Experimental Eye Research

Transcription

1 Experimental Eye Research 90 (2010) 478e492 Contents lists available at ScienceDirect Experimental Eye Research journal homepage: Anatomy of the human corneal innervation Carl F. Marfurt *, Jeremiah Cox 1, Sylvia Deek 1, Lauren Dvorscak Indiana University School of Medicine e Northwest, 3400 Broadway, Gary, IN 46408, United States article info abstract Article history: Received 4 November 2009 Accepted in revised form 16 December 2009 Available online 29 December 2009 Keywords: corneal nerves subbasal nerves The anatomy of the human corneal innervation has been the subject of much investigation; however, a comprehensive description remains elusive. The purpose of the present study was to provide a detailed description of the human corneal innervation using a novel approach involving immunohistochemically stained anterior-cornea whole mounts. Sixteen donor corneas aged 19e78 years were cut with a 6.0 mm trephine into a central plug and two peripheral rims. Each specimen was sectioned serially on a cryostat to produce several 100 mm-thick stromal sections and a 100e140 mm-thick anterior-cornea whole mount that contained the entire corneal epithelium and much of the anterior stroma. The corneal innervation was stained with a primary antibody against beta neurotubulin and subjected to rigorous quantitative and qualitative analyses. The results showed that a mean of , uniformly spaced, main stromal nerve bundles entered the cornea at the corneoscleral limbus. The bundles averaged mm in diameter, were separated by a mean spacing of mm, and entered the cornea at a mean distance of mm from the ocular surface. Each stromal bundle gave rise through repetitive branching to a moderately dense midstromal plexus and a dense subepithelial plexus (SEP). The SEP was comprised of modest numbers of straight and curvilinear nerves, most of which penetrated Bowman's membrane to supply the corneal epithelium, and a more abundant and anatomically complex population of tortuous, highly anastomotic nerves that remained largely confined in their distribution to the SEP. SEP density and anatomical complexity varied considerably among corneas and was less dense and patchier in the central cornea. A mean of stromal nerves penetrated Bowman's membrane to supply the central 10 mm of corneal epithelium (2.60 nerves/mm 2 ). The density of Bowman's membrane penetrations was greater peripherally than centrally. After entering the epithelium, stromal nerves branched into groups of up to twenty subbasal nerve fibers known as epithelial leashes. Leashes in the central and intermediate cornea anastomosed extensively to form a dense, continuous subbasal nerve plexus, while leashes in the peripheral cornea demonstrated fewer anastomoses and were less complex anatomically. Viewed in its entirety, the subbasal nerve plexus formed a gentle, whorl-like assemblage of long curvilinear subbasal fibers, 1.0e8.0 mm in length, that converged on an imaginary seam or gentle spiral (vortex) approximately mm inferonasal to the corneal apex. Mean subbasal nerve fiber density near the corneal apex was mm/mm 2 and mean subbasal and interconnecting nerve fiber diameters in the same region were mm and mm, respectively. Intraepithelial terminals originated exclusively as branches of subbasal nerves and terminated in all epithelial layers. Nerve terminals in the wing and squamous cell layers were morphologically diverse and ranged in total length from 9 to 780 mm. The suprabasal layers of the central corneal epithelium contained approximately terminals/mm 2. The results of this study provide a detailed, comprehensive description of human corneal nerve architecture and density that extends and refines existing accounts. An accurate, detailed model of the normal human corneal innervation may predict or help to understand the consequences of corneal nerve damage during refractive, cataract and other ocular surgeries. Ó 2009 Elsevier Ltd. All rights reserved. 1. Introduction * Corresponding author. Tel.: þ ; fax: þ addresses: cmarfurt@iun.edu (C.F. Marfurt), coxjj@iun.edu (J. Cox), sylvia. deek@comcast.net (S. Deek), ldvorsca@iun.edu (L. Dvorscak). 1 These two authors contributed equally to this work. The human cornea is the most densely innervated surface tissue in the body. In addition to their important sensory functions, corneal nerves help maintain the functional integrity of the ocular surface by releasing trophic substances that promote corneal epithelial homeostasis and by activating brainstem circuits that /$ e see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi: /j.exer

2 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e stimulate reflex tear production and blinking. Consequently, damage to corneal nerves as the result of surgery, trauma or disease leads to diminished corneal sensitivity and possible transient or long-term alterations in the functional integrity of the ocular surface. The anatomy of the human corneal innervation has been studied for many years by a variety of methods, including light and electron microscopy, immunohistochemistry and in vivo confocal microscopy (IVCM). Despite these efforts, a detailed, comprehensive description of human corneal nerve architecture remains elusive. Light and electron microscopic investigations of human corneal nerve distribution, density, and ultrastructure (Al-Aqaba et al., 2009; Muller et al., 1996, 1997; Schimmelpfennig, 1982; Ueda et al., 1989; Zander and Weddell, 1951) have generated most of the data on which current models of corneal innervation are based (Muller et al., 2003). More recently, IVCM has been used to image the innervation in healthy and diseased corneas and has provided considerable new information on the morphology, density, and disease- or surgical-induced alterations of corneal nerves, with special emphasis on the subbasal nerve plexus (Oliveira-Soto and Efron, 2001; Patel and McGhee, 2005; see Patel and McGhee, 2009, for review and additional references). IVCM is especially useful for imaging corneal nerves near the apex because of the relative ease of obtaining good quality tangential images in this region; however, although IVCM provides excellent resolution, it is often incapable of imaging reliably corneal epithelial terminals and very small diameter subbasal and stromal nerves. The purpose of the present study was to provide a detailed and comprehensive description of the human corneal innervation using a novel approach that involved immunohistochemical staining of 100e140 mm-thick anterior-cornea whole mounts and stromal sections. This method provides excellent visualization of the main stromal bundles, midstromal plexus, subepithelial plexus, subbasal nerve plexus, and intraepithelial terminals throughout the entire cornea. The results of this study were presented in preliminary fashion at ARVO (Marfurt et al., 2008). 