Morphology and Neurochemistry of Canine Corneal Innervation

Abstract

purpose. To determine the architectural pattern and neuropeptide content of canine corneal innervation.

methods. Corneal nerve fibers in normal dog eyes were labeled immunohistochemically with antibodies against protein gene product (PGP)-9.5, calcitonin gene-related peptide (CGRP), substance P (SP), vasoactive intestinal polypeptide (VIP), and tyrosine hydroxylase (TH). Relative innervation densities and distribution patterns for each fiber population were assessed qualitatively by serial line-drawing reconstructions and quantitatively by computer-assisted analyses.

results. More than 99% of all corneal PGP-9.5–immunoreactive (IR) nerves contained both CGRP and SP, approximately 30% contained TH, and none contained VIP. Distribution patterns of corneal PGP-9.5–, CGRP-, SP-, and TH-IR nerves were indistinguishable, except that TH-IR fibers were absent from the corneal epithelium. Morphologically, canine corneal innervation consisted of a rich anterior stromal plexus, divided on the basis of morphologic criteria into anterior and posterior levels, and a rich epithelial innervation, characterized by large numbers of horizontally oriented, basal epithelial “leash” formations. Leash axons in all quadrants of the corneal epithelium oriented preferentially toward a common locus in the perilimbal cornea.

conclusions. The results of this study demonstrate for the first time the detailed architectural features, distinctive basal epithelial leash orientations, and peptidergic content of canine corneal innervation. The normal innervation pattern described in this study will provide other investigators with essential baseline data for assessing corneal nerve alterations in canine patients with spontaneous chronic corneal epithelial defects (SCCED) and other ocular diseases or injuries.

The cornea is one of the most richly innervated tissues in the body, receiving dense sensory innervation from the trigeminal nerve and modest sympathetic innervation from the superior cervical ganglion. Corneal sensory and sympathetic nerves exert important neuromodulatory effects on corneal epithelial physiology, including, regulation of ion transport; cell proliferation, differentiation, adhesion, and migration; and wound healing. ¹ Damage to the corneal nerves by surgery, trauma, or disease deprives the corneal epithelium of essential neurotrophic influences and leads to the development of a serious degenerative condition known as neurotrophic keratitis. ² The pathophysiology of this condition is complex and is characterized by decreased epithelial cell proliferation, increased surface cell exfoliation, spontaneous recurrent epithelial erosion, and impaired wound healing after corneal injuries. ^{3 4} The mechanisms by which corneal nerves exert their trophic effects are currently under investigation in several laboratories, and involve, at least in part, the release of biologically active neuropeptides and neurotransmitters, such as substance P (SP), calcitonin gene-related peptide (CGRP), and norepinephrine. ^{5 6 7 8 9 10}  

Spontaneous chronic corneal epithelial defects (SCCED) are observed frequently in dogs in veterinary ophthalmic practice. The clinical features of the epithelial defects are similar to those in humans and consist of repeated episodes of spontaneous epithelial erosion, typically without any clear-cut history of previous corneal trauma. ^{11 12} Because corneal nerves are essential to the maintenance of a healthy corneal epithelium, it is tempting to hypothesize that the epithelial defects in canine eyes with SCCED may be related, at least in part, to anatomic or functional deficits in corneal innervation. 

The anatomy of mammalian corneal innervation has been well described in several species, including humans, rabbits, cats, and rats. ^{13 14 15 16} In contrast, the anatomy of canine corneal innervation has received only limited attention ¹⁷ and the peptidergic content of canine corneal nerves has, to our knowledge, never been investigated. The purpose of the present study was to provide a comprehensive morphologic description of canine corneal innervation in normal canine eyes and, in a companion article that appears in this issue of IOVS, ¹⁸ to describe alterations in corneal neuroanatomy, sensitivity, and SP content in animals with SCCED. 

Materials and Methods

Eight corneas from four dogs were examined in this investigation. All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Each animal was euthanatized for reasons other than use in this study and all were free of obvious ocular disease. The eyes were enucleated within 30 minutes of death and immersion fixed whole in 4% paraformaldehyde-0.2% picric acid in 0.1 M phosphate buffer (pH 7.4) for 24 to 72 hours. Each cornea, including approximately 1 to 2 mm of the contiguous corneoscleral limbus, was dissected from surrounding ocular tissues and stored in fresh fixative at 4°C until it was sectioned. 

