Palisade Endings: Cholinergic Sensory Organs or Effector Organs?

Abstract

purpose. This study aims to complement the authors’ prior findings on palisade endings in extraocular muscles (EOMs) of monkeys, and to clarify whether palisade endings are cholinergic motor or cholinergic sensory.

methods. Macaque monkeys (Macaca fascicularis, n = 10) of both sexes were analyzed using three-dimensional (3D) reconstructions, confocal laser scanning microscopy (CLSM), and conventional/immuno transmission electron microscopy (TEM). For CLSM, we used three combinations of triple fluorescent labeling. EOM wholemounts were labeled with cholinergic markers, including choline acetyltransferase (ChAT), choline transporter (ChT), vesicular acetylcholine transporter (VAChT), and a classical postsynaptic marker for motor terminals, namely α-bungarotoxin. Muscle fibers were counterstained with phalloidin. 3D reconstructions were done of triple-labeled palisade endings. For immuno TEM, tissue was labeled with antibody against ChAT.

results. Concordant with prior findings, the authors demonstrated that palisade endings at the muscle fiber tips arose from nerve fibers that are ChAT-positive. In 25% of the cases, axons forming palisade endings established multiple neuromuscular contacts outside the palisade complex. Such additional neuromuscular contacts were motor terminals, as demonstrated by α-bungarotoxin binding. All palisade endings established nerve terminals on the tendon. In 40% of the palisade endings, nerve terminals were observed on the muscle fiber as well. Neurotendinous contacts and neuromuscular contacts in palisade endings were ChT/ChAT/VAChT-immunoreactive. Neuromuscular contacts exhibited structural features of motor terminals and were also α-bungarotoxin positive.

conclusions. The present study ascertained that palisade endings are cholinergic motor organs. Therefore, it was concluded that palisade endings are not candidates to provide eye-position signals.

Knowledge of the position of objects in space is of practical importance for activities in everyday life. Simple tasks such as reaching for an object, as well as more complex tasks such as driving a car, need spatial information. To accurately locate objects in space, the central nervous system needs visual information from the retina as well as additional information from the eye’s position in the orbit. It is believed that such nonvisual information comes from sensory organs (proprioceptors) in extraocular muscles (EOMs). In fact, experimental findings in mammals and psychophysical studies in man provide important support that proprioceptive input from EOMs plays a role in the development of a normal binocular vision, depth perception, and orienting behavior. ^{1 2 3 4 5 6 7}  

Interestingly, classical proprioceptors (muscle spindles and Golgi tendon organs) as known from other skeletal muscles, are absent in the EOMs of most mammals. ⁸ Thus, it is not clear where the source of EOM proprioception lies. One possible origin of this information is the palisade ending (myotendinous cylinder), a nerve-end organ that is unique to EOMs. 

Palisade endings, which consist of a dense ramification of preterminal axons and their vesicle-loaded nerve terminals around the tip of a muscle fiber, are encapsulated organs located at the distal and proximal myotendinous junction. Dogiel ⁹ was one of the first scientists to describe palisade endings in the EOMs of several mammals. So far, palisade endings have been found in the EOMs of almost all species investigated, including monkeys, ^{10 11} cats, ^{12 13 14 15} rabbits, ^{16 17} sheep, ¹⁸ rats, ¹⁹ and humans. ^{20 21 22} In most species (cat, ^{12 15} sheep, ¹⁸ monkey, ¹⁰ and human ²⁰ ), palisade nerve terminals establish contacts on the tendon and on the muscle fiber. In fact, in cats, ^{12 15} sheep, ¹⁸ monkeys, ¹⁰ and humans, ²⁰ neurotendinous contacts are more numerous than neuromuscular contacts. Palisade endings in rabbits ¹⁷ and rats ¹⁹ appear to be an exception, because exclusively neuromuscular contacts have been observed. Palisade endings arise from nerve fibers that, coming from the muscle, extend into the tendon, and then turn back 180° to terminate around the tip of a muscle fiber. Palisade endings are exclusively found in the global (inner) layer of the EOMs and they are associated with non-twitch, multiply innervated muscle fibers that have several motor contacts along their length. ^{12 21 23} The multiply innervated muscle fibers have a unique innervation from small motoneurons located outside the borders of the main EOM nuclei. ²⁴  

Although direct physiologic evidence is still lacking, the literature to date suggests that palisade endings are sensory organs providing important information about the eye position. ^{2 3 6 12 13 19 25 26} In a single, earlier study, Sas and Schab ²⁷ suggested a motor role for palisade endings; more recently, Lukas et al. ²⁰ proposed a sensory/motor function. 

Surprisingly, we recently showed in cats and macaque monkeys that palisade endings are supplied by cholinergic axons, and that the palisade complexes are cholinergic as well. ^{11 14 15} In some cases, we determined that nerve fibers supplying palisade endings establish neuromuscular contacts outside the palisade complex. ^{11 14 15} In monkeys, such additional neuromuscular contacts were observed in 30% of the cases studied. ¹¹ Our recent findings ^{11 14 15} have reopened the debate about the functional significance of palisade endings and have advanced the question: Are palisade endings cholinergic sensory structures or effector organs involving collagen fibrils? The presence of nerve terminals in apposition to collagen fibrils seems to argue for sensory as opposed to motor function. However, in this case, the additional neuromuscular contacts established by the nerve fiber that supplies palisade endings would have to be interpreted as sensory as well. 

