Immunohistochemical study of rabbit choroidal innervation

Vision Research 39 (1999) 1249–1262

Immunohistochemical study of rabbit choroidal innervation

Jose M. Ramırez *, Alberto Trivino, Rosa De Hoz, Ana I. Ramırez, Juan J. Salazar,Julıan Garcıa-Sanchez

Instituto de In6estigaciones Oftalmologicas Ramon Castro6iejo, Facultad de Medicina, Pab VI, 4a Planta, Uni6ersidad Complutense,28040 Madrid, Spain

Received 8 August 1996; received in revised form 24 July 1998

Abstract

Immunocytochemical methods with antibodies to the light (68 kDa), medium (160 kDa), and heavy (200 kDa) chain subunitsof the neurofilament triplet have been used to visualize neuronal structures in rabbit choroids. Choroidal nerve fibers were presentin the suprachoroid and vascular laminae and absent in the choriocapillary layer. These fibers may be classified as perivascularand intervascular. Perivascular fibers surround all arterial and venous blood vessels and form a network; these fibers were labeledwith the three NF antibodies, although they were more easily visualized with anti NF-160 and anti NF-200 than anti NF-68.Intervascular fibers formed two groups. The first group consisted of fibers situated between the blood vessels and parallel to theblood vessel wall surface (paravascular fibers); these fibers were better observed using anti NF-160 and NF-200 than anti NF-68.The second group consisted of fibers which travel the entire length of the choroid until they reach the nerve plexus of the ciliarybody (long tract fibers). The plexus was observed with anti NF-68, anti NF-160 and anti NF-200; however, the long tract fiberswere more clearly visualized with anti NF-160 and anti NF-200 than with anti NF-68. Two types of choroidal cell were alsolabeled: ganglion cells and melanocytes. Ganglion cells are small, scarce neurons situated in the peripheral choroid; they werelabeled with anti NF-160 and anti NF-200. The melanocytes were only labeled with anti NF-200 and they were the only nonneuronal structure visualized using antibodies against neurofilaments. © 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Neurofilament; Choroid; Ciliary body; Innervation; Ganglion cells; Melanocytes

1. Introduction

The choroid is the vascular tissue of the eye and hasmultiple functions. The most important of these arenourishment of the external retinal tissue and regula-tion of ocular temperature; both are controlled by thechoroidal innervation, which depends on the sympa-thetic system (from the superior cervical ganglion),parasympathetic system (from the ciliary ganglion) andsensory system (from the trigeminal ganglion).

The morphology and distribution of the choroidalnerve fibers were originally studied using silver tech-niques (Wolter, 1960; Castro-Correia, 1967), whichwere not very selective because of the affinity of silverfor other structures such as collagen and elastic fibers.Subsequent studies sought greater specificity by mark-ing the choroidal nerve fibers. Initially, histochemical

techniques were used (Ehinger, 1966a,b,c,d), based onthe fluorescence emitted by the different neurotransmit-ters (noradrenaline and acetyl cholin) after chemicaltreatment. The later development of immunohistochem-ical techniques and the use of antibodies against vari-ous neuropeptides (Terenghi, Polak, Probert,McGregor, Blank, Butler, Unger, Zhang, Cole &Bloom, 1982; Miller, Coster, Costa & Furness, 1983;Stone, 1986a,b; Stone, Tervo, Tervo & Tarkhanen,1986) has made it possible to examine the choroidalnerve fibers more selectively from a physiological orfunctional point of view and to determine whether thefibers are sympathetic, parasympathetic or sensory.There are however few references to morphologicalstudies using immunohistochemical techniques.

The recent application of antibodies against neurofi-laments has now made it possible to selectively label thenerve fibers in the central and peripheral nervous sys-tems (Liem, Keith Leterrier, Trenchner & Shelanski,1981; Yen & Fields, 1981). Neurofilament (NF) protein

* Corresponding author. Tel.: +34-91-3941-363; fax: +34-91-3941-359; e-mail: [email protected].

0042-6989/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.

PII: S 0 0 4 2 -6989 (98 )00255 -7

J.M. Ramırez et al. / Vision Research 39 (1999) 1249–12621250

Table 1Comparison chart showing the intensity of immunoreactivity of the different antibodies in the various choroid zones

NF-68 NF-160Intensity of immunoreactivity NF-200

++ ++++Central choroid +++Perivascular fibers++++ ++++++–Intervascular fibers

+ ++++Peripheral choroid +++Perivascular fibers++++++++++–Intervascular fibers

– –Transition zone between choroid and cilliary body Perivascular fibers –+++ ++++Intervascular fibers ++++

is an important constituent of axons and is one of anumber of major classes of 10 nm proteins in cells(Wuerker & Kirkpattrick, 1972). It is built up of atriplet of polypepetides of approximately 68, 160 and200 kDa (Liem, Yen, Solomon & Shelanski, 1978).The utilization of these antibodies in whole-mounts ofchoroids can be useful to show their morphologicalcharacteristics.

The aim of the present study was to apply specificNF antibodies and PAP techniques to rabbit choroidwhole-mounts and analyze the morphology and distri-bution of the different nerve fibers.

2. Materials and methods

In total 24 eyes from 12 adult New Zealand albinorabbits were used. The animals were killed with anoverdose of sodium pentobarbital and perfused peri-mortem with 4% paraformaldehyde (Sigma, USA) in0.1 M phosphate buffered saline (PBS) at pH 7.4.Eyes were enucleated. After removing the cornea, lens,vitreous humor and retina the resulting eyecup wasimmersion-fixed for 4 h at 4°C in 4% paraformalde-hyde in 0.1 M PBS, pH 7.4. Afterwards the choroidwas removed from the eyecup and washed in PBS forseveral hours at 4°C.

