons in the trigeminal ganglion innervate the palisade endings. In anterograde tracing experiments Billig et al. ²³ injected neuronal tracers into the trigeminal ganglion of cats and found three different kinds of labeled nerve endings in the EOMs, one type resembling palisade endings. 

Palisade Endings as Effector Organs

We convincingly showed that exclusively ChAT-positive nerve fibers formed palisade endings in cat EOMs. All nerve terminals in palisade endings were ChAT positive, too. We further unambiguously demonstrated that neuromuscular contacts in palisade endings were α-bungarotoxin positive. In their vesicle content and molecular properties, nerve terminals in palisade endings were indistinguishable from motor contacts (see Table 1 ). Morphologically we provided evidence that palisade endings arose from nerve fibers that established motor contacts on the muscle fibers outside the palisade endings. Taken together, these observations strongly suggest that palisade endings in cat EOMs fulfill an effector function. 

Support that palisade endings are effectors is available in a previous report. ³⁵ In cats, after small stereotactic lesions of the motor nuclei (oculomotor, trochlear, and abducens nuclei) innervating the EOMs, Sas and Scháb ³⁵ found degenerated motor end plates and also degenerated palisade endings in the EOMs. The authors concluded that palisade endings must be innervated by motor neurons that are located in the EOM motor nuclei and suggested a motor function for these nervous end organs. 

Recently, we analyzed palisade endings in rabbit EOMs. We found neuromuscular contacts that had a basal lamina in the synaptic cleft, thereby resembling motor terminals which was confirmed by α-bungarotoxin staining. ²⁸ These findings indicate that palisade endings in rabbit EOMs have a motor function. ²⁸ Among mammals, palisade endings in rabbits are unique because neurotendinous contacts, which are a feature of palisade endings in all other species including humans, were not found. It is therefore difficult to compare palisade endings in rabbit EOMs with those in other species and humans. 

Summing up, besides the present investigation, there is only the study accomplished by Sas and Scháb ³⁵ that provides evidence for an effector or motor function of palisade endings. 

Conclusions

We think that our study provides convincing evidence in favor of an effector role of these EOM-specific organs. First, we showed that nerve fibers forming palisade complexes also established motor contacts outside the palisade endings. Second, nerve fibers forming palisade endings contained ChAT. ChAT is the synthesizing enzyme of acetylcholine that is presumably stored in the clear synaptic vesicles occurring massively in the nerve terminals at the tendon and on the muscle fiber. Third, neuromuscular contacts in palisade endings were positive for α-bungarotoxin, which labels postsynaptic acetylcholinergic receptors. 

In humans, the hypothesis was recently put forward by Lukas et al. ³⁰ that palisade endings might fulfill both motor and sensory functions. Such a concept, however, implies that the palisade complex is supplied by an afferent and an efferent nerve fiber. Based on our present study, we can definitively exclude the notion that palisade endings have sensory and motor innervation, because exclusively cholinergic nerve fibers were observed to form palisade endings. 

In the literature, there are examples that indicate that sensory nerve terminals store neurotransmitters and may have a secretory function. ^{46 47 48 49} In two studies, it has been shown that sensory nerve fibers can contain ChAT, which indicates the ability to synthesize acetylcholine. ^{50 51} Sann et al. ⁵⁰ described ChAT-positive neurons, most of them without a myelin sheath, in the dorsal root ganglion of rats. Recently, Yasuhara et al. ⁵¹ detected cholinergic sensory nerve fibers, extending from the trigeminal ophthalmic nerve, that innervate the rat iris. These iridial nerve fibers contained a splice variant of the conventional ChAT enzyme, the so called “peripheral ChAT.” ⁵¹ The conventional ChAT was not detected in the iridial nerve fibers of the rat. ⁵¹ The functional importance of cholinergic sensory nerve fibers remains unclear. 

Based on these two studies, ^{50 51} one could claim that palisade endings are cholinergic sensory organs. However, the following findings in our study rather disapprove this assumption. In many cases, we observed, first, that palisade endings arose from nerve fibers that established motor contacts on muscle fibers and, second, that neuromuscular contacts in palisade endings exhibited α-bungarotoxin staining. 

Palisade endings were numerous in the EOMs (81 and 94 palisade were counted in two medial recti, 30 and 34 in two inferior obliques) and their nerve terminals exhibited no signs of morphologic alterations. These observations indicate functional importance. It is, however, difficult to imagine which function palisade endings with motor nerve terminals might have. Most nerve terminals in palisade endings are located within the tendon, and it is unclear what effect a release of neurotransmitter (acetylcholine) may have on the surrounding collagen. Neurotendinous contacts are mostly at great distance from the muscle fiber surface. Neurotransmitter release far away from the target surface indicate that palisade endings belong to a motor system that is different from that in skeletal muscle, because motor terminals are usually in close contact with the muscle fiber surface. Thus, our present findings are at the moment inexplicable and physiological studies on palisade endings, which are have not been undertaken to date, are highly warranted. 