In this study, we have continued our analysis on palisade endings in monkeys to complement our prior findings on this EOM-specific organ in a primate species. Specifically, we labeled palisade endings with all commercially available cholinergic markers, including antibodies against choline acetyltransferase (ChAT), vesicular acetylcholine transporter (VAChT), and choline transporter (ChT). In the nervous system, ChT is used for the uptake of choline and ChAT synthesizes acetylcholine, which is packed into vesicles by VAChT. To distinguish between sensory and motor terminals, we used α-bungarotoxin, a form of snake venom that binds to post-synaptic nicotinic acetylcholine receptors and is widely used to detect motor terminals in vertebrate skeletal muscles including EOMs. ^{28 29} For our analysis, we used various techniques including three dimensional (3D) reconstructions, confocal laser scanning microscopy (CLSM), immuno light microscopy (LM), and conventional/immuno transmission electron microscopy (TEM). Here, we provide novel data indicating that palisade endings are effector (motor) organs. 

Materials and Methods

All animals used in this study were treated in accordance with the ARVO statement for the Use of Animals in Ophthalmic and Vision Research. 

Captive-bred cynomolgus monkeys (Macaca fascicularis, n = 10), five females and five males, their age varying between 4 and 6 years and their weight between 2.2 and 4.1 kg, were used for this study. The eyeballs including the distal parts of the EOMs were obtained from Novartis Pharma AG (Basel, Switzerland). The eyeballs were immersion fixed with 4% paraformaldehyde (for CLSM and immuno LM/TEM) or with 4% paraformaldehyde and 2.5% glutaraldehyde (for conventional TEM) in 0.1 M phosphate buffer (PB; pH 7.4) for 24 hours. Then tissue was rinsed in 0.1 M phosphate buffered saline (PBS; pH 7.4). The eyeballs were dissected and the rectus EOMs as well as the oblique EOMs, including the distal tendons, were removed. The proximal tendons of the EOMs could not be analyzed because the muscles were cut frontal to their posterior attachments. The EOMs were further processed for CLSM, immuno LM, conventional TEM, and immuno TEM. 3D reconstructions were made from immunolabeled palisade endings. 

Wholemount Immunostaining and Confocal Laser Scanning Microscopy

The EOMs of five animals were analyzed by CLSM. Distal EOM myotendons (60) were divided into three groups, each group containing 20 EOMs including both rectus and oblique muscles. The whole EOM myotendons were labeled with: 1) phalloidin (counterstaining the muscle fibers), an antibody against ChAT and α-bungarotoxin; 2) phalloidin and antibodies against ChAT and ChT; and 3) phalloidin and antibodies against neurofilament and VAChT. The sources and working dilutions of phalloidin, α-bungarotoxin, primary antibodies, and secondary antibodies are listed in Table 1 . 

Labeling with Phalloidin, Anti-ChAT and α-Bungarotoxin.

Fixed tissue was rinsed in PBS, frozen in liquid nitrogen, and subsequently thawed in PBS containing 1% Triton X. After blocking the nonspecific binding sites in 10% donkey serum for 1 hour, wholemounts were incubated in the primary antibody goat anti-ChAT at 20°C and in darkness for 48 hours. Wholemounts were extensively rinsed in PBS and then incubated in the secondary antibody (conjugated donkey anti-goat, Alexa Fluor 488; Invitrogen, Carlsbad, CA) and diluted in PBS containing 1% Triton X for in darkness for another 4 hours. After rinsing again, specimens were incubated overnight in a mixture containing rhodamine conjugated α-bungarotoxin and conjugated phalloidin (Alexa Fluor 633; Invitrogen). Finally, wholemounts were rinsed once more and mounted (Citifluor; Agar Scientific Ltd., Stansted, Essex, UK). 

Labeling with Phalloidin, Anti-ChAT, and Anti-ChT.

Wholemounts were processed using the method described above. After freezing and thawing, wholemounts were blocked in 10% donkey serum and incubated in a mixture of the primary antibodies goat anti-ChAT and mouse anti-ChT. Then, wholemounts were incubated in the secondary antibodies (Alexa Fluor 488 conjugated donkey anti-goat and rhodamine conjugated donkey anti-mouse) in each antibody for 4 hours at 20°C. After phalloidin incubation, wholemounts were mounted (Citifluor; Agar Scientific Ltd.). Between the incubation steps, specimens were extensively rinsed. 

Labeling with Phalloidin, Anti-Neurofilament and Anti-VAChT.

Frozen and thawed wholemounts were blocked in 10% donkey serum and incubated in the primary antibodies rabbit anti-neurofilament and goat anti-VAChT (Santa Cruz Biotechnology, Santa Cruz, CA). Then, specimens were labeled with the secondary antibodies rhodamine conjugated donkey anti-rabbit, conjugated donkey anti-goat (Alexa Fluor 488; Invitrogen), and phalloidin. Finally, wholemounts were mounted (Citifluor; Agar Scientific Ltd.). 

Triple-labeled EOM wholemounts were examined under a confocal laser scanning microscope (Zeiss LSM 410; Zeiss, Oberkochen, Germany). When palisade endings were identified, series of longitudinal virtual sections of 0.4 to 1 μm thickness were cut through the specimens. Each section was photo-documented and subsequently 3D projections were calculated using imaging software (Zeiss LSM Image Examiner; Zeiss). 

3D Reconstruction

3D reconstructions were made of palisade endings labeled with phalloidin, anti-ChAT and α-bungarotoxin or alternatively with phalloidin, anti-neurofilament and anti-VAChT. Wholemounts were analyzed by CLSM and series of virtual longitudinal sections were cut through the palisade endings. Each section was photo-documented and from these image stacks the structures of interest were segmented by the use of a dynamic threshold algorithm using imaging software (Photoshop CS 8.0.1; Adobe, San Jose, CA). An individual threshold for one of the structures of interest was set once by the operator in a single sample image and was then automatically adjusted to all other images by a macro. The segmentation procedure resulted in multiple series of binary (black and white) image files, each containing one of the structures of interest (muscle fiber, nerve fiber, palisade ending). 3D models of the required structures were generated by recombining the respective sets of binary image files into a single 3D object using information clustering software (Velocity 4.2; Vivisimo, Pittsburgh, PA) running on a Mac computer (Power Mac G5; Apple, Cupertino, CA). The sectional outlines of the respective structures were placed in a z-distance corresponding to the sectional thickness. Polygonal surfaces were then generated using a marching cube algorithm. The final models were visualized either alone or as combinations of solid or transparent objects. 