Before sectioning, each cornea was cut with a razor blade into six to eight wedge-shaped segments extending from corneal apex to limbus. Each segment was soaked for 30 to 60 minutes in 0.1 M phosphate buffer containing 30% sucrose, followed by an additional soaking for 5 minutes in optimal cutting temperature (OCT) compound (Miles Laboratories, Elkhart, IN). Approximately one half of the corneal wedges were sectioned in a cryostat in the anterior–posterior direction (tangential to the corneal surface), and the remainder were sectioned perpendicular to the corneal surface. Serial 30-μm-thick sections were collected in tissue culture wells filled with chilled phosphate-buffered saline (PBS). 

Immunohistochemical labeling of corneal nerves was performed on free-floating tissue sections by using a standardized avidin-biotin-horseradish peroxidase procedure. Sections were incubated overnight at 4°C in primary antisera directed against one of five neuronal markers: PGP-9.5 (1:5000; Chemicon International, Inc., Temecula, CA), CGRP (1:5000; Amersham, Arlington Heights, IL), SP (1:4000; Peninsula, Belmont, CA), tyrosine hydroxylase (TH, 1:400; Pel Freeze Biological; Rogers, AR), and vasoactive intestinal polypeptide (VIP; 1:500; Peninsula). Immunolabeled nerve fibers were visualized by using a kit (Vectastain ABC Elite; Vector Laboratories, Burlingame, CA) with diaminobenzidine (DAB) as the substrate. Specificity of the immunocytochemical procedure was confirmed for each antiserum by incubating randomly selected sections in normal serum without the appropriate primary antibody. 

All sections were critically examined in a light microscope (BH2; Olympus, Lake Success, NY). The innervation density and distribution pattern for each nerve fiber population under investigation was documented by making a series of line drawings with a drawing tube attached to the microscope and by photomicrographs (T-Max 100 film; Eastman Kodak, Rochester, NY). 

Computer-assisted quantitative analyses of immunostained corneal nerve fibers were performed in two corneas from different animals to determine the percentages of corneal PGP-9.5–immunoreactive (IR) nerves that contained CGRP, SP, TH, and VIP. Fifty, 30-μm-thick perpendicular sections from each cornea were collected in serial order, and every fifth section was processed immunohistochemically for PGP-9.5, CGRP, SP, TH, or VIP. Immunolabeled nerve fibers in four randomly selected sections from each group were then drawn at ×50 magnification by using a drawing tube attached to the light microscope, and the area occupied by immunolabeled nerve fibers was determined by image analysis software (NIH Image, provided in the public domain by the National Institutes of Health, Bethesda, MD, and available at http://www.nih.gov/od/oba). Nerve-density analyses were conducted on the central half of each corneal section, extending from the corneal apex for a distance of 4.75 mm toward the limbus. The results were then averaged for each group, and the innervation densities of corneal CGRP-, SP-, TH-, and VIP-IR nerve fibers were calculated as percentages of baseline PGP-9.5–IR innervation density. 

Results

Corneal specimens stained immunohistochemically with primary antisera against PGP-9.5, CGRP, or SP contained abundant immunolabeled nerve fibers. The numbers and distribution patterns of the PGP-9.5–, CGRP-, and SP-IR fiber populations were essentially identical, except for an increased density of PGP-9.5 innervation in the corneoscleral limbus. Thus, the following anatomic descriptions of canine corneal innervation represent composites of complementary observations from PGP-9.5–, CGRP-, and SP-immunostained material. 