2. Materials and methods Sixteen research corneas (6 pairs and 4 single corneas), from donors ranging in age from 19 to 81 years old (mean: 57.6 years) were obtained from various eye banks. Seven of the corneas were oriented at time of harvesting by placing an indelible ink mark at the superior pole. The remaining nine corneas were not oriented. Most of the corneas (n ¼ 14) were placed directly into room temperature 10% neutral buffered formalin; two corneas were placed in optisol preservative medium for 6 h prior to immersion fixation. Death-to-preservation (DTP) times ranged from 4.5 to 18.0 h. All corneas were shipped overnight in cold fixative solution to our laboratory at which time they were transferred immediately into ice cold 0.1 M phosphate buffered saline (PBS) containing 30% w/v sucrose. In preparation for immunohistochemical staining, fourteen of the corneas were cut into three large standardized pieces. The central corneal button was removed with a 6.0 or 6.5 mm corneal trephine. The remaining cornea, comprised of a 2.0e3.0 mm wide rim of peripheral cornea and about 0.5e1.0 mm of attached sclera, was then cut with a razor blade into nasal and temporal halves (oriented corneas) or two random halves (non-oriented corneas). Each of the three pieces was sectioned tangentially at 20 Cin a cryostat according to the following protocol. Each specimen was placed on a glass microscope slide and pressed, epithelial side facing down, on a flat platform of frozen OCT compound on a cryostat chuck. A total of four 100 mm-thick sections were then cut from the posterior surface of each central corneal button, and five or six 100 mm-thick sections were cut from the posterior surface of each peripheral corneal rim. All sections were collected in serial order in ice cold 0.1 M PBS. Tissue sectioning was then halted and the residual corneal specimen, still embedded in its frozen OCT matrix on the cryostat chuck, was immersed in a petri dish filled with ice cold PBS. As the OCT compound melted, the residual tissue, consisting of the entire corneal epithelium and approximately 50e100 mm of anterior-corneal stroma, was released intact into the PBS. The specimens thus obtained will be referred to henceforth in this paper as anterior-cornea whole mounts. To increase their permeability, all anterior-cornea whole mounts and 100 mm-thick stromal sections were incubated overnight at 37 C in 0.01% hyaluronidase (type IV-S, SigmaeAldrich, Inc., St. Louis, MO) and 0.1% ethylenediaminetetraacetic acid (EDTA; Sigma) in 0.1 M PBS ph 5.3 (Barrett et al., 1999). The next morning the tissues were rinsed three times for 15 min each in PBS containing 0.3% Triton X-100 (PBS-TX), and incubated for 2 h in blocking serum (1% bovine serum albumin in PBS-TX). The tissues were then incubated overnight at room temperature (RT) on a rocker table in a mouse monoclonal antibody directed against neuronal class III beta-tubulin (TuJ1, 1:500, Covance Research Products, Berkeley, CA). After three more PBS-TX rinses, the tissues were incubated for 2 h at RT in secondary antibody (biotinylated horse anti-mouse IgG, 1:200; Vector Laboratories, Burlingame, CA), rinsed again in PBS- TX, and incubated for 2 h at RT in avidin-biotin-horseradish peroxidase complex (ABC reagent; Vector Laboratories). After three more PBS-TX rinses, the tissues were incubated for 8 min at RT in 0.1% diaminobenzidine (Sigma) and 0.009% H and then rinsed three times in PBS and twice in distilled water. The tissues were then mounted in serial order on chrome alum-gelatin coated slides, air-dried, dehydrated in graded alcohols, cleared in xylene, and coverslipped with Permount under weighted coverslips. Two additional, non-oriented corneas were processed separately from those described above and were used mainly to study the distances from the ocular surface at which stromal nerve bundles entered the peripheral cornea. Each of these two corneas was cut from apex to limbus with a razor blade into eight, equalsized wedge-shaped pieces. Each wedge was then frozen in OCT compound and six serial sections per specimen were prepared in a cryostat by cutting the tissue parallel to the corneoscleral limbus at a distance precisely 6.0 mm from the corneal apex. The sections were collected in PBS and processed free-floating for neurotubulin immunohistochemistry as described above except without hyaluronidase/edta pretreatment Qualitative observations Qualitative analyses of corneal nerve architecture and morphology were performed using a Leica DM4000 research microscope. Comprehensive schematic line drawings of corneal main stromal bundles, midstromal plexus, subepithelial plexus, subbasal nerve plexus, and intraepithelial terminals were prepared at final magnifications of 30e300 by using a drawing tube attached to the microscope. Color images were captured with a Leica DFC420C digital camera Quantitative assessments Main stromal nerve bundles The main stromal nerve bundles were analyzed at their point of entry from the corneoscleral limbus into the peripheral cornea. Five distinct parameters were assessed: total number, distribution, spacing between adjacent bundles, diameter, and distance from the ocular surface. The first three parameters were determined by plotting the locations of all main stromal nerve bundles in the

3 480 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 anterior-cornea whole mount and serial 100 mm-thick sections onto composite illustrations made with the aid of a drawing tube at a final magnification of 80. The location of the corneoscleral limbus was approximated on each completed illustration by drawing a 12.0 mm diameter circle centered on the corneal apex. The total number of stromal nerve bundles, and the distances between adjacent bundles at their points of intersect with the edge of the circle, were recorded. The diameters of 130 randomly selected main stromal nerves from four different corneas were determined at their points of entry into the peripheral cornea by using the measuring tool device of Image J (NIH) on calibrated digital images taken on a Leica microscope at a magnification of 40. The depths at which the main stromal nerve bundles entered the cornea at the limbus were investigated in 30 mm-thick sections cut parallel to the corneoscleral limbus. The best section from each corneal wedge, as judged by nerve staining quality and lack of