Conventional Electron Microscopy

After immersion fixation was completed, EOMs containing the tendon were cut longitudinally into small strips. Specimen were postfixed in 1% osmium tetroxide, dehydrated in graded solutions of alcohol and embedded in resin (Epon; Hexion Specialty Chemicals, Houston, TX). Semithin cross-sections were cut through the tissue blocks and examined in the light microscope. When palisade endings were identified, ultrathin sections were cut at appropriate intervals. Sections were mounted on dioxane formvar-coated (Formvar; SPI-Chem, West Chester, PA) copper grids and stained in a 2% uranyl acetate solution followed by 0.4% lead citrate solution. Sections were analyzed with a transmission electron microscope (Zeiss EM 10; Zeiss). 

Immuno Light/Immuno Electron Microscopy

For immunolabeling we used the pre-embedding method. After immersion fixation, EOMs including the tendons were rinsed in PBS containing 0.1% Tween 20. Then, specimens were frozen and thawed. To inhibit the endogenous peroxidase, wholemounts were incubated in 0.05 M tris buffered saline pH 7.4 (TBS) containing 0.05% phenylhydrazine. After blocking in 10% rabbit serum, tissue was incubated in the primary antibody goat anti-ChAT (1:100) for 48 hours at 20°C followed by the secondary antibody horseradish peroxidase conjugated rabbit anti-goat (1:200; Chemicon International Inc., Temecula, CA). After antibody labeling was completed, wholemounts were immersion fixed in 2% PFA and 0.2% glutaraldehyde for superior preservation of ultrastructure. For enzymatic detection of horseradish peroxidase, specimens were incubated for 30 minutes in 3,3′ diaminobenzidine tetrahydrochloride (DAB) at 0.5% including 0.01% H₂O₂. The reaction was stopped by rinsing specimens in distilled water. EOMs were cut into small strips, postfixed in 1% osmium tetroxide, and embedded in Epon (Hexion). 

Semithin cross-sections were cut from the tendon toward the muscle and examined in the light microscope. ChAT-labeled axons were identified by the brown deposits of the DAB reaction product. ChAT-positive axons were traced in serial semithin sections and when palisade endings were identified by LM, ultrathin cross-sections were cut. These sections were mounted on dioxane formvar-coated copper grids, stained with in 2% uranyl acetate solution followed by 0.4% lead citrate solution, and analyzed under a transmission electron microscope (Zeiss EM 10; Zeiss). Alternatively, tissue blocks were re-orientated and semithin longitudinal sections were cut. When palisade endings were identified, ultrathin longitudinal sections were cut and stained as described above. 

Control Experiments

In negative controls, the primary antibodies were omitted and the secondary antibodies were used exclusively. In all cases, the omission of the primary antibodies resulted in a complete lack of immunostaining. 

To demonstrate the specificity of the cholinergic markers, positive controls were performed. Cryostat sections of monkey EOM muscle bellies were labeled with α-bungarotoxin and then either with antibodies against ChT, ChAT, or VAChT. Muscle fibers were counterstained with phalloidin. Examinations by CLSM demonstrated that α-bungarotoxin positive motor endplates were also positive for ChT, ChAT, and VAChT. 

Results

Number of Palisade Endings

We observed palisade endings in the distal myotendons of all types of EOMs. The number of palisade endings was counted in three medial recti and in three superior oblique. In the medial recti we counted between 50 and 73 palisade endings, and in the superior oblique, between 35 and 45. Generally, the number of palisade endings was higher in the rectus EOMs than in the oblique EOMs. 

Morphology of Palisade Endings

3D Reconstruction and Confocal Laser Scanning Microscopy.

In 3D reconstructions and immunolabeled wholemounts, the overall morphology of palisade endings became apparent. We observed nerve fibers that came from the muscle belly and extended for variable distances into the tendon. The diameter of these nerve fibers varied from 2 to 4 μm. Within the tendon the nerve fibers made a 180° loop and returned to the muscle. At the muscle-tendon junction, the returning axons divided into preterminal branches which surrounded single muscle fiber tips. Preterminal axons established nerve terminals in the tendon and around the muscle fiber tips. Such a neural specialization at the junction of a muscle fiber with the tendon represents the principle of a palisade ending. Palisade endings were usually supplied by a single axon with few exceptions in which two axons contributed to supply palisade endings (Figs. 1 2 3 and 4) . 

In immunolabeled EOM wholemounts, the nerve fibers supplying palisade endings were traced backward. In about one-fourth of the cases, we observed that the nerve fibers forming palisade endings established numerous delicate neuromuscular contacts outside the palisade/muscle fiber complex. Such additional neuromuscular contacts were found either on the same muscle fiber or on the neighboring muscle fiber (Figs. 2 and 3) . In other cases, the axons supplying palisade endings came from far away and when we traced the axons, we observed that they intermingled with others. Within such an axonal bundle it was not possible to trace individual axons further. However, in such cases we did not observe other nerve fibers which establish nerve terminals on the muscle fibers associated with palisade endings. 

Conventional Electron Microscopy.