2.1. Immunocytochemical procedure

The immunocytochemical study was performed us-ing the peroxidase antiperoxidase (PAP) method withthe following procedure. The choroids were pretreatedwith 0.3% hydrogen peroxide in PBS for 30 min atroom temperature before washing in PBS (3×10min). They were then incubated for 24 h at 4°C in10% normal goat serum (NGS, Sigma, USA) and0.2% Triton X-100 (MERCK, Germany) diluted inPBS. Whole-mount choroids were incubated withmouse monoclonal antisera against the following anti-gens: neurofilament 68 kDa (NF-68, clone N124,Biomakor, Israel), neurofilament 160 kDa (NF-160,clone NN18, Biomakor), and neurofilament 200 kDa

(NF-200, clone NE14, Biomakor) in a 1/100 dilutionat 4°C for 3–4 days and then washed in PBS (3×3h). The choroids were incubated at 4°C for 2 dayswith the immunoglobulin fraction from goat anti-mouse serum (Sigma) diluted 1/100, and then washedin PBS (3×3 h). After this, they were incubated at4°C for 24 h with mouse peroxidase antiperoxidasecomplex (PAP, Sigma) diluted 1/700, and rinsed inPBS (3×3 h). The whole choroids were treated with0.03% DAB (diaminobenzidine tetrahydrochloride,SIGMA) in PBS for 5 min, followed by 0.03% DAB/0.01% hydrogen peroxide in PBS for an additional 7min. After final rinsing in PBS (3×10 min), thechoroids were flat-mounted and fixed onto gelatinchrome-alum subbed slides, dehydrated in ascendingalcohols, cleared in xylene and mounted in Canadabalsam (MERCK).

All antibody dilutions were carried out in 0.2% Tri-ton X-100, 0.1% NGS in 0.1 M PBS at pH 7.4. Twotypes of controls were included: firstly, incubation inthe primary antibody solution was eliminated todemonstrate that the secondary antibodies reactedonly with their respective primary antibody. Secondly,incubation in the primary antibody solution was sub-stituted by normal mouse serum to check for primaryantiserum nonspecificity.

3. Results

3.1. Anti-NF 68

All the choroidal regions (nasal, temporal, upperand lower) exhibited very low NF-68 immunoreactiv-ity; this was more intense at the central choroid thanin the periphery. However the most intense NF-68immunoreactivity was located at the ciliary body andin its transition zone with the choroid (Table 1). Im-munoreactivity was absent in the choriocapillary layer(Fig. 1A).

NF 68(+ ) fibers can be classified as perivascularand intervascular (Fig. 1A).

J.M. Ramırez et al. / Vision Research 39 (1999) 1249–1262 1251

Fig. 1. Anti NF-68. (A) Perivascular nerve fibers (arrowhead) surround the arterial and venous blood vessels (*). Intervascular fiber (arrow).(bar=50 mm). The intervascular fiber is shown at high power in (B) (arrow) (bar=25 mm). (C) Intervascular fibers. These are either single(arrowhead) or grouped in pairs (arrow). (bar=25 mm). (D) Ciliary body. A long ciliary nerve is observed circumferentially (arrow). (CP: ciliaryprocesses; bar=250 mm). (E) Ciliary body. Individual smooth fibers lie between the ciliary processes. Some fibers go towards the ciliary body (�)and other fibers go towards the choroid (J). (bar=100 mm). (F) Control: no incubation in anti-NF-68. (bar=50 mm). (G) Control: incubationin normal mouse serum. (bar=100 mm) (A, B, C, D, E, F, G): Whole mount preparation, anti NF-68, PAP; (C, E): Nomarski optics.

3.2. Peri6ascular ner6e fibers

Very low NF-68 immunoreactivity was observedaround all arterial and venous blood vessels; butnerve fibers could not be observed individually. (Fig.1A).

3.3. Inter6ascular ner6e fibers

Course through the choroidal stroma between theblood vessels (Fig. 1A, B). These fibers are scarce andshow higher NF-68 immunoreactivity than perivascu-lar fibers. Their main characteristics are their discon-tinuous appearance and thin caliber; many are singleor grouped in pairs (Fig. 1C). These fibers are foundthroughout the choroid and it is impossible to ob-serve the complete course of the fibers since they dis-appear on many occasions. The intervascular fiberscan make contact with the vascular walls.

At the ciliary body, a thick, highly NF-68 (+ )bundle containing approximately 50–70 fibers was

observed circumferentially (Fig. 1D). The situation ofthis bundle suggested that it could be a long ciliarynerve as it enters the anterior uvea, branching outinto numerous smooth individual fibers to form afine-meshed plexus between the ciliary processes (Fig.1E).

4. Anti NF-160

All the choroidal regions (nasal, temporal, upperand lower) showed high immunoreactivity to NF-160(Fig. 2A). The labeled tissue was limited to the supra-choroidal and vascular layers (large and medium sizevessels), and was absent from the choriocapillarylayer. The NF-160 immunoreactivity was more in-tense at the central choroid and peripheral choroidthan in the zone between choroid and ciliary body(Table 1).

Depending on their location, nerve fibers can beclassified as perivascular or intervascular (Fig. 2A).


Fig. 2. AntiNF-160. (A) Perivascular nerve fibers (arrows), and intervascular nerve fibers (arrowhead). (bar=100 mm). (B) Perivascular fibers.Fiber bundles with a band morphology surround the vascular wall. Nerve fiber with a tuning fork morphology (arrow) (bar=50 mm). Detail ofthis fiber in the upper right corner (bar=10 mm). (C) Intervascular fibers. Long ciliary nerve (arrowhead). (bar=50 mm). (D) Fiber bundlecontaining 10–20 fibers (arrowhead). This bundle branches out into medium size bundles made up 3 or 4 fibers (arrow). (bar=50 mm). (E)Occasional widening of the intervascular fibers (arrowhead). (bar=25 mm). (F) Paravascular fibers (arrows) (blood vessels (*)). (bar=50 mm). ((A,B, C, D, E, F): Whole mount preparation, antiNF-160, PAP. (B, C, D, E): Nomarski optics).

4.1. Peri6ascular fibers

Many thin-caliber fibers surround the vascular walls(arteries and veins) along their pathway (Fig. 2A, B). Inthe large vessels the nerve fibers exhibit a circumferen-tial pattern, but when the vessels branch out, the fiberssurrounding the vessels wall form a mesh. This meshconsists of lengthwise and crosswise fibers covering theentire surface of the vessel. In these thinner vessels thecircumferential pattern is less apparent. These fibersshow high immunoreactivity to NF-160 and thus allowvisualization of all the choroidal blood vessels. Somevery thin fibers with tuning fork morphology wereobserved over the large blood vessel walls (arteries orveins), but they were not found along the course of theblood vessel (Fig. 2B).