It was shown that afferent signals from mammalian EOMs reach several regions within the central nervous system, including the superior colliculus, the lateral geniculate body, the pulvinar of the thalamus, the tegmentum, the vestibular nucleus, the nucleus prepositus hypoglossi, the cerebellum, Brodman areas 17 and 18, the Clare Bishop area, and the frontal cortex. ⁵ Palisade endings were supposed to be the source of proprioceptive input to the central nervous system, particularly in those mammals in which muscle spindles and Golgi tendon organs are few or absent in the EOMs (rhesus monkey, cat, dog, horse). ⁶ Our study provides evidence that palisade endings are effectors and are not candidates for a sensory input to the central nervous system. As humans have muscle spindles and palisade endings in their EOMs, our findings mean that muscle spindles are the only source of proprioceptive information from the EOMs. Furthermore, as palisade endings are located at the myotendinous region, knowledge about an effector function may be of significance in strabismus surgery. 

 Supported by Grant P15478 from the Fonds zur Foerderung der wissenschaftlichen Forschung.

 Submitted for publication September 14, 2004; revised October 15, 2004; accepted October 22, 2004.

 Disclosure: K.Z. Konakci, None; J. Streicher, None; W. Hoetzenecker, None; M.J.F. Blumer, None; J.-R. Lukas, None; R. Blumer, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

3-D projections of palisade endings viewed by CLSM. Nerve fibers are labeled with anti-neurofilament (red) and the terminals with anti-synaptophysin (green). The muscle fibers are stained with phalloidin (white). The tendon attached to the muscle fiber tips are not stained. Nerve fibers come from the muscle and extend into the tendon. Within the tendon, the nerve fibers make a loop and turn back to the myotendinous region. Nerve fibers split into several axon branches that established nerve terminals around the muscle fiber tip. Scale bar, 100 μm.

Figure 2.

 

3-D visualization of two palisade endings reconstructed from virtual serial sections cut with the CLSM. Nerve fibers were labeled with anti-ChAT (green), the nerve terminals with anti-synaptophysin (red), and the muscle fibers with phalloidin (gray). The tendon is not shown. CLSM micrographs of the palisade endings are shown in Figure 6 . (A, B) The whole formation of the palisade ending is shown from two different views. Two nerve fibers run beside the muscle fiber and extend into the tendon. Within the tendon the nerve fibers turn 180° and approach the muscle fiber tip. At its attachment to the tendon the muscle fiber exhibits several longitudinal finger-shaped processes. The nerve approaching the muscle fiber tip divides into several axon branches that repeatedly establish nerve terminals in the tendon compartment, between the muscle fiber processes and around the muscle fiber. (C, D) Another palisade ending from two different views. From the nerve fiber supplying the palisade ending, only the final section is shown. The nerve fiber splits into axon branches that establish nerve terminals. (E) Visualization of the palisade ending illustrated in (C) and (D) with the nerve fibers removed. The localization of the nerve terminals in the tendon compartment, between the muscle fiber processes and around the muscle fiber, is shown.

Figure 5.

 

3-D projections of palisade endings viewed with the CLSM (A, C, D). (B) CLSM picture of a single section through palisade endings shown in (A). Nerve fibers are immunostained with anti-neurofilament (red) and with anti-ChAT (green). Muscle fibers are labeled with phalloidin (white). The tendon is not stained. (A) Bundles of nerve fibers coming from the muscle extend into the tendon and turn back to form typical palisade endings on the muscle fiber tips. All nerve fibers stained positively for neurofilament, as well as for ChAT. (B) Single confocal section through the palisade ending of (A) with all nerve fibers double stained. (C) A single nerve fiber that is positive for neurofilament and ChAT supplies a palisade ending on the tip of a muscle fiber. (D) A nerve fiber double positive for ChAT and neurofilament runs alongside a muscle fiber, thereby establishing several motor contacts. At the muscle fiber tip, this nerve fiber immediately branches to form a palisade ending. Scale bars, 100 μm.

Figure 3.

 

TEM micrograph showing a cross section through a palisade ending, at the level of the myotendinous junction. The whole formation of the palisade ending is encircled by a capsule (C) of fibrocytes. The myelinated nerve fiber (N) forming the palisade ending has already passed through the capsule layers. Two other myelinated nerve fibers outside the capsule are seen at the top of the picture. Inside the capsule, collagen (Col), fibrocytes (F), a process of the muscle fiber (MF), and preterminal axons (arrow) are present. Scale bar, 10 μm.

Figure 4.

 

TEM micrographs showing preterminal axons and nerve terminals in palisade endings. (A) Cross section through preterminal axons (A) completely encircled by Schwann cells (S). A basal lamina (arrow) covers the surface of the Schwann cells. The axons contain mitochondria, neurotubuli and neurofilaments. C, capsule. (B) Cross sections through a nerve terminal (T) in the tendon compartment establishing contact to the surrounding collagen fibrils. The areas of the nerve terminal contacting the collagen lack a Schwann cell (S) investment (arrowheads). The nerve terminal contains mitochondria and is full of small, clear vesicles. F, fibrocyte, arrow: basal lamina. (C) Cross section through nerve terminals (T) lying between muscle fiber processes. The nerve terminals establish contacts to the surrounding collagen. S, Schwann cell; MF, muscle fiber; arrow: basal lamina. Inset: detailed image of a nerve terminal. (D) Cross section through a nerve terminal (T) contacting the muscle fiber. The synaptic cleft (arrowhead) is free of the basal lamina (arrow). The nerve terminal is packed with small, clear vesicles. MF, muscle fiber. Inset: detailed image of the contact. Scale bars, 1 μm.