We analyzed the fine structure of 50 palisade endings in different EOMs. Muscle fibers associated with palisade endings had a capsule sleeve that continued forward into the tendon. The capsule consisted of two to five layers of fibrocytes. Nerve fibers forming palisade endings penetrated the capsule at the level of the tendon. In the tendinous compartment, the axon lost its perineural envelope and myelin sheath. The nerve fiber divided into preterminal axons, which were directed toward the muscle fiber tip. Preterminal axons were completely encircled by Schwann cells, which were covered with a basal lamina. Along their further course, preterminal axons exhibited repeated expansions that obviously represented nerve terminals on the neighboring collagen fibrils. These so-called neurotendinous contacts were only partly covered with Schwann cells, whereas in the Schwann cell free areas, only a basal lamina was interposed between the axolemma and the collagen. Neurotendinous contacts were found in each palisade ending and all contained mitochondria and were full of clear vesicles (Fig. 5) . 

In 20 of 50 palisade endings, we observed that preterminal axons established contacts on the muscle fiber surface. Such neuromuscular contacts were found on the fingerlike muscle fiber processes that attach the muscle fiber to the tendon. Neuromuscular contacts were indistinguishable in their cytoplasmic content from neurotendinous contacts, and contained mitochondria and clear vesicles. The synaptic cleft of the neuromuscular contacts was filled with a basal lamina and the width measured between 90 and 110 nm. In very few neuromuscular contacts, we detected areas in which the basal lamina was dissolved, resulting in a narrowing of the synaptic cleft to 50 nm (Fig. 6) . 

Summing up, the fine structural analysis of 50 palisade endings showed that in 60% (30 of 50) of the cases, palisade endings had established exclusively neurotendinous contacts. In 40% of the cases (20 of 50), we found that palisade endings established neurotendinous contacts as well as neuromuscular contacts. 

Molecular Characteristics of Palisade Endings

Confocal Laser Scanning Microscopy.

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

In wholemounts labeled with phalloidin, anti-ChAT, and α-bungarotoxin, we observed ChAT-immunoreactive nerve fibers forming palisade endings. In some cases, we detected that ChAT-positive nerve fibers forming palisade endings also established multiple nerve contacts on the same muscle fiber or neighboring muscle fiber. These additional neuromuscular contacts were ChAT/α-bungarotoxin-positive (Figs. 2A 2B) . 

The palisade endings, including preterminal axons and palisade nerve terminals, were ChAT-positive. In about two-thirds of the palisade endings, we observed exclusively neurotendinous contacts which were ChAT-positive but α-bungarotoxin-negative (Fig. 2A) . In about one-third of the palisade endings, we found neuromuscular contacts as well. Neuromuscular contacts exhibited ChAT/α-bungarotoxin-reactivity (Figs. 2B 2C) . The ratio of palisade endings exhibiting neuromuscular contact matches our findings from TEM, in which 40% of the palisade endings had neuromuscular contacts. 

In wholemounts labeled with phalloidin, anti-neurofilament, and anti-VAChT, we found neurofilament positive nerve fibers supplying palisade endings. We also observed additional neuromuscular contacts exhibiting VAChT-immunoreactivity either on the same or neighboring muscle fiber (Fig. 3C) . Within the palisade ending, the nerve terminals were VAChT-immunoreactive as well (Figs. 3A 3B 3C) . 

In wholemounts labeled with phalloidin, anti-ChAT, and anti-ChT, we detected ChAT-positive nerve fibers forming palisade endings. Palisade nerve terminals exhibited ChAT/ChT-immunoreactivity (Fig. 4) . Unfortunately, we did not find additional neuromuscular contacts due to the fact that nerve fibers forming palisade endings intermingled with others and could not be traced any further. 

Immuno Light Microscopy and Immuno Electron Microscopy.

We analyzed 30 palisade endings of different EOMs. By LM, ChAT-labeled axons were identified by the brown deposits of the DAB reaction product. By TEM, ChAT-immunoreactivity was identified by the electron-dense deposits of the DAB reaction product. 

ChAT-positive axons formed palisade endings at the muscle fiber tips. In the tendinous compartment of the palisade endings, the ChAT-immunoreactive axons divided into preterminal axons sheathed by Schwann cells. Preterminal axons established ChAT-positive neurotendinous contacts that were partly invested with Schwann cells; at the point of contact, only a basal lamina was interposed between the axolemma and the collagen fibrils (Figs. 7 8) . As demonstrated by TEM, the internal structure of the neurotendinous contacts was partly masked by the dark DAB-reaction product. Whereas mitochondria were visible, clear vesicles were completely covered. ChAT-positive neurotendinous contacts were observed in each palisade ending. In 11 of 30 palisade endings, we observed ChAT-positive neuromuscular contacts as well. Such contacts were found on the finger-like muscle fiber processes attaching the muscle fiber to the tendon (Figs. 7 8) . Neuromuscular contacts had a basal lamina in the synaptic cleft (Fig. 8) . Importantly, we did not find ChAT-negative nerve terminals in the palisade endings. 

Motor Terminals

We analyzed the structural and molecular characteristics of motor terminals on singly innervated muscle fibers (SIFs) and multiply-innervated muscle fibers (MIFs) and compared them with palisade nerve terminals. Observations by TEM showed that en plaque motor endplates on SIFs and en grappe motor terminals on MIFs contained mitochondria and clear vesicles and had a basal lamina in the synaptic cleft. Whereas en plaque motor endplates exhibited deep folding of the subsynaptic membrane, en grappe terminals showed only shallow subsynaptic folding. The size of en plaque motor endplates varied between 30 and 50 μm and that of en grappe terminals between 5 and 8 μm. En plaque motor endplates and en grappe motor terminals bound α-bungarotoxin and immunohistochemically, they exhibited ChAT/VAChT/ChT reactivity (Figs. 9 10) . The morphologic and molecular characteristics of motor terminals on SIFs and MIFs and palisade nerve terminals are summarized in Table 2 . 