sectioning artifacts, was selected for study. The distance from the epithelial surface to the midpoint of each main stromal bundle contained within that section was measured by using the Image J measuring tool Subepithelial plexus (SEP) Nerve fiber diameters in four corneas were determined from digital images of the SEP taken with a 40 objective. The SEP contained two main types of nerve fibers (see Results): straight or curvilinear nerves that penetrated Bowman's membrane to enter the corneal epithelium, and tortuous anastomotic nerves that remained confined in their distribution to the SEP. The diameters of 100 fibers from each population were determined by using the measuring tool device of Image J on calibrated images. The diameters of the straight or curvilinear nerves were measured within 10e50 mm of where the fibers penetrated Bowman's membrane Bowman's membrane penetration sites The sites where stromal nerve fibers passed through Bowman's membrane and continued into the epithelium as subbasal nerves are known as Bowman's membrane penetration sites. The total number, distribution, and density of penetration sites were determined in the central 10 mm areas from three anterior-cornea whole mounts by plotting the locations of every penetration site onto high magnification line drawings prepared with a drawing tube. Subbasal nerves in the perilimbal cornea, most of which originated directly from the limbal plexus, were not included in these counts Subbasal nerve fibers Subbasal nerve fiber density (NFD) and diameters were determined in the central cornea of six donors aged 19e78 years old (mean age, 51.3 years). For NFD measurements, a 0.5 mm 2 rectangular area (1.0 mm horizontal 0.5 mm vertical) centered precisely on the corneal apex was designated for analysis. Digital images of the central subbasal nerve plexus were taken on a light microscope with a 10 objective and enlarged to a final magnification of 200. All subbasal nerves and interconnecting fibers located within the 0.5 mm 2 field were traced carefully with a calibrated tracing tool (Image J). Nerve fiber density was reported as the total combined length in millimeters of all subbasal and interconnecting nerve fibers within the 0.5 mm 2 sample area and reported as mm/mm 2. Subbasal nerve diameters were measured in photomontages of the central cornea (4.0 mm horizontal 0.5 mm vertical) taken with a 40 objective. A 3 mm horizontal line centered on the corneal apex was positioned across the montage using Adobe Photoshop. The diameter of every subbasal nerve or interconnecting fiber that transected the horizontal guide line was determined at its point of intersect with the line by using the measuring tool of Image J. A total of 460 subbasal nerves and 97 interconnecting fibers were measured Epithelial nerve terminal density Epithelial nerve terminal densities in small areas of the central cornea in three anterior-cornea whole mounts were determined by mapping the location and morphology of all nerve terminals in the suprabasal cell layers onto high magnification line drawings prepared with the aid of a drawing tube attached to the light microscope. The areas that were evaluated measured 1.0 mm 2 (two cases) or 0.4 mm 2 (one case) and were selected on the basis of optimal nerve terminal staining quality. 3. Results 3.1. Technical note: effect of death-to-preservation time on corneal nerve staining Anterior-cornea whole mounts with death-to-preservation (DTP) times between 7 and 13 h (n ¼ 6) yielded optimal nerve staining and provided superior demonstrations of the corneal SEP, subbasal nerves and intraepithelial terminals. Anterior-corneal whole mounts with DTP times in excess of 13 h (n ¼ 6) also Fig. 1. a. Main stromal bundles (arrows) entering the peripheral cornea at the corneoscleral limbus in an anterior-cornea whole mount. b. High magnification image of a main stromal bundle in a 100 mm-thick section.

4 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e Fig. 2. Main stromal nerve bundles enter the peripheral cornea uniformly from all directions. The dashed line indicates the approximate location of the corneoscleral limbus. contained numerous intensely stained nerves; however, many epithelial nerves showed varying signs of degeneration, such as swelling of subbasal nerves and fragmentation or loss of intraepithelial terminals. Anterior-cornea whole mounts with DTP times less than 7 h (n ¼ 4) were unexpectedly resistant to immunohistochemical staining and contained very few well-stained subbasal nerves or intraepithelial terminals, except at cut edges of the tissue or in areas where the epithelium had been damaged inadvertently during processing Main stromal nerve bundles A mean of main stromal nerve bundles (range, 53e89) entered the human cornea at the corneoscleral limbus Fig. 4. Distances from the corneal surface at which main stromal nerve bundles enter the peripheral cornea at the corneoscleral limbus. n ¼ 131 nerves. (Fig. 1). The nerve bundles were distributed uniformly about the corneal circumference (Fig. 2) and were separated from one another by a mean distance of 0.48 mm 0.40 mm. The stromal bundles entered the peripheral cornea at a mean distance of mm from the corneal surface; however, occasional bundles entered as far anterior as 56 mm or as deep as 543 mm (Figs. 3 and 4). The mean diameter of the main stromal bundles was mm (Fig. 5) Midstromal nerve plexus Soon after entering the cornea, each stromal nerve bundle gave rise through repetitive branching to varying numbers of progressively smaller and smaller stromal nerves that anastomosed frequently, often at highly acute branch points, to form a moderately dense midstromal plexus. The distal branches of the midstromal nerves often coursed centrally for several millimeters and on occasion crossed the geographic center of the cornea to reach the opposite side. The midstromal plexus in the peripheral stroma occupied roughly the anterior one-half of the stroma while in the central cornea the plexus occupied approximately the anterior onethird. On rare occasions a few isolated nerves were observed in the posterior half of the stroma; however, nerves were never seen adjacent to Descemet's membrane or in the corneal endothelium. The midstromal plexus was most dense in the peripheral cornea and decreased progressively in density and anatomical complexity Fig. 3. Main stromal nerve bundles (arrows) entering the peripheral cornea near the corneoscleral limbus. In some instances, multiple stromal bundles enter the cornea at the same location but at different depths (right side of figure). Fig. 5. Main stromal nerve bundle diameters.