4.2. Inter6ascular fibers

At the suprachoroidal lamina, a thick fiber bundlemade up of 50–70 fibers with a band morphology wasobserved. This bundle entered the suprachoroid in theequatorial zone of the nasal and temporal regions,

transiting through the suprachoroid lamina withoutforming a real plexus. The situation of this bundlesuggests that it could be a long ciliary nerve (Fig. 2C).

In the underlying vascular lamina some strongly NF-160(+ ) fiber bundles were visible containing approxi-mately 10–20 fibers (Fig. 2D). These bundles werelocated in the vicinity of the long ciliary nerves, but wewere unable to pinpoint their exact origin. The bundlesbranched into medium-sized bundles made up of 3 or 4fibers (Fig. 2D) which branched again into very thinfibers which made contact with the blood vessel walls.The fibers presented occasional widening (Fig. 2E) andat some points were located parallel to the blood vesselwall surface (paravascular fibers) (Fig. 2F).

In the choroidal stroma we observed NF-160(+ )fibers which ran through the choroid without disap-pearing. We have called these ‘long tract fibers’. Longtract fibers can gather together with neighboring fibersto form bundles made up of two, three or more fibers.There is normally a thick fiber next to a thin fiber (Fig.3A). Occasionally, the fiber caliber is large. All longtract fibers make contact with the blood vessel atnumerous points (Fig. 3B). The fiber bundles ran as far


Fig. 3. Anti-NF 160. (A) Long tract fibers. Next to a thin caliber fiber (arrow) there is a thick caliber fiber (arrowhead). (bar=250 mm). (B) Longtract fiber making contact (arrowhead) with a blood vessel (arrows). (bar=100 mm). (C): Ciliary body. Fiber network (arrow) between the ciliaryprocesses. (bar=50 mm). (D) Control: no incubation in anti NF-160. (bar=100 mm). (E) Control; incubation in normal mouse serum. (bar=100mm) (A, B, C, D, E): Whole mount preparation; antiNF-160, PAP.

as the peripheral choroid, the number of fibers increas-ing as they progressed. Close to the ciliary body thebundles were made up of 15–20 fibers. At the ciliarybody there was a fiber network with intense NF 160immunoreactivity (Fig. 3C).

With anti NF-160 we can also label the choroidalganglion cells (Fig. 4A, B, C). These cells are veryscarce and are among the constituents of the intervas-cular fiber network located in the peripheral choroid.Most of the neurons have bipolar phenotype, but thereare multipolar neurons. These cells are gathered inclusters of two to three (Fig. 4A, B) but on occasionsthey can be single (Fig. 4C).

5. Anti NF-200

NF-200(+ ) immunoreactivity was very high in allchoroidal regions (nasal, temporal, upper and lower) atthe level of the suprachoroidal and vascular lamina butabsent in the choriocapillaries (Fig. 5A). The NF-200label was principally related to the blood vessel walls(arteries and veins) and was more intense at the centralchoroid than at the periphery (Table 1). Immunoreac-tivity was due to the presence of many NF-200(+ )nerve fibers and star-shaped cells identified as choroidalmelanocytes (Fig. 5B), which were absent from therespective controls (Fig. 5C, D).

5.1. Peri6ascular fibers

All choroidal blood vessels were accompanied by thinperivascular lengthwise and crosswise fibers, forming ameshwork on their wall surface (Fig. 5E). A fibernetwork was also observed at the branching point of ablood vessel.

Fig. 4. AntiNF-160. (A, B, C): Choroidal ganglion cells. (A, B)Ganglion cells (arrowhead) gathered in clusters of two or more cells.(bar=50 mm). (C): Single ganglion cell with a bipolar phenotype(arrowhead). (bar=50 mm). ((A, B, C): Whole mount preparation;antiNF-160; PAP; Nomarski optics).


Fig. 5. Anti NF-200. (A) Perivascular nerve fibers (arrowhead), intervascular nerve fiber (arrow). Blood vessel (*). (bar=250 mm). (B) Choroidalmelanocytes. Intervascular fiber (arrowhead) positioned close to the melanocytes (arrow). (Blood vessel (*)). (bar=25 mm). (C) Control: noincubation in anti-NF 200 (bar=100 mm). (D) Control: incubation in normal mouse serum (bar=100 mm). (E) Perivascular fibers (arrowhead),paravascular fibers (arrow) with enlargement of this caliber. Blood vessel (*). (bar=50 mm). (F) Long ciliary nerve (arrow). Network of the ciliarynerve at the ciliary body (arrowhead). (Ciliary processes (CP)). (bar=250 mm) (G) Long ciliary nerve (arrow). This nerve branches out intomedium size bundle (arrowhead) (Ciliary processes (CP)). (bar=250 mm). ((A, B, C, D, E, F, G): Whole mount preparation; anti NF-200; PAP.(B, E, G): Nomarski optics).

5.2. Inter6ascular fibers

At the suprachoroidal lamina two thick band-shapedfiber bundles with strong NF-200 immunoreactivitywere observed. The bundles, made up of 50–70 fiberswere situated in the equatorial zone of the nasal andtemporal regions of the choroid. These bundles couldbe long ciliary nerves (Fig. 5F, G). In the vascular layerthere were bundles with 5–10 fibers and high im-munoreactivity. In some cases these were located paral-lel to the vascular wall and were denominatedparavascular fibers (Fig. 5E). Thin fibers emerged fromthe bundles and made vascular contact (Fig. 5E). Atsome points there was enlargement of the caliber of theintervascular fibers (Fig. 5E).

Besides the nerve branches issuing from the longciliary nerves (Fig. 6A), we observed bundles of thinnerfibers. These bundles issued from the posterior pole andbranched out progressively in their journey to the cil-iary body, producing both paravascular branches andvessel terminals. These nerve fibers could be thebranches of the short ciliary nerves (Fig. 6B, C).

At the ciliary body the fiber bundles formed anextensive network (Fig. 6D, E). The bundles, highlyNF-200(+ ), were made up of approximately ten fibers(Fig. 6E); these emerged from bundles which ran theentire length of the choroid, up to the ciliary body.These last bundles could be terminal branches of thelong and short ciliary nerves.

With anti NF-200 we can label two types of cell:Ganglion cells and melanocytes.

5.3. Ganglion cells

These cells were very scarce and were located prefer-entially at the ciliary body. They had a multipolar orbipolar phenotype and were gathered in clusters of twoor three neurons (Fig. 6F).