Discussion

We have recently demonstrated in cats ^{14 15} and monkeys ¹¹ that palisade endings are cholinergic. These novel findings have reopened the debate about the functional significance of palisade endings and have advanced the question: Are palisade endings cholinergic sensory organs or effector organs involving collagen fibrils? We have extended our prior finding on palisade endings in monkeys. ¹¹ Herein, we show that nerve fibers supplying palisade endings establish α-bungarotoxin-positive neuromuscular contacts outside the palisade complex; that all palisade endings exhibit ChT/ChAT/VAChT-immunoreactivity; and that neuromuscular contacts, when present in palisade endings, exhibit features of motor terminals and are α-bungarotoxin-positive as well. The implications of these findings with respect to the function of palisade endings are discussed below. 

In line with our prior CLSM findings, we observed that ChAT-immunoreactive nerve fibers supply palisade endings in monkey and also establish multiple neuromuscular contacts outside the palisade complex. ¹¹ Such additional neuromuscular contacts were observed either on the same muscle fiber of the palisade/muscle fiber complex or—in a new finding—on a neighboring muscle fiber. Additional neuromuscular contacts were found in about 25% of the palisade endings, which is slightly less (30% of the palisade endings) than in our prior study ¹¹ on monkeys. Concordant with our prior study, ¹¹ we showed that additional neuromuscular contacts are ChAT-immunoreactive. Here, we further demonstrated that these contacts are VAChT-immunoreactive and, more importantly, α-bungarotoxin-positive as well. α-Bungarotoxin binding proves that additional neuromuscular contacts are motor; this finding has a direct consequence with respect to palisade endings. In fact, palisade endings arising from axons that supply motor neuromuscular contacts in other locations would have to be interpreted as motor as well. 

By TEM, we demonstrated that each palisade ending in monkey establishes nerve terminals targeting collagen fibrils. In 40% of the palisade endings, we observed nerve terminals targeting the muscle fibers as well. Such neurotendinous and neuromuscular contacts contain mitochondria and clear vesicles. Neuromuscular contacts in palisade endings of monkeys usually have a basal lamina in the synaptic cleft, which is a defining feature of motor terminals. ³⁰ We observed that the basal lamina is discontinuous in only a few neuromuscular contacts. Such an interruption of the basal lamina was also detected in motor terminals on muscle fibers of rat EOMs. ³¹ By immunohistochemistry, we confirmed our previous findings ¹¹ that neurotendinous and neuromuscular contacts in palisade endings of this primate species are ChAT/VAChT immunoreactive; here we demonstrated that they are ChT-immunoreactive as well. Neuromuscular contacts, when present in palisade endings, also exhibit α-bungarotoxin staining. The co-localization of ChT, ChAT, and VAChT demonstrates that neurotendinous and neuromuscular contacts in palisade endings contain all components for the synthesis of acetylcholine. α-Bungarotoxin-binding shows that neuromuscular contacts in palisade endings have nicotinic acetylcholine receptors. Applying morphologic and molecular criteria, the present study proves that neuromuscular contacts in palisade endings are definitely motor. In neurotendinous contacts of palisade endings, α-bungarotoxin staining, which is typical for motor terminals, is absent. On the other hand, we provided evidence that the palisade endings themselves are motor structures and in this case, palisade neurotendinous contacts would have to be interpreted as motor as well. 

Taken together, the present study confirms our prior findings that palisade endings are cholinergic and provides novel data clearly indicating that palisade endings are cholinergic motor and not cholinergic sensory. 

A major argument to classify palisade endings as putative effectors is based on the finding that nerve fibers supplying palisade endings also supply motor neuromuscular contacts outside the palisade complex. Such additional neuromuscular contacts were observed in 25% of the cases, and in fact, it is a critical question whether all palisade endings have these contacts. To detect additional neuromuscular contacts, the nerve fibers supplying palisade endings had to be traced over a long distance, and often the axons intermingled with others and could not be traced any further. In such cases we did not observe another nerve fiber that establishes neuromuscular contacts on a muscle fiber associated with a palisade ending, and we assume that the axon supplying palisade endings might establish additional neuromuscular contacts more proximal. Nevertheless, due to technical difficulties, we cannot conclude with certainty whether additional neuromuscular contacts of nerve fibers supplying palisade endings are a general feature. It is, however, difficult to imagine that palisade endings in which such a construction could not be confirmed represent another population of palisade endings. If this were indeed the case, we would have to distinguish between two categories of palisade endings. 

With respect to innervation, EOMs in mammals have two kinds of muscle fibers, SIFs and MIFs. SIFs have a single motor endplate (en plaque ending), whereas MIFs receive multiple neuromuscular contacts (en grappe endings) throughout their length. ^{32 33} Palisade endings are exclusively found on the tip of a particular muscle fiber that is the MIF of the global EOM layer. ^{12 21 23} There can be no doubt that the additional neuromuscular contacts ascertained in palisade endings of the present study are en grappe endings. We therefore conclude that palisade endings arise from axons that also supply MIFs. Interestingly, palisade endings were either found on the tip of the same muscle fiber innervated by the MIF motoneurons or, alternatively, on a neighboring muscle fiber, indicating a variability of the MIF motoneuron/palisade ending unit. 

To date, physiological investigations on palisade endings are completely missing; their morphology, however, is well described. Ultrastructural investigations which focused on palisade nerve terminals have presented anatomic evidence to classify palisade endings as sensory organs. Specifically, in palisade endings of cats, ^{12 15} sheep, ¹⁸ and humans, ²⁰ and in Ruskell’s ¹⁰ earlier study on palisade endings in rhesus monkeys, neurotendinous contacts were constantly observed and, despite the presence of clear vesicles, nerve terminals in apposition to collagen fibrils seem to point to a sensory function. With the exception of humans, neuromuscular contacts in palisade endings of these species lack a basal lamina in the synaptic cleft, a feature common with sensory nerve terminals on intrafusal muscle fibers of muscle spindles. ^{34 35 36 37} Surprisingly, there are differences between Ruskell’s ¹⁰ findings and ours with respect to neuromuscular contacts in palisade endings of monkeys. In Ruskell’s study, ¹⁰ neuromuscular contacts appear morphologically sensory-like, whereas in the present study neuromuscular contacts are morphologically motor-like, which is confirmed by α-bungarotoxin binding. By TEM, we analyzed 50 palisade endings of different EOMs in detail and it is extremely unlikely that we missed morphologically sensory-like neuromuscular contacts. At the moment, the discrepancies between Ruskell’s ¹⁰ and our study regarding neuromuscular contacts in palisade endings of the same species are not explicable. 