Limbal Plexus

The limbal plexus comprised a dense, superficial nerve network arranged as an 0.8- to 1.0-mm-wide, ringlike band around the peripheral cornea. The origins of the limbal fibers were numerous and complex and included collateral branches of stromal and subconjunctival fibers in passage to the cornea, recurrent collaterals from the peripheral corneal plexus, and perivascular fibers associated with the rich limbal vasculature. Morphologically, the limbal plexus was subdivided into two zones. The outer (periscleral) zone (Fig. 1A) was dominated by large numbers of predominantly perivascular nerve fascicles and a relatively modest stromal plexus whose individual axons traveled apparently randomly through the limbal stroma, unrelated to vascular elements. The inner (pericorneal) zone (Fig. 1B) comprised a considerably more dense meshwork of highly branched and anastomotic axons and small-diameter fascicles. Many of the fibers in the inner zone formed intimate associations with vascular elements of the superficial limbal arcade; other fibers continued through the corneoscleral transition zone and anastomosed with axons in the peripheral anterior stromal plexus (Fig. 1B) . The limbal and conjunctival epithelia contained modest numbers of short, wavy, beaded axons with predominantly radial orientations. 

Corneal Stromal Innervation

Most nerve fibers entered the peripheral cornea at the corneoscleral limbus in a series of 14 to 18 prominent, radially directed, superficial stromal nerve bundles. The bundles were of relatively uniform size and were distributed at regular intervals around the limbal circumference (Fig. 2) . Each bundle contained approximately 30 to 40 light microscopically visible axons. Additional, smaller nerve fascicles (not illustrated) entered the peripheral cornea between, and slightly superficial to, the main stromal bundles. 

Soon after entering the cornea, the main stromal bundles exhibited complex series of repetitive, dichotomous branchings to form elaborate axonal trees. The distal branches of these trees anastomosed extensively with one another at angular junctions to form a dense, anatomically complex stromal plexus that extended uninterrupted into all areas of the cornea from limbus to limbus. The latter plexus occupied approximately the anterior half (0.4–0.5 mm) of the corneal stroma and was subdivided on the basis of innervation density and morphologic criteria into distinct posterior and anterior levels. The posterior level of the stromal plexus (Fig. 3A) contained modest numbers of primarily small- to medium-diameter bundles and scattered individual axons. Many of the axons in this location were unusually long, straight, and filamentous (Fig. 4A) . The anterior level of the stromal plexus was much more densely innervated and morphologically complex (Fig. 3B) . An especially delicate meshwork of exceptionally thin, preterminal axons occupied the region immediately beneath the epithelial basement membrane (Fig. 4B) . 

In contrast to the richly innervated anterior stroma, the posterior half of the canine corneal stroma was largely devoid of innervation. However, a few rare nerve fibers were seen in some animals in an extreme posterior location adjacent to the corneal endothelium (Fig. 4C) . 

Epithelial Innervation

The corneal epithelium contained dense accumulations of PGP-9.5–, CGRP-, and SP-IR nerve fibers and terminals. On entering the basal epithelial cell layer, most intraepithelial axons formed unique preterminal arborizations known as “epithelial leashes” (Fig. 5A) . Each epithelial leash formation comprised a family of two to six axons attached to a single subepithelial fiber. Individual axons in a formation coursed horizontally through the basal epithelial cell layer tangential to the corneal surface and roughly parallel to one another for distances of 1.0 to 1.4 mm. Each axon measured 1.2 to 3.5 μm in diameter, but most were less than 2.5 μm. It could not be determined within the limits of resolution of the light microscope whether individual immunostained leash axons represented single unmyelinated axons or tightly packed collections of multiple unmyelinated axons. ^{19 20}  

Leash formations demonstrated a high degree of directional specificity. Adjacent leash formations were always aligned in an orderly, coordinated fashion, as if converging on some common, but nonapical, corneal region (Fig. 3C) . Corneas donated for use in this study were not marked for orientation before enucleation; thus, it was not possible to determine whether the leashes arched consistently toward a particular anatomic locus (e.g., nasomedial, temporolateral). 

As the leash axons coursed horizontally through the basal epithelium, they gave origin to a profusion of thin, prominently beaded ascending branches (Figs. 3D 5B) . The ascending axons divided extensively and formed irregular clusters of short terminal branches that ended throughout the basal, wing, and squamous epithelial layers. Most axonal endings were tipped by a single conspicuous, bulbous terminal expansion. 

