Abstract
purpose. To characterize human choroidal ganglion cells (CGCs) further, regarding their immunohistochemical and ultrastructural appearance and their pre- and postsynaptic connections.
methods. Choroidal wholemounts and serial sections of human donor eyes were stained with antibodies against neuronal nitric oxide synthase (nNOS), vasoactive intestinal peptide (VIP), tyrosine hydroxylase (TH), vesicular monoaminergic transporter (VMAT)-2, vesicular acetylcholine transporter (VAChT), neuropeptide Y (NPY), substance P (SP), calcitonin gene-related peptide (CGRP), calretinin, galanin, synaptophysin, and α-smooth muscle actin. Ultrathin sections of glutaraldehyde-fixed eyes were studied with an electron microscope.
results. All CGCs stained for nNOS, most for VIP, approximately 45% for calretinin, and only single neurons for NPY and galanin. Ultrastructurally, the CGCs showed an incomplete glial sheath and, in places, showed close contact to surrounding collagen fibrils. The CGCs were in close contact with numerous boutons staining for the different neuronal markers including synaptophysin, nNOS, VIP, NPY, TH, VMAT-2, VAChT, calretinin, and NPY.
conclusions. The data indicate a complex integrative function of CGCs. The immunohistochemical and ultrastructural characteristics also indicate that the CGCs may have mechanosensory properties. The complex synaptic information points to a specific regulative CGC function in parallel with ciliary muscle contraction (accommodation). Axons originating from CGCs mainly supply the choroidal vasculature, thus implicating the CGCs as vasodilative neurons, but single CGCs may also innervate other structures such as nonvascular choroidal smooth muscle cells.
It is well established that the uvea in most species is innervated by two parasympathetic pathways, namely by nerves deriving from the ciliary ganglion and by those deriving from the sphenopalatine ganglion.
Within the uvea of human eyes, an additional group of nitric oxide synthase (NOS) and vasoactive intestinal peptide (VIP)-immunoreactive (IR) neurons has recently been discovered.
1 2 3 Approximately, 2000 of these neurons are located in the inner portion of the ciliary muscle and presumably are involved in fine regulation of the accommodation.
3 In addition, approximately 2000 neurons are present in the choroid. Most of the postganglionic nerve fibers of these choroidal ganglion cells (CGCs) join the perivascular nerve fiber plexus that supports the vasodilative innervation of the choroidal vasculature.
1 2 A small number of these CGCs is located within the ciliary nerves. The postganglionic nerve fibers of this group of neurons do not enter the choroid but join the nerve fiber plexus that innervates the outflow system.
4
When different mammalian eyes were compared, substantial numbers of CGCs in the posterior choroid were found only in higher primates and human eyes containing a fovea centralis.
5 CGCs are also present in a number of birds with a specialized accommodative system and one or more foveae.
6 7 Further characterization of the CGCs in the duck revealed that in addition to NO and VIP, the CGCs express galanin. The postganglionic fibers of these cells innervate not only the choroidal vasculature, but also the numerous non–vascular smooth muscle cells (NVSMCs) present in the avian choroidal stroma.
8 Duck CGCs are surrounded by tyrosine hydroxylase (TH)/dopamine-β-hydrolase (DBH)–immunoreactive nerve fibers forming synaptic contacts with the CGCs.
9 Evidence has shown that these fibers derive from the superior cervical ganglion. The CGCs in the duck also receive calcitonin gene-related peptide (CGRP)–positive efferent collaterals of trigeminal afferents that may indicate precentral reflex arcs.
10 Innervation of CGCs by postsynaptic sympathetic fibers in the avian eye, however, indicates a more complex integrative function of the ganglion cell plexus, similar to that in the enteric nervous system.
9
It is not yet known whether CGCs in primate eyes have a similar complex function. In the present study we investigated human CGCs to clarify which neurotransmitters are expressed by these cells, to characterize their presynaptic input, and to define their target tissue.
Thirty-seven eyes of human donors ranging in age between 12 and 95 years were observed 11 to 36 hours after death. All eyes were incised equatorially and fixed either for 4 hours in neutral buffered 4% paraformaldehyde (PFA) or for 12 hours in Zamboni fixative, containing 4% PFA and 0.01% picric acid. The tissue was then rinsed in phosphate-buffered saline (PBS, pH 7.4) several times. In 27 eyes, the posterior segments were divided into four quadrants and wholemounts of the choroid and sclera were performed. From some of the wholemounts and from the remaining eyes, serial 14- to 16-μm-thick sagittal and tangential sections were cut through the choroid and sclera with a cryostat (Leica, Bensheim, Germany) and mounted on poly-l-lysine–coated glass slides.
Incubation with the primary antibody was performed overnight at room temperature, with the antibody diluted in PBS containing 1% bovine serum albumin and 0.1% Triton X. Control experiments were performed by incubating the sections only with the dilution solution. The primary antibodies used are listed in
Table 1 . The sections and wholemounts were then rinsed in PBS and incubated for 1 hour with an appropriate fluorescent-dye–conjugated secondary antibody, diluted in PBS (Dianova, Hamburg, Germany). In the double-staining procedure, the steps were the same, but the incubation time of the primary antibody was reduced to 4 to 6 hours.