5 482 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 Fig. 8. Stromal free nerve endings (arrows). Fig. 6. Low magnification survey image of the peripheral corneal innervation. Main stromal bundles (msbs) branch and give origin to a moderately dense midstromal plexus composed of straight/curvilinear fibers (arrowheads) and tortuous fibers (arrows). Some of the straight or curvilinear fibers penetrate Bowman's membrane and give rise in a more superficial plane to subbasal nerve fibers (snf). in a central direction. In the peripheral cornea, the plexus was comprised mainly of medium and small diameter nerves with straight or curvilinear trajectories, and varying numbers of tortuous fibers (Fig. 6). In the central cornea, the midstromal plexus was less dense and consisted mostly of small diameter, curvilinear nerves (Fig. 7). A few nerve fibers in all areas of the cornea terminated in the stroma as free nerve endings (Fig. 8) Subepithelial plexus Most midstromal nerve fibers continued into the narrow band of anterior stroma located immediately beneath Bowman's membrane and gave rise to a dense, roughly two-dimensional, subepithelial plexus (SEP). The SEP contained two main types of nerve fibers: a modest population of straight or curvilinear nerves, and a more abundant and anatomically complex population of tortuous, highly anastomotic nerves (Fig. 9a, b). The straight or curvilinear anterior stromal nerves ranged in size from 0.37 to 9.06 mm in diameter and had a mean diameter of mm (Fig. 10). Most of these nerves penetrated Bowman's membrane and continued into the corneal epithelium as subbasal nerves (see below). Other straight or curvilinear nerves gave origin at their distal termini to the tortuous SEP nerve fibers. The tortuous SEP fiber population consisted of mixtures of small and medium diameter fibers that ranged in size from 0.24 to 3.28 mm. The nerves coursed in seemingly random directions and often anastomosed extensively to form complex, grid-like meshworks (Fig. 11). The density and anatomical complexity of the SEP tortuous fiber population varied considerably from cornea to cornea and from region to region in the same cornea. In general, the SEP was much more dense in the peripheral and intermediate cornea, and less dense, inconsistent, and patchy in the central cornea (Fig. 12). It was impossible because of the plexiform nature of the SEP (e.g., Fig. 11) to determine how most tortuous nerves terminated; however, a few terminated in the subepithelial stroma as free nerve endings, and a very small number (typically only 1e4 per cornea) gave origin to collaterals that penetrated Bowman's membrane and formed subbasal nerves Nerve penetration into the corneal epithelium Fig. 7. Mid-stromal plexus in the central and pericentral cornea. The area illustrated is 6 mm in diameter and is centered on a point 0.5 mm from the corneal apex (asterisk). The drawing is a composite of an anterior-cornea whole mount and two 100 mm-thick stromal sections and illustrates in two-dimensions the entire midstromal nerve plexus. Deeper sections of the corneal stroma from this specimen contained no nerve fibers. The distal continuations of the midstromal nerves (e.g., arrows) give origin to the subepithelial plexus (not illustrated). The area inside the box is illustrated in greater detail in Fig. 12. Relatively modest numbers of nerve fibers in the SEP penetrated Bowman's membrane to give rise to subbasal nerves (Fig. 13). A mean of SEP nerves (range: 156e269) penetrated Bowman's membrane inside the 10 mm diameter zone centered on the corneal apex; this corresponds to an overall density of 2.60 penetrations/mm 2 (Fig. 14). The density of nerve penetration sites was about twice as high (3.18/mm 2 vs. 1.55/mm 2 ) in the pericentral cornea (3e5 mm from the corneal apex) than in the central cornea (0e3 mm from the apex). Nerve penetration density was lowest in the area surrounding and inferonasal to the corneal apex. Additional subbasal nerves in the extreme peripheral (i.e., perilimbal) corneal epithelium originated directly from the limbal plexus or from short, radially directed collaterals of limbal nerves (Fig. 15). The latter penetrations were difficult to quantify and are not included in the data shown in Fig Subbasal nerve plexus Immediately after penetrating Bowman's membrane, each stromal nerve branched into one or more subbasal nerves that

6 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e Fig. 9. a. Peripheral corneal innervation, focused on the subepithelial plexus. The boxed area near the upper left corner of fig a is shown at higher magnification in b. a. The SEP consists of modest numbers of straight or curvilinear fibers (arrowheads) and a dense, plexiform network of tortuous nerve fibers (arrows). b. SEP straight or curvilinear fibers (arrowheads) penetrate Bowman's membrane (at open circles) to give rise to subbasal nerves. The tortuous nerve fibers (arrows) anastomose frequently and give the SEP its highly characteristic plexiform appearance. coursed parallel to the ocular surface near the interface of Bowman's membrane and the basal epithelium (Fig. 16). Most of the subbasal nerves observed in this study were of smooth caliber; however, some were beaded in appearance. The term epithelial leash is defined as a group of subbasal nerves that derives from the same parent anterior stromal nerve (Rozsa and Beuerman, 1982; Schimmelpfennig, 1982; Chan-Ling, 1989). At their points of origin, each leash consisted initially of 1e20 subbasal nerves; however, the absolute number of subbasal fibers in a given leash fluctuated continuously in the proximo-distal direction due to repetitive nerve branching and anastomotic connections via thin, obliquelyoriented interconnecting fibers. Epithelial leashes in the central Fig. 10. Diameters of SEP straight and curvilinear nerve fibers. Measurements were taken within 50 mm of where the nerves penetrated Bowman's membrane. Fig. 11. Schematic line drawing of a small area of the SEP located approximately 3 mm from the corneal apex. Abundant, thin tortuous nerve fibers anastomose frequently to form a dense, felt-like meshwork. Embedded within this meshwork are two straight/ curvilinear nerves (arrows) that penetrate Bowman's membrane (open circles) and continue into the basal epithelial layer as subbasal nerves (e.g., arrowheads).