Neurochemistry of Corneal Innervation

Quantitative analyses of immunolabeled corneal nerve fibers in semiadjacent perpendicular sections through the central cornea (Fig. 6) demonstrated that SP, CGRP, TH, and VIP were expressed within more than 99%, more than 99%, 29.7%, and 0%, respectively, of all corneal PGP-9.5–IR nerves. 

TH-IR axons were located mainly in small- to medium-sized nerve bundles in the anterior stromal plexus; however, TH-IR fibers were never observed in the corneal epithelium (Figs. 6 7) . TH-IR nerves were extremely abundant in the corneoscleral limbus; the latter fibers mainly formed rich perivascular plexuses and were many times more numerous than limbal CGRP- and SP-IR nerves. VIP-IR fibers were absent in the corneal stroma or epithelium; however, modest numbers of perivascular VIP-IR fibers were consistently seen in the corneoscleral limbus. 

Discussion

Corneal Nerve Anatomy

The major anatomic features of canine corneal innervation demonstrated in the present study are similar to those described previously in human, cat, rabbit, and rat corneas. ^{13 14 15 16} Thus, canine corneas contain a rich limbal plexus, multiple radially directed stromal nerve bundles, a dense highly anastomotic subepithelial plexus, and dense epithelial innervation distinguished by families of horizontal axons known as basal epithelial leashes. The results of this study confirm and extend earlier morphologic observations in the canine cornea by Barrett et al. ¹⁷ and suggest that comparable patterns of corneal nerve organization exist, with minor interspecies differences, in all mammals. 

A novel finding of the present study concerns the directional orientation of canine corneal epithelial leashes. Previous descriptions of leash axon orientation in mouse, rat, cat, rabbit, and human eyes suggest that leash axons pursue radial or quasiradial, whorled trajectories and are aligned so that the leashes appear to converge on the corneal apex or a region slightly adjacent to the apex. ^{14 15 16 19 21} In marked contrast to these observations, leash formations in the canine corneas examined in the present study course horizontally in different directions (circumferential, oblique, or radial) depending on the area of cornea in which they are located. To our knowledge, similar observations have been described in the ophthalmologic literature only once. ²² In that study, all epithelial leash formations in the rabbit cornea were conspicuously oriented toward the nasomedial limbus. 

The regulatory influences that determine directional leash orientation in the mammalian cornea have received little attention. It is tempting to speculate that radially directed leash axons, as seen in many mammalian species, reflect comigration of axons and corneal basal epithelial cells. According to the X, Y, Z hypothesis of corneal epithelial renewal, ²³ new basal epithelial cells continuously develop from proliferating stem cells in the corneoscleral limbus and migrate in a centripetal direction toward the central cornea. Ultrastructurally, basal epithelial cells are tightly anchored to one another by extensive lateral membrane interdigitations and desmosomal attachments, and epithelial leash axons occupy narrow intercellular spaces between adjacent basal epithelial cells or lie sequestered inside cytoplasmic infoldings of the cells. ²⁴ Thus, leash axons wedged between adjacent columns of migrating cells may undergo compensatory, horizontal elongation as they are dragged along in a radial direction with the comigrating cells. In support of this hypothesis, time-lapse, scanning slit lamp confocal observations of living human corneas show that basal epithelial cells and leash axons migrate centripetally (radially) in concert with one another at a rate of approximately 10 to 20 μm per day and that this tandem movement is made possible by the continual addition of new nerve material at the site of entry of the nerve into the epithelium. ^{21 25}  

Alternatively, it is possible that leashes develop radial orientations independent of epithelial cell dynamics. In the latter case, migrating basal epithelial cells may use the preexisting, horizontal axons as scaffolds for centripetal-directed movement. 

The predominantly nonradial leash orientations seen in dog corneas and in one study of rabbit corneas ²² suggest that leash orientations in these species may be governed by alternative mechanisms. Leash axons in the rabbit corneas examined by deLeeuw and Chan ²² consistently coursed toward the nasomedial limbus. Leash axons in the dog eyes examined in the present study also appeared to converge on some peripheral, perilimbal site; however, the precise site of nerve convergence could not be determined. All the canine eyes examined in this study demonstrated comparable patterns of leash orientation, and none of the dogs had histories of ocular disease or trauma; thus, the pattern of leash orientation in this study is unlikely to have been caused by injury-induced, nerve-remodeling events. Possibly, leash axons in rabbits and dogs elongate prenatally and/or postnatally in response to concentration gradients of one or more neuronotrophic factors released by corneal epithelial cells. ²⁶ The functional significance of a predominantly nasomedial convergence of leash axons, such as has been described in the rabbit cornea, ²² remains to be determined. 