The most compelling argument that palisade endings are sensory structures has come from Billig et al., ¹³ who injected neuronal tracer into the trigeminal ganglion, which is presumed to exclusively contain cell bodies of afferent nerve fibers. In cats, Billig et al. ¹³ found three kinds of labeled nerve endings, one resembling palisade endings. Recently, Wang et al. ²⁶ provided indirect evidence that palisade endings are sensory. Using rhesus monkeys, Wang and colleagues ²⁶ recorded eye position signals from the contralateral side in the primary somatosensory cortex. Since muscle spindles and Golgi tendon organs are rare or absent in monkey EOMs and palisade endings are numerous, the authors concluded that the signals arise from the palisades. ^{3 26 38}  

Despite the impressive support in the literature that palisade endings are sensory organs, the present report challenges this view for three reasons. First, nerve fibers supplying palisade endings also supply motor neuromuscular contacts. Second, palisade endings contain acetylcholine, the neurotransmitter of motor terminals. Third, neuromuscular contacts in palisade endings are endowed with nicotinic acetylcholine receptors, which are otherwise present in motor terminals. Indication that palisade endings are putative effectors was provided by α-bungarotoxin binding in palisade endings of humans ²⁰ and cats. ^{14 15} Specifically, in palisade endings of humans, neuromuscular contacts are α-bungarotoxin-positive, and in palisade endings of cats, the sparse neuromuscular contacts are α-bungarotoxin positive as well. ^{14 15 20} Further evidence that palisade endings are putative effectors comes from a nerve degeneration experiment. Sas and Schab ²⁷ made lesions of the oculomotor nuclei and showed that in addition to the expected loss of motor terminals on the EOMs, palisade endings were degenerated. The authors concluded that palisade endings are supplied by axons which originate from the EOM motor nuclei. ²⁷  

Functional Considerations

In the present study, we ascertained that palisade endings are supplied by axons that also supply MIFs. The palisade endings are associated with the same muscle fibers innervated by MIF motoneurons or, alternatively, with neighboring muscle fibers. MIF motoneurons establish en grappe endings and on activation, such motoneurons induce local contractions of the muscle fiber at the site of the nerve terminals. ^{39 40 41 42} But what can be the function of palisade endings arising from MIF motoneurons? 

On activation, MIF/palisade motoneurons would excite en grappe endings as well as palisade endings, including palisade neurotendinous contacts and, when present, palisade neuromuscular contacts at the muscle fiber tip. After neurotransmitter release, en grappe endings would elicit focal contractions of the muscle fiber body, whereas palisade neuromuscular contacts would elicit contraction of the most terminal part of the muscle fiber, either of the same or neighboring muscle fiber. Neurotendinous contacts in palisade endings are surrounded by collagen fibrils and it is unclear what an effect neurotransmitter release could have on the tendon. With respect to their position, palisade neurotendinous contacts lie in serial to the muscle fibers, which is analogous to nerve terminals in Golgi tendon organs. ⁴³ Thus, neurotendinous contacts in palisade endings are ideally located to register muscle fiber contraction and could, theoretically, function as sensory endings. However, in this case the nerve fibers supplying palisade endings would have to conduct information in two directions: from the central nervous system to the palisade neuromuscular contacts (efferent) and, in the opposite direction, from palisade neurotendinous contacts to the central nervous system (afferent). In fact, such an idea is highly speculative and at present it is too early to determine the functional sense of palisade endings. Nevertheless, based on their frequency, morphology, and molecular characteristics, there can be no doubt that palisade endings are functional. 

It is unlikely that palisade endings receive a double innervation arising from a motor and a sensory nerve fiber. First, the present study shows that, with few exceptions, palisade endings are supplied by a single axon. Second, by double-labeling of palisade endings with a general marker for nerve fibers (antibody against neurofilament) and a marker for cholinergic nerve fibers (antibody against ChAT) all axons of the palisade complexes exhibit ChAT-immunoreactivity. ^{11 14 15} Third, by immuno TEM, we observed that all palisade nerve terminals are ChAT-positive. Fourth, by labeling palisade endings with substance P, which is present in some sensory axons, no staining was detected at all in palisade endings of humans, monkeys, and rats. ^{44 45}  

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 area 17 and 18, the Clare Bishop area, the frontal cortex, and the somatosensory cortex. ^{7 26} Muscle spindles and Golgi tendon organs are absent in the EOMs of most mammalian species, whereas palisade endings, so far, have been found in each species investigated. Therefore, it has been suggested that palisade endings could be the source of proprioceptive input to the central nervous system. The findings of the present study call this view into question. Palisade endings are present in human EOMs and due to their localization at the myotendinous junction, a finding of particular interest for strabismus surgeons. Observations of patients after strabismus surgery indicate alteration in spatial perception, and it is supposed that the damage of the palisade endings during the surgical procedure could be the reason. ^{1 2} The results of the present study could indicate that surgical procedures to treat strabismus would have no side effects with respect to eye position signals. 

 Supported by Grant P15478-B06 from the Fonds zur Foerderung der Wissenschaftlichen Forschung.

 Submitted for publication August 19, 2008; revised September 29, 2008; accepted January 19, 2009.

 Disclosure: R. Blumer, None; K.Z. Konakci, None; C. Pomikal, None; G. Wieczorek, None; J.-R. Lukas, None; J. Streicher, 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.