7 484 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 Fig. 12. High magnification line drawing of the midstromal and subepithelial plexuses in the central corneal region indicated by the box in Fig. 7. SEP density in the central cornea is often patchy in nature and areas of moderate-to-high tortuous nerve fiber density (e.g., large asterisks) often intermingle with areas of lower tortuous nerve fiber density (smaller asterisks) in seemingly random fashion. and intermediate cornea anastomosed extensively with one another in both the lateral (side-to-side) and proximal-to-distal axes to form a dense, homogenous subbasal nerve plexus in which clear-cut boundaries between adjoining leashes were no longer discernible (Fig. 17). Epithelial leashes in the peripheral cornea, in contrast, were less numerous, more widely separated, demonstrated fewer side-to-side anastomoses, and contained few thick subbasal nerves (Fig. 18). When viewed in its entirety, the subbasal nerve plexus comprised a gentle, spiral-like assemblage of long, curvilinear subbasal nerve fibers that converged on an imaginary center, or vortex, located inferior and slightly nasal to the corneal apex (Figs. 19 and 20). As a consequence of the whorl-like arrangement, subbasal nerves in the region of the corneal apex coursed in a predominantly near-vertical (1 o'clock-to-7 o'clock) direction, while subbasal nerves in the intermediate and peripheral cornea coursed in various directions consistent with their relative positions within the whorl. Subbasal nerves in the extreme peripheral cornea, especially in the inferonasal quadrant, often coursed parallel to the limbus for varying distances before changing direction gradually or abruptly to reorient towards the vortex (Figs. 15 and 21). The geographical center of the subbasal nerve vortex was located between 2.18 and 2.92 mm (mean ¼ 2.51 mm 0.23 mm) from the corneal apex. Near the center of the vortex, the distal segments of the subbasal fibers in some corneas fused to form an anastomotic network that spiraled gently in either a clockwise (two corneas, one OD one OS) or counterclockwise (one cornea, OD) direction (Fig. 22). In other cases (four corneas), the subbasal nerves did not form a prominent spiral but ended on opposing sides of an imaginary seam-like interface (e.g., Fig. 19). The mean subbasal nerve fiber density in the central corneas from six different donors was mm/mm 2 (range: 39.6e53.3 mm/mm 2 ). No correlation between subbasal nerve density and donor age was observed (Fig. 23). A horizontal line drawn through the corneal apex intersected a mean of subbasal nerves/millimeter (not including interconnecting axons). Mean subbasal and interconnecting nerve fiber diameters in the central cornea were mm and mm, respectively (Figs. 24 and 25). Subbasal nerve fiber length varied widely. The shortest subbasal nerves measured less than 1 mm in total length and were concentrated near the center of the vortex (e.g., Fig. 22b) and in the perilimbal region. The longest subbasal nerves were concentrated in the superior corneal quadrant (e.g., Fig. 19) and often traveled distances of 6.5e8.0 mm or longer Intraepithelial terminals Intraepithelial terminals were observed in all corneas examined in the current study; however, the morphological integrity of these delicate structures varied widely from cornea to cornea in a manner that did not correlate well with death-to-preservation time. The descriptions that follow are based on observations from three Fig. 13. Stromal nerve penetrations through Bowman's membrane in anterior-cornea whole mounts (aec) and in a 30 mm-thick perpendicular section (d). a. A straight stromal nerve (arrow) penetrates Bowman's membrane (open circle) and branches into multiple subbasal nerves (arrowheads). b, c. Some stromal nerves (arrows) split into multiple branches (arrowheads) immediately prior to, or while penetrating, Bowman's membrane. Each branch then gives rise to one or more subbasal nerves. d. A stromal nerve (arrow) penetrates Bowman's membrane (asterisk) and continues as a subbasal nerve (arrowheads).