Most of the antibodies showed sufficient staining in all specimens obtained, regardless of the donor age, time since death, or fixation. For some antibodies, certain special conditions must exist in the tissue to promote sufficient staining. Therefore, the number of tissue specimens was limited for these markers. The antibody against VIP showed sufficient staining of CGCs only in tissue fixed for 24 hours in Zamboni fixative and with a time since death of less than 24 hours. PFA-fixed specimens showed axonal VIP staining but no staining of the CGCs. Staining with the antibody against neuropeptide Y (NPY) was only possible by using cryosections of both PFA- and Zamboni-fixed specimens. Wholemounts did not show sufficient staining of either neurons or nerve fibers.
All sections were mounted with Kaiser glycerol jelly and viewed with a fluorescence microscope (Aristoplan; Leica) or with a confocal laser scanning system (MRC 1000; Bio-Rad, Munich, Germany) attached to an inverted microscope (Diaphot 300; Nikon, Tokyo, Japan). In estimating the size of single neurons, a diameter was defined as the shortest distance between opposing cell boundaries, when passing through the center of the cell.
The ultrastructural morphology and the surrounding tissue of the CGCs was investigated in 13 eyes of donors ranging in age between 56 and 91 years. The eyes were obtained 4 to 24 hours after death, incised equatorially, and fixed in 4% PFA and 1% glutaraldehyde for at least 24 hours.
Small samples of the choroid and sclera were prepared and embedded in Epon. Sagittal and tangential semithin serial sections (1 μm thick) were performed and stained with toluidine blue. From localized CGCs, ultrathin sections were performed, stained with leaded citrate and uranyl acetate, and viewed with a transmission electron microscope (model EM 902; Carl Zeiss Meditec, Oberkochen, Germany).
In human choroidal sections and wholemounts incubated with antibodies against neuronal NOS (nNOS), all 1820 visualized CGCs showed positive staining
(Fig. 1A) . No additional CGCs, either identified by their content of autofluorescent lipofuscin or by a nonspecific light fluorescent background in the neuronal cytoplasm, were detected in the nNOS-stained specimens. The nNOS staining was independent of the size (10–37 μm in diameter) and the location of the neurons. Numerous axons within the suprachoroidal nerve plexus and around choroidal vessels were also nNOS IR.
Staining with antibodies against VIP revealed that most of the CGCs (354/362 CGCs evaluated) were VIP IR. Colocalization was documented by double staining with antibodies against nNOS and VIP
(Fig. 1B) . It appeared that the few VIP-negative CGCs, visualized by autofluorescent lipofuscin or by a nonspecific light fluorescent background, were of smaller size (12–18 μm in diameter).
Less frequent IR in CGCs was seen in choroidal tissue stained with antibodies against calretinin. Wholemounts revealed that 82 of the 146 CGCs evaluated were calretinin IR
(Fig. 1C) . In some cases, these CGCs were preferentially of larger size (24–36 μm in diameter), but there were also other cases showing mainly small calretinin-positive cells (17–21 μm in diameter).
Only few small CGCs (14–18 μm in diameter) were NPY IR (7/152 evaluated CGCs). Single larger CGCs (25–34 μm in diameter) revealed faint staining with antibodies against galanin (3/189 evaluated CGCs). Most of the neurons in NPY- and galanin-stained sections, however, were unstained. There was no staining of CGCs with the antibodies against TH, DBH, vesicular monoaminergic transporter (VMAT)-2, vesicular acetylcholine transporter (VAChT), substance P (SP), and CGRP (total number of evaluated CGCs: 1340).
Ultrastructural investigations showed, that the CGCs appeared as typical neurons with a large nucleus containing light-appearing chromatin and a clear nucleolus and a cytoplasm containing ribosomes, stacks of rough endoplasmic reticulum, and mitochondria
(Fig. 2A) . In CGCs of aged donor eyes, lipofuscin granules were frequently seen within the cytoplasm of the neurons. numerous vesicle-filled boutons were present around the CGC, forming synaptic contacts with the neuron
(Fig. 2B) . The surface of the CGCs toward the surrounding connective tissue space was only incompletely covered by glial cell processes. In places, the basement membrane of the neurons directly contacted the adjacent collagen fibrils
(Fig. 2C) without forming specialized contacts, such as with hemidesmosomes. Those CGCs arranged in doublets or clusters of even more cells were partly apposed to each other without being completely separated by glial cell processes
(Fig. 2D) . Neither single cells nor groups of CGCs were supplied by or adjacent to capillaries. All CGCs were embedded in a nerve fiber plexus.
A dense perivascular network of nerve fibers was seen around the arteries, arterioles, and veins when using antibodies against VAChT, nNOS, or VIP as the marker for parasympathetic, and TH, DBH, VMAT-2, or NPY as the marker for sympathetic nerve fibers. The larger arteries were also accompanied by CGRP-positive nerve fibers. SP-positive nerve fibers formed a fine network in the outer choroid without showing close affiliation to the vessels but rather contacting cells in the choroidal stroma. In the inner choroid, only single TH/VMAT-2/NPY-positive nerve fibers reached the choriocapillary layer, forming a fine network at the level of the precapillary arterioles.