5.4. Melanocytes

The melanocytes were distributed evenly within thedifferent choroidal quadrants. These cells were located


Fig. 6. Anti NF-200. (A) Branches of the long ciliary nerves (arrowhead) which split into smaller-caliber branches, some of them situated parallelto the vessel wall (paravascular fibers) (arrow) (blood vessels (*)). (bar=250 mm). (B) Short ciliary nerves (arrow) (blood vessel (*). (bar=250mm). (C) Short ciliary nerves. Some fibers follow a course parallel to the course of the vessels (arrow) and others make vascular contact(arrowhead). (bar=50 mm). (D) Fiber bundles in the ciliary body (Ciliary processes (CP)). (bar=100 mm). (E) Fiber network between the ciliaryprocesses. (F) Choroidal ganglion cell in the ciliary body with multipolar phenotype (arrow) (Ciliary processes (CP)). (bar=50 mm). ((A, B, C,D, E, F): Whole mount preparation; anti NF-200; PAP. (C, E, F): Nomarski optics).

near the blood vessel walls or in the choroidal stroma.They were star-shaped and were interconnected by theirprocesses, forming a three-dimensional net. Perivascu-lar and intervascular fibers were positioned close to themelanocytes (Fig. 5B).

6. Discussion

In this study the neurofilament protein antibodies(68, 160, 200 kDa) were used to analyze the morphol-ogy and distribution of the choroidal nerve fibers. Theneurofilaments are purely structural, serving primarilyto consolidate the axon and to enhance axonal caliber(Zhao & Szaro, 1995). These antibodies have been usedto describe the morphology of neuronal structures inthe central and peripheral nervous system (Liem et al.,1978, 1981; Yen & Fields, 1981). As far as we know,this is the first study done on the choroid using anti-bodies against the three NF (68, 160, 200 kDa)proteins.

In choroidal whole-mounts, we observed that thechoroidal innervation was composed of nerve fibersgathered into groups of different origins. The firstgroup of fibers reached the choroid as perivascularfibers that surrounded the short posterior ciliary arter-ies. These nerve fibers stem from the perivascularplexuses in the carotid system, which runs to the oph-thalmic artery. Thereafter, the fibers spread out tofollow the short posterior ciliary arteries in thechoroidal vascular layer (Guglielmone & Cantino,1982; Nuzzi, Gugglielmone & Grignolo, 1992).

The perivascular plexus in the short posterior ciliaryarteries and their main branches formed a meshwork inwhich fibers in circumferential pattern were very appar-ent when observed with the antibodies to neurofila-ments. This is similar to the pattern described in thesearteries with NADPH-diaphorase (Flugel-Koch,Tamm, Mayer, Lutjen-Drecoll, 1994) and is consistentwith the pattern found in the inner carotid artery usingneuropeptides (VIP and NPY) (Tsai, Tew & Shipley,1992).


Scheme 1. The schematic shows a topographic map of the distribution of the nerve fibers in the choroidal territory. This is a general view of therabbit choroid. The optic nerve (ON) is in the central area; the ciliary body (CB) and their processes (CP) are located in the peripheral zone.Choroid vasularization is shown in light grey, while the nerve fibers are shown in black. As the schematic shows the long ciliary nerves (1)penetrate the nasal (N) and temporal (T) zones of the equatorial choroid. These nerves occupy a triangular area as they travel towards the iris,issuing vascular branches on the way. Also shown are the entries of the short ciliary nerves (2) in the vicinity of the ON and their main branchesto the vessels. The fiber meshwork (3) at the CB and the branches (4) toward the CP are also shown. V: vorticous veins, S: Upper, I: lower.

The second group of fibers consists of nerve fiberbundles which travel along the course of the ciliaryarteries. These fibers stem from the short posteriorciliary nerves from the ciliary ganglion which penetratethe eyeball close to the short posterior ciliary arteries.These nerves separate into bundles running parallel oroblique to the ciliary arteries (Scheme 1) (Guglielmone& Cantino, 1982; Nuzzi et al., 1992).

Finally, the third group of fibers are composed ofbranches of the long ciliary nerves that divide inside thevascular layer. These fibers penetrate the eyeball in theequatorial zone of the nasal and temporal regions andtransit through the suprachoroid lamina without form-ing real plexuses, to form an extensive network in theunderlying vascular lamina (Scheme 1) (Miller et al.,

1983; Ramırez, Trivino, Ramırez & Garcıa-Sanchez,1989). This general distribution of nerve fibers is inagreement with observations made using silver tech-niques (Wolter, 1960; Castro-Correia, 1967) and im-munohistochemical techniques with neuropeptides(Terenghi et al., 1982; Miller et al., 1983; Terenghi,Polak, Allen, Zhang, Unger & Bloom, 1983; Uusitalo,Lehtosalo, Palkamo & Toivanen, 1984; Terenghi, Po-lak, Ghatel, Mulderry, Butler, Unger & Bloom, 1985;Terenghi, Zhang, Unger & Polak, 1986; Stone, 1986a,b;Stone et al., 1986; Stone & McGlinn, 1988; Stone,McGlinn & Kuwayama, 1988; Ramırez, de Hoz,Salazar, Ramırez, Trivino & Garcıa-Sanchez, 1990;Grimes, McGlinn, Koeberlein & Stone, 1994; Klooster,Beckers, Ten Tusscher, Vrensen, van der Want &Lamers, 1996).


The choroidal innervation is principally related to thelarge and medium sized vessels. Although previousauthors have found nerve fibers in the choriocapillarylayer (Castro-Correia, 1967; Ehinger, 1966a,b,c,d; Mat-susaka, 1981), like Ruskell (1971), Guglielmone andCantino (1982), Flugel-Koch, Kaufman and Lutjen-Drecoll (1994) and Flugel-Koch et al. (1994), we havenot been able to demonstrate the choriocapillaryinnervation.