8 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e Fig. 14. a. Location of stromal nerve penetrations (solid circles) through Bowman's membrane in a 48-year-old cornea. The area illustrated is 10 mm in diameter and centered on the corneal apex (A). V, center of subbasal nerve vortex. The numbers, 1 through 5, indicate distance in millimeters from the apex. b. Density of Bowman's membrane nerve penetration sites as a function of distance from the corneal apex. n ¼ 611 nerve penetration sites from three different corneas. corneas in which the majority of intraepithelial terminals were reasonably well preserved. The term intraepithelial terminal as used in this study refers to the entire epithelial axon distal to its point of origin from a subbasal nerve and includes all of its collateral branches and terminal expansions ( nerve endings ). Intraepithelial terminals originated exclusively as branches of subbasal nerves. The terminals were distributed abundantly throughout all layers of the epithelium and varied considerably in total length, predominant directional orientation, and morphological complexity. Terminals in the basal epithelial cell layer generally coursed parallel to the parent subbasal nerve fibers, branched relatively infrequently, and gave rise to small numbers of bulbous nerve endings (e.g., Fig. 18). Intraepithelial terminals in more superficial epithelial layers were generally more complex. Some of latter axons ended as single terminal expansions; however, the majority branched one or more times at acute angles and supported small numbers of variously oriented, preterminal collaterals in complex treelike morphologies (Figs. 26 and 27). In some cases, groups of intraepithelial terminals that originated from the same subbasal nerve were oriented in a remarkably uniform direction (Fig. 28). Nerve terminal density in the wing and squamous cell layers of the central cornea was mm/mm 2, or terminals/mm 2. The total lengths of intraepithelial terminals in the wing and squamous cell layers, including all collateral branches and nerve endings, ranged from 9.0 to mm Summary of major findings The major findings of this study are summarized in Table Discussion Fig. 15. Nerve entry into the peripheral corneal epithelium. Subbasal nerves in the peripheral cornea originate either directly from the limbal plexus (e.g., arrows), or from short, radially directed branches of the limbal plexus (arrowheads). Open circles, stromal nerve penetrations through Bowman's membrane. A main stromal bundle (msb) is visible entering the cornea from the limbus in a deeper plane of focus. The results of this study provide a detailed and comprehensive description of the human corneal innervation that adds to existing knowledge in the field. The use of immunohistochemically stained, thick anterior-cornea whole mounts made it possible to visualize the entire corneal innervation, except for the deepest stromal nerves, in a single preparation and revealed important threedimensional relationships among midstromal nerves, SEP, subbasal plexus and epithelial nerve terminals. Diaminobenzidine (DAB) was used to label the nerves in this study because the chromogen is permanent and does not fade under prolonged illumination; this made it possible to prepare large-format, detailed illustrations of the immunohistochemically stained nerves with a drawing tube attached to the light microscope. The results of this study demonstrated a critical and unexpected relationship between death-to-preservation (DTP) time and successful corneal nerve staining. Anterior-cornea whole mounts

9 486 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 Fig. 16. Subbasal nerve fibers in an anterior-cornea whole mount (a) and in a perpendicular, 30 mm-thick section (b). a. An anterior stromal nerve (arrow) divides and penetrates Bowman's membrane (open circles) to form an epithelial leash comprised of approximately 16e18 subbasal nerve fibers. The black line shows the approximate plane of section of the 30 mm-thick section (from a different cornea) shown in figure b. b. Subbasal nerve fibers (e.g., arrows) travel roughly parallel to one another in the basal epithelial cell layer close to Bowman's membrane (bm). A small diameter stromal nerve (arrowhead) penetrates Bowman's membrane near the center of the field. e, corneal epithelium. s, stroma. Calibration bar equals 20 mm in both figures. with DTP times between 7 and 13 h provided the best results, whereas whole mounts with DTP times less than 7 h contained very few well-stained epithelial nerves. The latter observation suggests that some postmortem degradation of the corneal epithelium prior to immersion fixation may be required for successful hyaluronidase penetration and permeabilization of the thick specimens Stromal nerve bundles The mean number of main stromal nerve bundles that supply the human corneas examined in this study ( ) is slightly less than the 80 bundles per cornea reported by Zander and Weddell (1951), and greater than the 44 bundles reported by Al-Aqaba et al. (2009). The main stromal nerve bundles seen here are distributed uniformly around the corneal circumference and enter the cornea radially from all directions (Al-Aqaba et al., 2009). This observation contradicts previous statements that stromal nerves enter the human cornea in greater numbers at the nasal and temporal poles (Darwish et al., 2007; Muller et al., 2005; Solomon et al., 2004). The reason for this apparent discrepancy is not known. The human cornea receives most of its sensory innervation from two long ciliary nerves that enter the posterior globe medial and lateral to the optic nerve and course forward in the suprachoroidal space at the nasal and temporal meridians (Bron et al., 1997; Vaughn, 1992). Prior to reaching the corneoscleral limbus, the nerves branch repetitively into smaller bundles and anastomose extensively with branches of the short ciliary nerves (May, 2004; Trivino et al., 2002) to form 60e80, uniformly distributed nerve Fig. 17. a. Detailed schematic line drawing of the central subbasal nerve plexus. The area illustrated is 2.28 mm wide 2.80 mm high (6.38 mm 2 ) and is from the same corneal region shown in Fig. 12. b. Digital image of a small area of the subbasal nerve plexus illustrated schematically in figure a. Large arrow, thick subbasal nerve fiber. Small arrows, thin subbasal nerve fiber. Arrowheads, interconnecting fibers.

10 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e Fig. 18. Subbasal nerves in the peripheral cornea. The SNFs anastomose less frequently and are generally thinner and more uniform in diameter than are central SNFs. Arrows, nerve terminals in the basal epithelial cell layer (see later text for details). bundles that approach the limbus radially from all directions. Clinical observations seem consistent with this view, in that block excisions of anterior uveal tumors or epithelial ingrowths located near the limbus at the medial or lateral poles are not associated with higher incidences of neuroparalytic keratitis (Groh et al., 2002). The results of the present study may have implications for understanding the sensory loss that occurs following cataract surgery. The mean distance between main stromal bundles reported in the present study is about 0.5 mm; thus, a 2.8e3.0 mm long curvilinear, clear corneal incision such as is commonly used for foldable and injectable intraocular lenses would transect a mean of w6 major stromal nerve bundles, or approximately 6e11% of the total corneal innervation. Following phacoemulsification, corneal sensitivity is reduced in the vicinity of the incision and, to a more variable degree, in the central cornea (Khanal et al., 2008; Sitomopul et al., 2008). This sensory loss is most likely explained by combined damage to main stromal bundles and subbasal nerves in the vicinity of the incision. Fig. 20. Low magnification photomontage of the subbasal nerve plexus in a 6 mm diameter, anterior-cornea whole mount from a 48-year-old cornea. The orientation notch at the top of the whole mount is the superior pole. The results of this study also suggest that transection of main stromal bundles may produce a significant corneal denervation during LASIK surgery. The most common corneal flap thickness used in LASIK surgery is approximately 160 mm and photoablation may remove as much as an additional 100 mm of stromal tissue. The results of our study showed that approximately 37% (48 out of 131) Fig. 19. Subbasal nerve plexus in a 6 mm diameter central button from a 61 year-old cornea. For clarity, only the largest diameter subbasal nerve fibers have been illustrated. Individual subbasal nerves follow straight or curvilinear trajectories and converge on an imaginary center, or vortex (asterisk), located approximately 2.5 mm inferonasal to the corneal apex. Fig. 21. Subbasal nerves in the perilimbal region of the inferonasal quadrant. The nerves travel roughly parallel to the limbus for varying distances before altering their direction and orienting towards the center of the vortex.