Figure 6.

 

3-D projections of palisade endings viewed with the CLSM (A, B). Nerve fibers are labeled with anti-ChAT (green) and nerve terminals with anti-synaptophysin (red). Muscle fibers are stained with phalloidin (white). The tendon is not stained. (A) Two ChAT positive nerve fibers form a palisade ending on a muscle fiber tip. Axon branches exhibit varicosities that establish nerve terminals. All nerve terminals are double positive for ChAT and synaptophysin. (B) Another palisade ending with the nerve terminal positive for ChAT and synaptophysin. Scale bars, 100 μm.

Figure 7.

 

3-D projections of palisade endings viewed with the CLSM. Nerve fibers are immunostained with anti-ChAT (green) and the terminals with anti-synaptophysin (white) and α-bungarotoxin (red). Muscle fibers come from the left. Muscle fibers and tendon are not stained. (A) ChAT positive nerve fibers forming two palisade endings. Nerve terminals contacting collagen stain positively for ChAT and synaptophysin only. Neuromuscular contacts are triple positive for ChAT, synaptophysin, and α-bungarotoxin. (B) Palisade endings in (A) showing only the ChAT and α-bungarotoxin staining. Synaptophysin labeling is omitted. Scale bar, 100 μm.

Figure 8.

 

TEM micrographs from immunoelectron microscopy. Nerve fibers and nerve terminals are labeled with anti-ChAT. Immunoreactivity was identified by the electron-dense DAB reaction product. (A) Cross section through a ChAT-positive myelinated nerve fiber (N) inside the capsule (C) of a palisade ending. The staining of the nerve fiber is weak but evident when compared with unstained nerve fibers in Figure 3 . The nerve fiber is surrounded by perineural cells (PC) coated with a basal lamina (arrow). Myelin sheath ( ). Scale bar, 1 μm. (B) Cross section through a palisade ending viewed at low magnification. Two ChAT-positive preterminal axons (arrows) are visible inside the capsule (C). (C) Oblique section through a ChAT-positive preterminal axon (A), encircled by a Schwann cell (S), which is coated by a basal lamina (arrow). Capsule (C). (D–F) Cross sections through ChAT-immunoreactive nerve terminals (T) establishing contacts to the neighboring collagen fibrils. Parts of the nerve terminals lack a Schwann cell (S) and are only covered by a basal lamina (arrow). The nerve terminals are lying in different positions. (D) Within the tendon compartment of the palisade ending, (E) between two muscle fiber processes, and (F) close to the muscle fiber surface. MF, muscle fiber; F, fibrocyte. Scale bars, 1 μm.

Figure 9.

 

TEM and CLSM micrographs showing the motor innervation of muscle fibers associated with palisade endings. (A) Unlabeled motor terminal; (B) motor terminal stained with anti ChAT. (C) Nerve fibers are labeled with anti-ChAT (green), nerve terminals with anti-synaptophysin (red), and muscle fibers with phalloidin (white). (D) Nerve fibers are stained with anti-ChAT and terminals with anti-synaptophysin (white) and α-bungarotoxin (red). (A, B) An unlabeled and a ChAT-positive motor terminal (MT) outside the palisade ending. The unlabeled motor terminal is full of small, clear vesicles. A basal lamina (arrow) fills the synaptic cleft, and the subsynaptic membrane has shallow foldings. MF, muscle fiber; SC, Schwann cell. (C, D) Outside the palisade ending the muscle fiber receives several motor contacts along its length. Motor contacts are positive for ChAT and synaptophysin (C) as well as for α-bungarotoxin (D). Scale bars: (A, B) 1 μm; (C, D) 100 μm.

Table 1.

 

Morphological and Molecular Characteristics of Nerve Terminals in Palisade Endings and of Motor Terminals on Multiple Innervated Muscle Fibers

	Palisade Endings			Motor Terminals on Multiple Innervated Muscle Fibers
		Neurotendinous Terminals	Neuromuscular Terminals
Basal lamina in the synaptic cleft	+*	−	+
Small clear vesicles in the terminal	+	+	+
Synaptophysin-reactivity	+	+	+
ChAT-reactivity	+	+	+
α-bungarotoxin-reactivity	−	+	+

*  Neurotendinous contacts are covered by a basal lamina, which is also a feature of sensory neurotendinous contacts in Golgi tendon organs. ^{13 14 41 42}

The authors thank the veterinarians Ingrid Harant, Thomas Czedik-Eysenberg, and Mario Pichler for kindly providing material; Christiane Krivanek and Mariella Lipowec for valuable technical assistance; and Adolf Ellinger for helpful advice regarding immunoelectron microscopy.