Table 1.

 

 

List of the Markers for Muscle Fibers, Neurotoxin, Primary Antibodies, and Secondary Antibodies, as well as Working Dilution and Sources

Table 1.

 

List of the Markers for Muscle Fibers, Neurotoxin, Primary Antibodies, and Secondary Antibodies, as well as Working Dilution and Sources

Marker for Muscle Fibers	Neurotoxin	Primary Antibodies	Secondary Antibodies
AlexaFluor 633 conjugated phalloidin (1:80)	α-Bungarotoxin rhodamine conjugated (1:500)	Goat anti choline acetyltransferase (1:100); mouse anti-choline transporter (1:200); rabbit anti-neurofilament (1:200); goat anti-VAChT (1:100)	Donkey anti-goat AlexaFluor 488 (1:500); donkey anti-mouse rhodamine (1:400); donkey anti-rabbit rhodamine (1:200)

Figure 1.

 

3D reconstructions of palisade endings from virtual CLSM sections. (A) The nerve fibers are labeled with anti-neurofilament (green), the palisade nerve terminals with anti-VAChT (red), and the muscle fibers with phalloidin (white). The tendon not labeled continues the muscle fiber tip to the right. A nerve fiber coming from the muscle extends into tendon. There, the nerve fiber turns back and divides into preterminal axons to form a palisade ending around the muscle fiber tip. The palisade ending establishes nerve terminals on the tendon and around the muscle fiber tip. (B) Nerve fibers are labeled with anti-ChAT (green), nicotinic acetylcholine receptors in the muscle fiber membrane with α-bungarotoxin (red), and the muscle fiber with phalloidin (white). The tendon not labeled continues the muscle fiber tip to the right. Nerve fibers coming from the muscle form a palisade ending around a muscle fiber tip. Exclusively palisade nerve terminals contacting the muscle fibers are labeled. A CLSM image of this palisade ending is shown in Figure 2D . Scale bars, 100 μm.

Figure 2.

 

CLSM images of palisade endings. Nerve fibers are labeled with anti-ChAT (green), nicotinic acetylcholine receptors with α-bungarotoxin (red), and muscle fibers with phalloidin (white). The tendon not labeled continues the muscle fiber tip to the right. Neuromuscular contacts which exhibit ChAT/α-bungarotoxin reactivity appear yellow in the overlay. (A) A ChAT-positive nerve fiber running alongside the muscle fiber forms palisade ending at the muscle fiber tip. Outside the palisade complex the nerve fiber establish multiple, ChAT/α-bungarotoxin-positive neuromuscular contacts (arrows) on the same muscle fiber. (Details of the neuromuscular contacts are shown in the insets.) The preterminal axons and the nerve terminals of the palisade ending are ChAT-positive as well. The palisade nerve terminals exhibit no α-bungarotoxin-binding, indicating that no neuromuscular contacts are present. (B) ChAT-positive nerve fiber forming a palisade ending at a muscle fiber tip and establishing ChAT/α-bungarotoxin-positive neuromuscular contacts on a neighboring muscle fiber outside the palisade complex. The palisade ending establishes neuromuscular contacts that are ChAT/α-bungarotoxin-positive. (C) Palisade endings with neuromuscular contacts exhibiting ChAT/α-bungarotoxin-reactivity. A detail of another palisade ending with a single ChAT/α-bungarotoxin-positive neuromuscular contact is shown in the inset. Scale bars, 100 μm.

Figure 3.

 

CLSM images of palisade endings. Nerve fibers are labeled with anti-neurofilament (red), nerve terminals with anti-VAChT (green), and muscle fibers with phalloidin (white). The tendon not labeled continues the muscle fiber tip to the right. (A) A neurofilament-positive nerve fiber forms a palisade ending at a muscle fiber tip. Outside the palisade complex the nerve fiber establishes VAChT/neurofilament-positive neuromuscular contacts on the same and on a neighboring muscle fiber. Within the palisade complex, preterminal axons are neurofilament-positive whereas palisade nerve terminals exhibit VAChT/neurofilament-immunoreactivity. (B) Only the final part of the nerve fiber is shown that runs within the tendon and then curves back to supply a palisade ending exhibiting VAChT/neurofilament-positive palisade nerve terminals. (C) Two neurofilament-labeled nerve fibers running alongside a muscle fiber and establishing VAChT/neurofilament-positive neuromuscular contacts on a neighboring muscle fiber outside the palisade complex. Both nerve fibers supply a palisade ending on the same muscle fiber and also extend further into the tendon. Palisade nerve terminals are VAChT/neurofilament-positive. Scale bars, 100 μm.

Figure 4.

 

CLSM images of palisade endings. Nerve fibers are labeled with anti-ChAT (green), nerve terminals with anti-ChT (red), and muscle fibers with phalloidin (white). The tendon not labeled continues the muscle fiber tip to the right. (A, B) ChAT-immunoreactive nerve fibers forming palisade endings at the muscle fiber tip. Preterminal axons of the palisade endings are ChAT-immunoreactive as well. The palisade nerve terminals exhibit ChAT/ChT-immunoreactivity. Scale bars, 100 μm.

Figure 5.

 

Micrographs from conventional transmission electron microscopy. (A) Micrograph at low magnification showing a cross-section through a palisade ending at the level of its tendinous compartment. The palisade ending is sheathed by a thin capsule (C) of fibrocytes. The myelinated nerve fiber (N) supplying the palisade ending has not yet penetrated the capsule and resides outside. Preterminal axons (arrows), fibrocytes (F), and collagen fibrils (COL) are inside the capsule. (B, C) Micrographs at high magnifications showing cross-sections through palisade nerve terminals (T) contacting the COL. Such neurotendinous contacts are only partly invested by Schwann cells (S) which are covered with a basal lamina (arrows). Areas of the terminals contacting the collagen are free from S cells and are indicated with arrowheads. The nerve terminals contain mitochondria and clear vesicles. Scale bars: (A) 10 μm; (B, C) 1 μm.