Corneal Neuropeptides and Neuroenzymes: Methodological Considerations

The results of the quantitative analyses reported in the present study show that more than 99% of all the canine corneal nerve fibers contained CGRP and SP and that approximately 30% contained TH. The validity of these estimates depends on the assumption that PGP-9.5 immunohistochemistry successfully labels all canine corneal nerve fibers and provides an accurate baseline indicator of overall innervation density. In support of this presumption, the results of an earlier immunohistochemical study of canine dorsal root ganglia have shown that virtually all canine primary sensory neurons express PGP-9.5. ²⁷ Work in other laboratories has shown that PGP-9.5 is contained in extremely high percentages (if not all) of cutaneous sensory, cholinergic, and peptidergic axons. ²⁸ PGP-9.5 immunohistochemistry has yielded spectacular demonstrations of peripheral innervation density in a diverse number of tissues and organs, including some canine tissues, ^{29 30} and side-by-side comparisons with other peripheral-nerve–staining methods has shown that PGP-9.5 immunohistochemistry is qualitatively and quantitatively superior to acetylcholinesterase, neuron-specific enolase, and neurofilament triplet protein histochemical staining procedures. ²⁸ Thus, it seems reasonable to conclude that the pan-neuronal marker PGP-9.5 is ubiquitously distributed within canine corneal nerve fibers. 

CGRP and SP

The high percentages of PGP-9.5–IR corneal nerves that expressed CGRP (>99%) and SP (>99%) make it virtually certain that these two neuropeptides colocalize in most, if not all, canine corneal nerves. ³¹ The results confirm previous immunohistochemical reports of CGRP- and SP-IR nerves in other mammalian corneas ^{32 33} ; however, the percentages of corneal nerves that contain these neuropeptides in other species remains unknown. The results of a recent comparative immunoassay study suggest that there may be considerable interspecies differences in corneal neuropeptide concentration. ³⁴  

The origins of the canine CGRP- and SP-IR nerves in the present study remain to be determined; however, ocular CGRP- and SP-IR nerves in other species are predominantly, ^{16 35} but perhaps not exclusively, ³¹ sensory. In contrast to the fact that nearly all (>99%) canine corneal nerves contain CGRP and SP, only 20% to 30% and 40% to 50% of mammalian trigeminal ganglion neurons express SP and CGRP, respectively. ³⁶ The exaggerated peptidergic innervation of the canine cornea most likely reflects the fact that most, if not all, corneal nerves are C-fiber and A-delta nociceptors ³⁷ and that nociceptive afferents generally originate from peptidergic, small- to medium-diameter sensory neurons. ^{36 38}  

The rich density of CGRP- and SP-IR nerves in the dog cornea makes it tempting to speculate that these peptides subserve ongoing trophic and regulatory processes in the corneal epithelium and that they, when released from corneal sensory nerves, stimulate corneal epithelial cells as part of the normal processes of tissue maintenance, physiologic renewal, and wound healing (for further discussion, see Murphy et al. ¹⁸ in this issue of IOVS). Recent evidence has shown that SP and CGRP modulate various aspects of corneal epithelial cell behavior, including, proliferation, adhesion, and migration. ^{5 6 8 9 39 40}  

Tyrosine Hydroxylase

In this study a substantial percentage (∼30%) of canine corneal nerves expressed TH. The general distribution pattern of the canine TH-IR fiber population mimicked that of the CGRP- and SP-IR fiber populations; however, compared with peptidergic axons, there were fewer individual TH-IR axons in the subepithelial plexus, an absence of nerves in the epithelium, and more fibers in the limbus. The absence of TH-IR intraepithelial axons could represent a real deficiency of TH-IR epithelial innervation or, alternatively, the amount of TH present in the thin, distal preterminal and terminal nerve segments could be below the sensitivity of the immunohistochemical technique. 