The ciliary arteries which subsequently enter thechoroid have a dense perivascular supply of nerve fibersand as they branch, their perivascular nerve supply isrestricted to medium sized blood vessels (Miller et al.,1983). In our study, NF immunoreactive fibers werefound on the walls of all large sized vessels with ameshwork pattern in which the fibers were arrangedaround the vessel both circumferentially and lengthwise(Scheme 2). The nerve fiber distribution is reminiscentof the layout of the connective tissue in the blood-vesselwalls (Sverdlick, 1945; Wolter, 1960), and in fact insome studies using silver techniques, some of themarked fibers could have been collagen (Sverdlick,1945; Mawas, 1952; Wolter, 1960; Castro-Correia,1961, 1967). However, with immunohistochemical tech-niques, and more specifically the use of antibodies tothe specific neurofilaments of the nerve fiber cytoskele-ton, we can mark the neurons more selectively. More-over, perivascular nerve fibers were not observed in anyof the two controls performed for the different neurofi-laments, where the choroid remained quite unmarked(Fig. 1F–G, Fig. 3D–E, Fig. 5C–D). In the firstcontrol, the primary antibody (anti-NF) was omitted todemonstrate that the secondary antibodies only react

with their respective primary antibodies and thus elimi-nate the possibility of a non-specific background(Priestley, 1987). And in the second control incubationin the primary antibody solution was substituted byincubation in normal mouse serum to control for pri-mary antiserum nonspecificity. In addition, the samenerve fiber pattern has been observed using neuropep-tides in all the vessels stemming from the carotid system(Tsai et al., 1992) and also in the short posterior ciliaryarteries using NDAPH-diaphorase (Flugel-Koch et al.,1994). According to Tsai et al. (1992), the perivascularnerve fibers exhibit this meshwork pattern when theyattain maturity and may be either sympathetic orparasympathetic. So similar a distribution of sympa-thetic and parasympathetic fibers in the perivascularplexuses indicates that there is an important balancebetween vasoconstriction and vasodilation in all vesselsstemming from the carotid system (Tsai et al., 1992).However, in the present work more perivascular fiberswere found than when antibodies to various neuropep-tides were used (Terenghi et al., 1982, 1983; Miller etal., 1983; Stone, 1986a,b). This could be because theselast authors used specific antibodies to sympathetic orparasympathetic or sensory fibers, whereas the antibod-ies to neurofilaments can mark all three fiber types atonce (Vitadello, Triban, Fabris, Dona, Gorio & Schi-afino, 1987; Vega, Vazquez, Naves, Del Valle, Calzada& Represa, 1994; Bleys, Cowen, Groen, Hillen &Ibrahim, 1996), so that the fibers in the overall patterncontains are more numerous.

In our study the perivascular network was morereadily visualized using anti NF-160 and NF-200 thananti NF-68 kDa; this can be explained by the structuralarrangement of the NF triplet. The three neurofilamentproteins NF-68, NF-160 and NF-200 contain a con-served a helical rod domain and their respective molec-ular masses reflect differences in the amino and carboxyterminal amino acid sequences that flank the rod do-main. The NF-160 and NF-200 proteins are anchoredto a core of NF-68 via their central rod domain and arethe protein determinants of the axonal diameter (Eyer& Peterson, 1994). This special role of NF-160 andNF-200 is probably mediated through their carboxylterminal domain, which forms extensive cross-bridgesbetween adjacent neurofilaments, thus creating a largenumber of extensively cross-linked neurofilaments andhence an increase in axonal caliber (Zhao & Szaro,1995). These macromolecular associations formed bymedium and high molecular weight neurofilaments thataugment axonal caliber, make it possible to observe thenerve fibers individually using anti NF-160 and NF-200, while with anti NF-68 only, immunoreactivity (+ )around blood vessel walls can be observed. Moreover,all these factors could account for the differences be-tween the pattern of perivascular marking in the vari-ous anti-neurofilaments. However, in some of our

Scheme 2. Three-dimensional schematic of one large choroidal bloodvessel showing perivascular fibers (1) and some very thin fibers withtuning fork morphology (2). Intervascular fibers are also shownlocated parallel to the blood vessel wall (paravascular fibers) (3),sending out a thin branch that makes vascular contact (4). Finally, afiber network can be observed at the branching point of the bloodvessel (5).


Scheme 3. Three-dimensional schematic of choroid showing vascular layers (large (1) and medium (2) size vessels), choriocapillaries (3), Bruch’smembrane (4) and pigmentary epithelium (5). The long ciliary nerve (6) penetrates the suprachoroidal lamina sending branches (7) between thelarge and medium size blood vessels. These branches are intervascular fibers.

whole-mount choroidal preparations, the perivascularfibers were more easily studied with anti NF-160 thananti NF-200. This was determined by the choroidalmelanocytes which were labeled with anti NF-200.These cells are situated near the vascular walls and theirimmunoreactivity could mask the perivascular fibers.

In addition, intervascular fibers were found in allchoroidal territory except in the choriocapillary tissue.Two groups of fibers were observed: the first group wasformed by branches of the long (Scheme 3) and shortciliary nerves, which were situated between the bloodvessels and parallel to the blood vessel wall surface(paravascular fibers) (Schemes 2 and 4). The secondgroup was formed by fibers, and also branches of thelong and short ciliary nerves, which travel along thechoroid up to the ciliary body plexus (long tract fibers)(Scheme 4). Antibodies against the three anti NFproteins showed positive staining in two groups ofintervascular fibers. Nerve fiber morphology in bothgroups were better observed using anti NF-160 andNF-200 than anti NF-68. Like Seiger, Dahl, Ayer-Lelievre and Bjoklund (1984), we found (using antiNF-68) that the intervascular fibers had a clearly dis-continuous appearance; however, when these were la-beled with NF-160 and NF-200, all the courses of thefiber could be followed. As suggested previously forperivascular fibers, the macromolecular associations be-tween NF triplets established by NF-160 and NF-200could allow continuous observation of nerve fibers,while NF-68, which makes up the filament backbone(Lee & Cleveland, 1994), could be more difficult to

visualize if the fibers are labeled continuously.When anti NF-160 and anti NF-200 were used, we

observed occasional enlargement of the axonal caliber,which we think could be an irregular accumulation ofneurofilaments along the axons. Nixon, Paskevich, Si-hag and Thayer (1994) observed that the transport ofsome highly phosphorylated isoforms of NF-160 andNF-200 in the ganglion cell axons slows sharply orstops for considerable periods of time. This process oflocal accumulation of neurofilaments would enable neu-rons to expand axon caliber with little or no increase inneurofilament gene expression.

The long tract fibers were readily observed with antiNF-160 and NF-200. As axons mature, the levels ofthese neurofilament proteins increase and the highestmolecular weight neurofilament protein NF-200emerges. During axonal maturation, both NF-160 andNF-200 are progressively phosphorylated and this pro-cess is accompanied by deceleration in slow axonaltransport and an increase in axonal caliber (Zhao &Szaro, 1995). This could explain why the long tractfibers, which may be formed by more mature axons, arereadily observed with anti NF160 and NF-200.