11 488 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 Fig. 22. Subbasal nerve vortices. Subbasal nerve fibers in some corneas rotate counterclockwise (a) while others rotate clockwise (b) about the center of the vortex. The circular region illustrated in a is 400 mm in diameter, centered on the vortex and superior is towards the top. In b, two thick subbasal nerves penetrate the epithelium (arrows) near the center of the vortex and arch conspicuously in a clockwise direction. of main stromal bundles entered the peripheral cornea at a depth of 250 mm or less from the corneal surface and would likely be damaged by LASIK surgery Midstromal nerves and the subepithelial plexus The density of the midstromal nerve plexus is greater in the peripheral cornea than in the central cornea. It seems reasonable to speculate that the central cornea contains fewer stromal nerve fibers because most of the epithelial innervation in this region comes from stromal nerves that penetrate Bowman's membrane in more peripheral locations (Fig. 14). The morphology of the SEP is exceedingly complex and raises interesting questions concerning the role of these fibers in corneal neurophysiology. The results of the present study confirm previous IVCM observations (Visser et al., 2009) that the SEP comprises two highly distinctive types of nerve fibers; however, only the straight and curvilinear nerves provide significant innervation to the corneal epithelium. The more numerous tortuous nerve fibers, which give the SEP its distinctive morphological appearance, originate as branches of the straight or curvilinear nerves and supply less than 5% of the epithelial innervation. It was not possible to determine the ultimate destination of most of the tortuous nerve fibers in this study and consequently their functional significance remains purely conjectural. Some tortuous fibers may constitute a reserve population of SEP nerve fibers that assist in epithelial reinnervation after corneal injuries. Others may make bouton en passant contacts on resident cells and possess sensory or trophic functions. Electron microscopic and IVCM studies have shown that some stromal axons form intimate contacts with keratocytes (Matsuda, 1968; Muller et al., 1996; Visser et al., 2009); however, because keratocytes are present throughout the corneal stroma in extremely high density, it remains to be shown whether these contacts are functional or coincidental. The relatively small numbers of SEP and midstromal free nerve endings observed in the present study suggest that they are likely of limited physiological significance. No attempt was made to calculate SEP density in this study because of the high intra- and inter-cornea variability and patchy nature of the plexus. The current study confirms previous reports based on IVCM that the SEP is denser in appearance in the peripheral cornea and less dense and highly variable in the central cornea (Auran et al., 1995; Visser et al., 2009; Oliveira-Soto and Efron, 2001). The patchy nature of the SEP in the central cornea was a consistent and factual observation and was not caused by incomplete immunohistochemical staining Stromal nerve penetrations through Bowman's membrane The results of this study show that the subbasal nerve plexus originates from a modest number (mean ¼ 204) of stromal nerves that penetrate Bowman's membrane within a 10 mm diameter circle centered on the corneal apex. The mean number of Fig. 23. Subbasal nerve fiber densities in the central corneas of six donors aged 19e78 years old. Fig. 24. Subbasal nerve fiber diameters in the central cornea. Subbasal nerves ranged in size from 0.40 to 5.66 mm. n ¼ 460 fibers from six different donors.

12 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e Fig. 25. Interconnecting fiber diameters in the central cornea. Interconnecting fibers ranged in size from 0.29 to 1.82 mm. n ¼ 97 fibers from four different corneas. penetrations seen here is slightly less than that reported by previous workers. Ueda et al. (1989) reported a total of 400 penetrations per cornea, and Al-Aqaba et al. (2009) reported a total of 155e185 penetrations within an 8 mm diameter zone centered on the corneal apex. The differences in the total number of penetration sites reported in these three studies are most likely explained by variations in the corneal area examined, and the criteria used to count the nerve penetration sites. In the present study, a cluster of penetrations that originated from the same stromal nerve (e.g., Fig. 13b, c) was counted as a single penetration site. The results of the present study have shown that nearly 80% of all Bowman's membrane penetration sites in the central (10 mm diameter) corneal button are located in the pericentral region between 3 and 5 mm from the corneal apex. Additional large numbers of subbasal nerves (not quantified in the current study) enter the perilimbal cornea (5e6 mm from the corneal apex) directly from the limbal plexus. Collectively, this high concentration of pericentral and peripheral subbasal nerves may provide an important source of central reinnervation after LASIK surgery, penetrating keratoplasty, and cataract surgery. Circular flaps in Fig. 27. High magnification schematic line drawing of intraepithelial nerve terminals in the wing and squamous cell layers of a 1 mm 2 area of central cornea. LASIK surgery and graft diameters in penetrating keratoplasty average 8.5e9.5 mm and 7.5e8.0 mm in diameter, respectively, and leave intact a 1e2 mm wide peripheral rim of tissue from which central reinnervation occurs slowly via subbasal nerve elongation across the wound margin (Calvillo et al., 2004; Erie et al., 2005; Niederer et al., 2007; Patel et al., 2007; Tervo et al., 1985). Nerve penetration sites in the central cornea of living eyes appear by IVCM as bright, irregular or disc-shaped areas approximately 20e40 mm in diameter (Oliveira-Soto and Efron, 2001; Patel and McGhee, 2005). The relative paucity and widespread distribution of these penetration sites in the central cornea makes them difficult to locate by electron microscopy (Matsuda, 1968; Muller et al., 1996). As the stromal nerve penetrates Bowman's membrane, new axoplasm and membrane are added to the nerve, thus causing the subbasal nerves to elongate continuously in a proximal-todistal direction (Auran et al., 1995; Patel and McGhee, 2008). Fig. 26. Intraepithelial nerve terminals as seen in an anterior-corneal whole mount (a) and in a perpendicular, 30 mm-thick section (b). a. Intraepithelial terminals originate (e.g., circle) exclusively as branches of subbasal nerves (arrowheads, deeper plane of focus). Nerve terminals in the suprabasal epithelium often possess multiple collateral branches, and each branch is capped by a bulbous terminal expansion (arrows). b. Nerve terminals end blindly as free nerve endings (arrows) in all layers of the corneal epithelium.