Figure 6.

 

Micrographs from conventional transmission electron microscopy. (A) Micrograph at low magnification showing a cross-section through a palisade ending at the level of the muscle fiber/tendon attachment. The muscle fiber (MF) exhibits processes that are separated by COL. Palisade nerve terminals (T) contact the muscle fiber surface. (B, C) Micrographs at high magnifications showing cross-sections through T contacting the MF. A basal lamina (arrow) fills the synaptic cleft. The neuromuscular contacts contain mitochondria and clear vesicles. (B, inset) Detail of a neuromuscular contact. (D) Micrograph at high magnification showing a cross-section through a palisade nerve terminal contacting the muscle fiber with the synaptic cleft only partly filled with a basal lamina (arrow). Other areas of the synaptic cleft lack a basal lamina investment (arrowheads). Scale bar: (A) 10 μm; (B, C, D) 1 μm.

Figure 7.

 

Micrographs from immuno light microscopy. Nerve fibers and nerve terminals are labeled with anti-ChAT. After the DAB-reaction, ChAT-positive neurons appear brown in color. (A) Semithin cross-section through the tendinous compartment of a palisade ending showing ChAT-positive preterminal axons (arrowheads) and a ChAT-positive palisade nerve terminal (arrow). (B) Semithin longitudinal section through a palisade ending. Palisade nerve terminals (arrowheads) contacting the collagen fibrils are visible. Other palisade nerve terminals (arrows) contact the muscle fiber processes that attach the MF to the tendon. On a neighboring muscle fiber a ChAT-positive motor terminal (asterisk) is visible. Scale bars, 10 μm.

Figure 8.

 

Micrographs from immuno transmission electron microscopy. Nerve fibers and nerve terminals are labeled with anti-ChAT and are identified by the electron-dense DAB-reaction product. (A) Micrograph at low magnification showing a cross-section through a palisade ending at the level of its tendinous compartment. The N, which supplies the palisade ending, is outside the C. The myelinated nerve fiber exhibits a weaker ChAT-immunoreactivity than the preterminal axons (arrows) inside the capsule. Inset shows a detail of a ChAT-positive preterminal axon. (B, C) Micrographs at high magnification showing ChAT-positive T contacting Col. Neurotendinous contacts are only partly invested with S cells. Areas of the terminals contacting the collagen are without an S cell and are indicated with arrowheads. Basal lamina (arrow). (D, E) Micrographs at high magnification showing a cross-section (D) and a longitudinal section (E) through T contacting the MF. Neuromuscular contacts exhibit ChAT-immunoreactivity. A basal lamina (arrow) fills the synaptic cleft. Scale bars: (A) 10 μm; (B, C, D, E) 1 μm.

Figure 9.

 

CLSM images showing singly-innervated muscle fibers with en plaque motor endplates and multiply-innervated muscle fibers with en grappe motor terminals. (A, B) Nerve fibers are labeled with anti-ChAT (green), motor terminals with α-bungarotoxin (red), and muscle fibers with phalloidin (white). (A) A ChAT-positive axon supplying an en plaque endplate that is positive for ChAT/α-bungarotoxin; (B) ChAT-positive axon supplying en grappe endings positive for ChAT/α-bungarotoxin; (C, D) Motor terminals are labeled with anti-ChT (green) and α-bungarotoxin (red), muscle fibers with phalloidin (white). (C) En plaque endplate and (D) en grappe terminals showing ChT/α-bungarotoxin reactivity. (E, F) Motor terminals are labeled with anti-VAChT (green) and α-bungarotoxin (red), muscle fibers with phalloidin (white). (E) En plaque endplate and (F) en grappe terminals showing VAChT/α-bungarotoxin-reactivity. Scale bars, 100 μm.

Figure 10.

 

Micrographs showing singly-innervated muscle fibers with en plaque motor endplates and multiply-innervated muscle fibers with en grappe motor terminals. (A, B) Conventional electron microscopy showing (A) an en plaque motor endplate (MT) and (B) an en grappe motor ending (MT). En plaque and en grappe motor terminals contain mitochondria and vesicles. The synaptic cleft is filled with a basal lamina (arrow). (C, D) Nerve terminals are labeled with anti-ChAT. (C) Immuno light microscopy showing a ChAT-positive en plaque motor endplate. (D) Immuno electron microscopy showing an en grappe motor ending (MT). Basal lamina (arrow). Scale bars: (A, B, D) 1 μm; (C) 10 μm.

Table 2.

 

 

Morphological and Molecular Characteristics of Palisade Nerve Terminals and Motor Nerve Terminals on Singly-innervated Muscle Fibers (SIFs) and Multiply-innervated Muscle Fibers (MIFs)

Table 2.

 

Morphological and Molecular Characteristics of Palisade Nerve Terminals and Motor Nerve Terminals on Singly-innervated Muscle Fibers (SIFs) and Multiply-innervated Muscle Fibers (MIFs)

	Palisade Nerve Terminals			Motor Terminals
		Neurotendinous Contacts	Neuromuscular Contacts	En Plaque Endplates on SIFs	En Grappe Endings on MIFs
Clear vesicles	+	+	+	+
Basal lamina in the synaptic cleft	−*	+	+	+
ChT/ChAT/VAChT reactivity	+	+	+	+
α-Bungarotoxin-reactivity	−	+	+	+

* Neurotendinous contacts are covered by a basal lamina, but are not separated by a synaptic cleft from a neighboring structure.

 

The authors thank Marietta Lipowec, Regina Mayer, and Christiane Hanefl-Krivanek for their valuable technical assistance.