Like Flugel-Koch, Kaufman and Lutjen-Drecoll(1994) and Flugel-Koch et al. (1994), we found gan-glion cells in rabbit choroid (Scheme 5). These authors,using NADPH-diaphorase, observed approximately 20cells per choroid. However, in our choroid whole-mounts we only found a few small-sized cells labeledwith anti NF-160 and NF-200. This paucity of cellscould be due to the neurofilament expression in the


Scheme 4. This schematic demonstrates the distribution pattern of the long tract intervascular fiber (1). This fiber issues branches consisting oftwo fibers (one thick (2) and one thin (3)); later they send out very thin branches that make vascular contact (4).

neuronal soma. In the peripheral sensory and auto-nomic ganglia, the largest neurons express neurofila-ments and the smallest neurons express peripherin, withroughly medium-sized neurons expressing both (Xu,Dong & Cleveland, 1994). The choroidal ganglion cellsare small neurons and could express peripherin likeautonomic ganglia not labeled with anti NF-triplet.Thus, the few neurons labeled with anti NF-160 andNF-200 could be medium-sized neurons. These neuronsshould be labeled with three neurofilament antibodies,but we were unable to observe these cells with antiNF-68. Liem et al. (1978), Sternberger and Sternberger(1983) and Trojanowski, Obrocka and Lee (1983) failedto detect NF-68 in the cell body of Purkinje cells. In alater study Trojanowski, Obrocka and Virginia (1985)explained that this can be ascribed to several factors,such as the choice of fixative or dilution conditions.

Finally, we labeled choroidal melanocytes with antiNF-200 in some of our mounts (Scheme 6). This nonneuronal cell expresses the 200 kDa neurofilamentprotein as does retinal pigment epithelium. This ispossible since these cells can express multiple neuron-associated markers (b111, 200 neurofilament and neuronspecific enolase) (Vinores, Derevjanik, Mahlow, Hack-ett, Haller, de Juan, Frankfurter & Campochiaro,1995). In addition, NF-200 is among the proteins thatappear in the early period of neurite elongation and isstimulated by a-melanotropin, which is a peptide with awell-documented effect on melanocytes (Vadoud-Seyedi, Lespargnard, Renard, Deraemaecker &Ghanem, 1994).

From the studies of Matsusaka (1982) it is knownthat melanocytes may play a role in choroidal circula-tion, the contractile nature of these cells providingadded structural support for the delicate capillaries.The cytoplasm of choroidal melanocytes containsfilaments that closely resemble actin filaments, suggest-ing the ability to generate contractile forces. This con-

traction-generating ability may allow melanocytes toregulate choroidal vessel caliber through neural inner-vation (Gartner & Fischer, 1989). Labeling with antiNF-200 is useful in studying the relationship betweenthe nerve fibers and melanocytes and has not beenreported previously.

7. Functional considerations

In this study using antibodies against the differentneurofilaments, we observed two large groups of fibers:fine-caliber perivascular fibers surrounding the vessels,and intervascular fibers. The caliber of the latter dif-fered according to location.

At the ciliary body we found a plexus made up ofvery fine fibers similar in caliber to the perivascularfibers. In the rest of the choroid the intervascular fibers,which are not very numerous, stem from nerves with aband morphology located at the suprachoroid, whichdivide into branches of different caliber in the choroidalterritory.

Seiger et al. (1984), using antibodies to neurofila-ments, observed that at the iris there were plexuses ofvery fine nerve fibers situated in the area of the sphinc-ter and the dilator. They also found other thicker fibersof varying caliber distributed along the iris. These lastfibers were reminiscent of the pattern of sensory fibersdemonstrated with silver techniques (Linder, 1978).Upon performing denervation, Seiger et al. (1984)found that the thicker fibers, which varied in caliber,disappeared upon lesion of the trigeminal ganglion,while the plexuses of very fine fibers in the sphincterand the dilator were unaffected.

At the choroid we observed two plexuses of very finefibers similar to the ones found at the iris sphincter anddilator (Seiger et al., 1984). These fibers are locatedsurrounding the blood vessels, and at the ciliary body.


Scheme 5. Three-dimensional schematic of choroidal ganglion cells. These cells participate in the intervascular fiber network. They are gatheredin clusters of two to three cells. Their phenotype can be bipolar or multipolar.

According to Seiger et al. (1984), the very fine fibers areof sympathetic and parasympathetic origin; this is inagreement with the observations of Tsai et al. (1992),who found a plexus of very fine fibers in a meshworkpattern around the inner carotid artery, stemming fromboth the sympathetic and the parasympathetic systems.We therefore believe that the perivascular plexus of finefibers in the choroid vessels could be sympathetic andparasympathetic fibers which are necessary to maintainthe balance between vasoconstriction and vasodilationthat regulates the choroidal blood flow (Bill, 1962;Nilsson et al., 1985; Klooster et al., 1996). At the ciliarybody we also observed a similar plexus of very finefibers, which could likewise be of sympathetic andparasympathetic origin and play a role in the au-tonomous innervation of the ciliary body (Ehinger,1966a).

In this study we also observed intervascular fibers ofvarying calibers, distributed throughout the length ofthe choroid and not very numerous. The thickest ofthese fibers resemble the fiber bundles observed bySeiger et al. (1984) at the iris with anti-NF, whichdisappeared following denervation of the trigeminalganglion. We therefore believe that these thicker fiberscould be sensory fibers, while the rest of the intervascu-lar fibers are sympathetic and parasympathetic fibersobserved on their course to the ciliary body.

Acknowledgements

We thank Alistair L. Ross for his linguistic assis-tance. This work was supported by Ministerio de Edu-cacion y Ciencia (DGICYT) grant PB92-0185.

References

Bill, A. (1962). Autonomic nervous control of uveal blood flow. ActaPhysiologica Scandina6ica, 56, 70–81.

Bleys, R. L., Cowen, T., Groen, G. J., Hillen, B., & Ibrahim, N. B.(1996). Perivascular nerves of the human basal cerebral arteries: I.topographical distribution. Journal of Cerebral Blood Flow andMetabolism, 16, 1034–1047.