13 490 C.F. Marfurt et al. / Experimental Eye Research 90 (2010) 478e492 architectural feature was not recognized in most previous light and electron microscopic studies because the orientations of the corneas were in many cases not known and the authors examined only limited numbers of small sample areas from the central or pericentral cornea (Muller et al., 1996; Schimmelpfennig, 1982). Muller et al. (2003) proposed a model of subbasal nerve architecture that showed subbasal nerves oriented in a superior-to-inferior direction over the corneal apex, and in a nasal-to-temporal direction in surrounding areas. Patel and McGhee (2005) subsequently published a 5 mm 2 IVCM photomontage of the central subbasal plexus in a living eye that confirmed the roughly vertical orientation of subbasal nerves over the corneal apex, and further showed that subbasal nerves in the central and pericentral cornea converged in a whorl-like pattern on a point located approximately 1e2 mm inferior to the apex. The architectural map of the subbasal nerve plexus demonstrated in the present study confirms the IVCM findings of Patel and McGhee (2005) and extends their findings to include observations on subbasal nerve orientation in more peripheral and perilimbal corneal regions Subbasal nerve vortex Fig. 28. Epithelial nerve terminals. The two peripheral leashes shown here are each comprised of only one or two thick subbasal nerves (arrows). Each subbasal nerve gives rise to a large cluster of intraepithelial terminals that radiates asymmetrically at roughly right angles to the parent nerve Architectural organization of the subbasal nerve plexus The results of the present study have shown that the human subbasal nerve plexus, when viewed in its entirety over the whole cornea, forms a gentle spiral-like pattern whose center is located about 2.5 mm inferonasal to the corneal apex. This unique The morphology and location of the subbasal nerve vortices described here confirm and extend previous reports of whorl-like patterns of subbasal nerves located just inferior to the corneal apex in human (Al-Aqaba et al., 2009; Auran et al., 1995; Patel and McGhee, 2005, 2008; Ueda et al., 1989) and rodent (Dvorscak and Marfurt, 2008; Leiper et al., 2009; Yu and Rosenblatt, 2007) corneas. The mechanisms that govern the formation and maintenance of this spiral-like arrangement remain largely conjectural. According to one hypothesis, basal epithelial cells near the corneoscleral limbus migrate centripetally in a whorl-like fashion towards the corneal apex in response to chemotropic guidance, electromagnetic cues, and population pressures (Collinson et al., 2002, 2004; Dua et al., 1996). Subbasal nerves, occupying narrow intercellular spaces within the migratory epithelial sheet (Muller et al., 1996), are pulled along by mechanical forces and undergo compensatory horizontal elongation (Auran et al., 1995; Patel and McGhee, 2008). It has also been postulated that shearing forces exerted on the corneal surface by the eyelids during spontaneous Table 1 Summary of quantitative data. Corneal nerve type and location Mean Range Main stromal nerve bundles Total number per cornea e89 Diameters mm 4.77e40.28 mm Distance between adjacent bundles mm 0.03e2.61 mm Distance from corneal surface at limbal entry point mm 56e543 mm Subepithelial plexus (SEP) Diameters of straight/curvilinear fibers mm 0.37e9.06 mm Diameters of tortuous fibers Highly variable 0.24e3.28 mm Bowman's membrane penetration sites (central 10 mm of cornea) Total number e269 Density, central cornea (<3 mm from corneal apex) /mm e1.84/mm 2 Density, peripheral cornea (3e5 mm from corneal apex) /mm e4.32/mm 2 Subbasal nerve plexus Distance from corneal apex to center of vortex mm 2.18e2.92 mm Subbasal nerve plexus density, central cornea mm/mm e53.34 mm/mm 2 Number of subbasal nerve fiber intersects/mm, central cornea e29.67 Subbasal nerve fiber diameters, central cornea mm 0.40e5.66 mm Interconnecting fiber diameters, central cornea mm 0.29e1.82 Intraepithelial nerve terminals, central cornea, suprabasal layers only Nerve terminal density (number of terminals/mm 2 ) /mm 2 331e972/mm 2 Nerve terminal density (mm/mm 2 ) mm/mm e37.6 mm/mm 2 Nerve ending density (number of endings/mm 2 ) /mm 2 593e1514/mm 2