TH is the rate-limiting enzyme of catecholamine synthesis and is abundantly expressed in sympathetic neurons; thus, its presence in peripheral nerves is often interpreted as evidence of a fiber’s sympathetic nature (but see additional discussion, to follow). Corneal sympathetic nerve fibers occur in most, if not all, mammalian corneas ⁴¹ ; however, their relative densities (and therefore potential functional significance) demonstrate considerable interspecies differences. ⁴² Functionally, ocular sympathetic nerves have been implicated in the modulation of corneal epithelial ion transport, cell proliferation, and cell migration after corneal wound healing. ^{6 7 10 43 44} In the limbus, TH-IR nerve fibers may protect against overperfusion and breakdown of limbal blood–ocular barriers during acute elevations of intraocular blood pressure. 

If the premise that PGP-9.5 stains 100% of canine corneal nerves is accepted, then the results of this study suggest that TH colocalizes with CGRP and SP in approximately 30% of canine corneal nerves. Whether the TH-CGRP-SP–IR fibers represent sympathetic, sensory, or parasympathetic corneal fibers could not be determined in the present study; however, extrapolation of data from previous immunohistochemical studies suggests several plausible hypotheses. For example, the triple-labeled nerves may come from TH-IR sympathetic neurons in the superior cervical ganglion (SCG) that also express CGRP and SP. Indeed, substantial percentages of neurons in the canine SCG express CGRP under normal physiological conditions ⁴⁵ and approximately 20% to 40% of human SCG neurons express both TH and CGRP. ⁴⁶ Similarly, SP is found in 10% to 15% of mammalian SCG neurons, ⁴⁷ and after 48 hours in cell culture, most rat SCG neurons express both SP and TH. ⁴⁸  

Alternatively, TH-CGRP-SP–IR corneal fibers may constitute a subpopulation of peptidergic, trigeminal sensory axons. TH is synthesized in numerous trigeminal and spinal sensory ganglion neurons, ^{49 50} and some trigeminal TH-IR nerves innervate the cornea. ⁵⁰ Human corneas also contain substantial numbers of TH-IR nerves ^{19 50} ; however, their origin(s) remains speculative. Human and other primate corneal nerves apparently do not have additional catecholamine-synthesizing enzymes and only rarely contain detectable levels of catecholamines. ⁵¹ Thus a functional role for TH in corneal sensory nerve physiology remains unclear. 

Finally, the TH-CGRP-SP–IR corneal nerves observed in the current study may derive from parasympathetic neurons. Cat and rat corneas receive sparse parasympathetic innervation from the ciliary ganglion, ^{52 53} and most mammalian ciliary ganglia contain large numbers of CGRP-, SP-, and/or TH-IR neurons. ^{54 55}  

Vasoactive Intestinal Polypeptide

The results of the present study provided no evidence for corneal VIP-IR innervation except in modest numbers of perivascular fibers in the corneoscleral limbus. Observations in other species suggest that limbal VIP-IR fibers are vasomotor in function and that they derive from ocular parasympathetic ganglia. ⁵⁶ Detectable levels of VIP have been measured in some mammalian corneas by radioimmunoassay, ^{34 57} and isolated VIP-IR axons have been demonstrated in the rat cornea by immunohistochemistry. ¹⁶ The paucity or absence of direct corneal VIP-IR innervation in these studies suggests a minor role for this peptide in corneal epithelial cell physiology; however, modulation of corneal cell biology after VIP release and diffusion from limbal and uveal axons cannot be ruled out. 

 Supported in part by National Institutes of Health Grant EY10841 (CJM).

 Submitted for publication June 5, 2000; revised September 19, 2000; accepted October 6, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Innervation of the corneoscleral limbus. (A) PGP-9.5–IR axons in the outer limbal zone formed intimate perivascular associations with limbal arteries and veins of various diameters. (B) PGP-9.5–IR fibers in the inner limbal zone comprised dense networks of fine-diameter axons, many of which continued into the peripheral cornea to anastomose with fibers in the peripheral corneal plexus (arrows). Scale bar, (A) 100 μm; (B) 200 μm.