Castro-Correia, J. (1961). Inervacao coroideia anales del instituto.Barraguer, 213, 487–518.

Castro-Correia, J. (1967). Studies on the innervation of the uvealtract. Ophthalmologica, 154, 497–520.

Ehinger, B. (1966). Connections between adrenergic nerves and othertissue components in the eye. Acta Physiologica Scandina6ica, 67,57–64.

Ehinger, B. (1966). Distribution of adrenergic nerves in the eye andsome related structures in the cat. Acta Physiologica Scandina6ica,66, 123–128.

Ehinger, B. (1966). Ocular and orbital vegetative nerves. Acta Physio-logica Scandina6ica (Suppl. 268), 67, 1–35.

Ehinger, B. (1966). Localization of ocular adrenergic nerves andbarrier mechanisms in some mammals. Acta Ophthalmologica, 44,814–822.

Eyer, J., & Peterson, A. (1994). Neurofilament-deficient axons andperikaryal aggregates in viable transgenic mice expressing a neu-rofilament-b-galactosidase fusion protein. Neuron, 12, 389–405.

Scheme 6. Three-dimensional schematic of choroid showing vascularlayers (large (1) and medium (2) size vessels) and choriocapillaries (3).At the suprachoroidal laminae intervascular nerve fibers with periodicwidening (4) can be observed. The widening extends thin fibers (5)towards the vessel walls. In this schematic the distribution pattern ofthe choroidal melanocytes is also shown (6).


Flugel-Koch, C., Kaufman, P., & Lutjen-Drecoll, E. (1994). Associa-tion of a choroidal ganglion cell plexus with the fovea centralis.In6estigati6e Ophthalmology and Visual Science, 35, 4268–4272.

Flugel-Koch, C., Tamm, E. T., Mayer, B., & Lutjen-Drecoll, E.(1994). Species differences in choroidal vasodilative innervation:evidence for specific intrinsic nitrergic and VIP-positive neurons inthe human eye. In6estigati6e Ophthalmology and Visual Science,35, 592–599.

Gartner, J., & Fischer, F. (1989). Experimental autonomic neuropa-thy in the choroid of streptozotocin-diabetic rats. Electron micro-scopic observations. Retina, 9, 49–58.

Grimes, P. A., McGlinn, A. M., Koeberlein, B., & Stone, R. A.(1994). Immunoreactivity in autonomic innervation of the cat eye.Journal of Comparati6e Neurology, 348, 234–243.

Guglielmone, R., & Cantino, D. (1982). Autonomic innervation ofthe ocular choroid membrane in the chicken. Cell and TissueResearch, 222, 417–431.

Klooster, J., Beckers, H. J. M., Ten Tusscher, M. P. M., Vrensen, G.F. J. M., van der Want, J. J. L., & Lamers, W. P. M. A. (1996).Sympathetic innervation of the rat choroid: an autoradiographictracing and immunohistochemical study. Ophthalmic Research,28, 36–43.

Lee, M. K., & Cleveland, D. W. (1994). Neurofilament function anddysfunction: involvement in axonal growth and neuronal disease.Current Opinion in Cell Biology, 6, 34–40.

Liem, R. K. H., Keith, C. H., Leterrier, J. F., Trenckner, E., &Shelanski, M. L. (1981). Chemistry and biology of neuronal andglial intermediate filaments. Cold Spring Harbour Symposia onQuantitati6e Biology, 46, 341–350.

Liem, R. K. H., Yen, S. H., Solomon, G. D., & Shelanski, M. L.(1978). Intermediate filaments in nervous tissues. Journal of CellBiology, 79, 637–645.

Linder, J. E. (1978). A simple and reliable method for silver impreg-nation of nerves in paraffin sections of soft and mineralizedtissues. Journal of Anatomy, 27, 543–551.

Matsusaka, T. (1981). An evidence for adrenergic involvement in thechoroidal circulation. Graefes Archi6e for Clinical and Experimen-tal Ophthalmology, 216, 17–21.

Matsusaka, T. (1982). Cytoarchitecture of choroidal melanocytes.Experimental Eye Research, 35, 461–469.

Mawas, J. (1952). Sur l’inervation vaso-motrice de le choroide.Comptes Rendus de l ’Association des Anatomistes, 69, 679.

Miller, A. S., Coster, D. J., Costa, M., & Furness, J. B. (1983).Vasoactive intestinal polypeptide immunoreactive nerve fibers inthe human eye. Australian Journal of Ophthalmology, 11, 185–193.

Nilsson, S. F. E., Linder, J., & Bill, A. (1985). Characteristcs of uvealvasodilation produced by facial nerve stimulation in monkeys,cats and rabbits. Experimental Eye Research, 40, 841–852.

Nixon, R. A., Paskevich, P. A., Sihag, R. K., & Thayer, C. Y. (1994).Phosphorylation on carboxyl terminus domains of neurofilamentproteins in retinal ganglion cell neurons in vivo: influences onregional neurofilament accumulation, interneurofilament spacing,and axon caliber. Journal of Cell Biology, 126, 1031–1046.

Nuzzi, R., Gugglielmone, R., & Grignolo, F. M. (1992). Fluorescenthistochemical demonstration of adrenergic terminations in thehuman choroid. Experimental Eye Research (Suppl. 1), 55, 50.

Priestley, J. V. (1987). Immunocytochemical techniques for the local-ization of neurochemically characterized nerve pathways. In A. J.Turner, & H. S. Bachelard, Neurochemistry : a Practical Approach.Oxford: IRL Press, 65–112.

Ramırez, J. M., de Hoz, R., Salazar, J. J., Ramırez, A. I., Trivino, A.,& Garcıa-Sanchez, J. (1990). Fibras amielınicas coroideas. Estu-dio inmunohistoquımico. Archi6os de la Sociedad Espanola deOftalmologıa In6est, 3, 77–82.

Ramırez, J. M., Trivino, A., Ramırez, A. I., & Garcıa-Sanchez, J.(1989). Study of the choroidal innervation through immunocyto-

chemical technics. In R.W. Flower, Abstract The II InternationalSymposium on the Choroid (pp. 32–42) Maryland, R.W. Flower.

Ruskell, G. L. (1971). Facial parasympathetic innervation of thechoroidal blood-vessels in monkeys. Experimental Eye Research,12, 166–172.

Seiger, A., Dahl, D., Ayer-Lelievre, C., & Bjorklund, H. (1984).Appearance and distribution of neurofilament immunoreactivityin iris nerves. Journal of Comparati6e Neurology, 223, 457–470.

Sternberger, L. A., & Sternberger, N. H. (1983). Monoclonal anti-bodies distinguish phosphorylated and nonphosphorylated formsof neurofilaments in situ. Proceedings of the National Academy ofSciences of The United States of America, 80, 6126–6132.

Stone, R. A. (1986). Neuropeptide Y and the innervation of thehuman eye. Experimental Eye Research, 42, 349–355.

Stone, R. A. (1986). Vasoactive intestinal polypeptide and the ocularinnervation. In6estigati6e Ophthalmology and Visual Science, 27,951–957.

Stone, R. A., & McGlinn, A. M. (1988). Calcitonin gene-relatedpeptide immunoreactive nerves in human and Rhesus monkeyeyes. In6estigati6e Ophthalmol and Visual Science, 29, 305–310.

Stone, R. A., McGlinn, A. M., & Kuwayama, Y. (1988). Galanin-likeimmunoreactive nerves in the porcine eye. Experimental EyeResearch, 46, 457–461.

Stone, R. A., Tervo, T., Tervo, K., & Tarkhanen, A. (1986). Vasoac-tive intestinal polypeptide-like immunoreactive nerves to the hu-man eye. Acta Ophthalmologica, 64, 12–18.

Sverdlick, J. (1945). Sobre la organizacion vasculo-conectiva de lacoroides. Archi6os de histologıa normal y patologica, 2, 513–529.

Terenghi, G., Polak, J. M., Allen, J. M., Zhang, S. Q., Unger, W. G.,& Bloom, S. R. (1983). Neuropeptide Y-immunoreactive nerves inthe uvea of guinea pig and rat. Neuroscience Letters, 42, 33–38.

Terenghi, G., Polak, J. M., Ghatel, M. A., Mulderry, P. K., Butler, J.M., Unger, W. G., & Bloom, S. R. (1985). Distribution and originof calcitonin gene-related peptide (CGRP) immunoreactivity inthe sensory innervation of the mammalian eye. Journal of Com-parati6e Neurology, 233, 506–516.

Terenghi, G., Polak, J. M., Probert, L., McGregor, G. P., Blank, M.A., Butler, J. M., Unger, W. G., Zhang, S. Q., Cole, D. F., &Bloom, S. R. (1982). Mapping, quantitative distribution andorigin of substance P- and VIP- containing nerves in the uvea ofguinea pig eye. Histochemistry, 75, 399–417.

Terenghi, G., Zhang, S. Q., Unger, W. G., & Polak, J. M. (1986).Morphological changes of sensory CGRP-immunoreactive andsympathetic nerves in peripheral tissues following chronic dener-vation. Histochemistry, 86, 89–95.

Trojanowski, J. Q., Obrocka, M. A., & Lee, V. M-Y. (1983). Acomparison of eight different chromogen protocols for thedemonstration of immunoreactive neurofilaments or glialfilaments in rat cerebellum using the peroxidase–antiperoxidasemethod and monoclonal antibodies. Journal of Histochemistry andCytochemistry, 31, 1217–1226.

Trojanowski, J. Q., Obrocka, M. A., & Virginia, M.-Y. L. (1985).Distribution on neurofilament subunits in neurons and neuronalprocesses: immunohistochemical studies of bovine cerebellumwith subunit-specific monoclonal antibodies. Journal of Histo-chemistry and Cytochemistry, 33, 557–563.

Tsai, S.-H., Tew, J. M., & Shipley, M. T. (1992). Development ofcerebral arterial innervation: synchronous development of neu-ropeptide y (NPY)- and vasoactive intestinal polypeptide (VIP)-containing fibers and some observations on growth cones. De6el-opment Brain Research, 69, 77–83.

Uusitalo, H., Lehtosalo, J., Palkamo, A., & Toivanen, M. (1984).Vasoactive intestinal polypeptide (VIP)-like immunoreactivity inthe human and guinea-pig choroid. Experimental Eye Research,38, 435–437.

Vadoud-Seyedi, J., Lespargnard, L., Renard, N., Deraemaecker, R.,& Ghanem, G. E. (1994). Immunohistological evaluation ofneurofi-


lament 200 as an early nerve regeneration marker. Research inSurgery, 6, 18–21.

Vega, J. A., Vazquez, E., Naves, F. J., Del Valle, M. E., Calzada, B.,& Represa, J. J. (1994). Immunohistochemical localization of thehigh-affinity NGF receptor (gp 140-trkA) in the adult humandorsal root and sympathetic ganglia and in the nerves and sensorycorpuscles supplying digital skin. Anatomical Records, 240, 579–588.

Vinores, S. A., Derevjanik, N. L., Mahlow, J., Hackett, S. F., Haller,J. A., deJuan, E., Frankfurter, A., & Campochiaro, P. A. (1995).Class III B-tubulin in human retinal pigment epithelial cells inculture and in epiretinal membranes. Experimental Eye Research,60, 385–400.

Vitadello, M., Triban, C., Fabris, M., Dona, M., Gorio, A., &Schiafino, S. (1987). A developmentally regulated isoform of150 000 molecular weight neurofilament protein specifically ex-

pressed in autonomic and small sensory neurons. Neuroscience,23, 931–941.

Wolter, J. R. (1960). Nerves of the normal human choroid. Archi6esof Ophthalmology, 64, 120–124.

Wuerker, R., & Kirkpattrick, J. B. (1972). Neuronal microtubules,neurofilaments and microfilaments. International Re6iews of Cy-tology, 33, 45–57.

Xu, Z., Dong, L.-Y., & Cleveland, D. W. (1994). Neuronal intermedi-ate filaments: new progress on an old subject. Current Opinion inNeurobiology, 4, 655–661.

Yen, S. H., & Fields, K. L. (1981). Antibodies to neurofilament, glialfilament, and fibroblast intermediate filament protein bind todifferent cell types of the nervous system. Journal of Cell Biology,88, 115–126.

Zhao, Y., & Szaro, B. G. (1995). The optic tract and tectal ablationinfluence the composition of neurofilaments in regenerating opticaxons of Xenopus lae6is. Journal of Neuroscience, 15, 4629–4640.

.

Documents

Immunohistochemical study of rabbit choroidal innervation