Figure 2.

 

Stromal nerve bundles (arrow) entering the cornea at the corneoscleral limbus. Fifteen large CGRP-IR nerve bundles are visible in this specimen. The cornea was artificially flattened by making six radial slits with a razor blade before sectioning.

Figure 3.

 

Densities and distribution patterns of immunolabeled PGP-9.5–IR corneal nerve fibers in representative 30-μm-thick sections cut tangential to the corneal surface. Orientation diagram (top) shows locations of drawings in (A–D). The sections are arranged sequentially from posterior to anterior to maintain consistency with the anatomic descriptions provided in text. (A) PGP-9.5–IR nerve fibers at midstromal level (posterior level of stromal plexus). The stromal nerve plexus at this level typically contained modest numbers of unusually thin, straight axons (arrows). The latter axons were randomly oriented, generated few collaterals, and coursed uninterrupted for long distances in a single stromal plane. (B) PGP-9.5–IR fibers in the subepithelial plexus. The nerve network extended ubiquitously from apex to limbus and comprised a strikingly complex, anastomotic meshwork of small-, medium-, and large-diameter nerve fascicles and numerous individual axons. (C) PGP-9.5–IR axon leashes in the basal epithelial cell layer. Individual leash axons demonstrated considerable regional variation in directional orientation; however, collectively they formed highly organized aggregates. In this particular specimen, leashes in the peripheral (perilimbal) cornea coursed circumferentially and parallel to the limbus, whereas leashes in progressively more central corneal areas demonstrated oblique, and then nearly radial, orientations. Arrows: direction in which the leashes traveled. The data presented suggest that the leash axons converged on a perilimbal site (not illustrated) to the lower left of the specimen. (D) PGP-9.5–IR terminal axons in the corneal epithelium. The 30-μm-thick section is composed mainly of superficial epithelium; however, a few basal epithelial leash axons (arrow) are visible. Multiple short, terminal branches (arrowheads) derived from each leash axon to provide rich innervation to the overlying superficial corneal epithelium.

Figure 4.

 

Innervation of the corneal stroma. (A) Long, filamentous SP-IR axons in the posterior layer of the stromal plexus. Compared with most stromal axons, filamentous axons were unusually thin, unbranched, and often crisscross haphazardly, with no obvious preferred directional orientation. (B) SP-IR nerve fibers in the subepithelial stroma. Individual axons followed tortuous trajectories and anastomose frequently to form a delicate plexiform meshwork. (C) A prominently beaded, SP-IR axon (arrows) in the extreme posterior stroma adjacent to the endothelium.

Figure 5.

 

PGP-9.5–IR innervation of the corneal epithelium as seen in horizontal (tangential) sections through the central cornea. (A) Several prominently beaded leash axons (arrows) meandered horizontally through the basal epithelial cell layer. (B) Cluster of fine terminal branches emanating from a single horizontal leash axon. The terminal axons originated from a common preterminal axon in a deeper plane of focus (arrow) and followed gentle, sweeping courses through the superficial epithelium before ending as bulbous terminal expansions (arrowheads). Scale bar, (A) 100 μm; (B) 50 μm.

Figure 6.

 

Distribution patterns and relative densities of corneal nerve fibers stained immunohistochemically for PGP-9.5, CGRP, SP and TH. Quantitative analyses of tissue sections such as those shown revealed that more than 99%, more than 99%, and 29.7% of canine PGP-9.5–IR fibers contained CGRP, SP, and TH, respectively.

Figure 7.

 

(A) TH-IR fibers in the subepithelial stroma were found mainly in small- and medium-sized bundles; however, small numbers of individual TH-IR axons were also seen. (B) Perpendicular section through the central cornea showing multiple TH-IR nerve fibers (arrows) in the anterior stromal plexus, but an apparent absence of TH-IR fibers in the corneal epithelium. Length of scale bar in (B) represents 1 mm when applied to (A) and 140 μm when applied to (B).

The authors thank Sean Campbell for superior technical assistance.