Architecture of human corneal nerves. - ARVO...

Architecture of Human Corneal Nerves

Linda J. Mutter, Gijs F.J. M. Vrensen, Liesbeth Pels, Bob Nunes Cardozo,and Ben Willekens

Purpose. The corneal innervation, mainly analyzed in light microscopical studies, has beendescribed as radially oriented stromal nerve bundles that ramify as leashes in the subbasalplexus. The current study aims to determine the orientation, the size, and the postmortemchanges of the nerve fibers in the subbasal plexus of the human cornea.

Methods. Before processing for light and electron microscopy, the position of the corneaswithin the enucleated eyes of persons with melanoma and pairs of postmortem eyes wasmarked. The orientation and postmortem changes of the fibers were studied in serial "enface" semithin sections, and the size was determined in random, ultrathin cross-sections.

Results. Thirteen and a half hours after death, the majority of the nerve fibers were degeneratedor gone. Nerve fiber bundles in the subbasal plexus run first in die 9-3 hours direction, thenafter bifurcation in the 12-3 hours direction and after a second bifurcadon again in the 9-3hours direction. From die main straight bundles, single-beaded fibers branch and runobliquely. Quantification of the nerve fibers shows an equally dense innervated central andcentral-peripheral cornea (mean fiber diameter, 0.4 /j,m) and a five to six times lower inner-vated peripheral cornea (mean fiber diameter, 0.67 fitn).

Conclusions. The nerve bundles in the subbasal plexus of the human cornea form a regulardense meshwork with equal density over a large central and central-peripheral area. Becauseof their size, the majority of the fibers can be classified as C-fibers. Invest Ophthalmol VisSci. 1997;38:985-994.

i. he nerve fiber pattern of the densely innervatedmammalian cornea has been described as radially ori-ented nerve bundles entering the corneal domainthrough the sclera.1"3 These bundles originate fromthe ophthalmic division of the trigeminal nerve2'4"6

and are located mainly in the stroma.1'2'6"8 After pass-ing Bowman's layer, they ramify and end within theepithelium as free nerve endings.8'9 These nerve end-ings are derived from myelinated A-<5 and unmyelin-ated C-nerve fibers.10"12 In the skin, A-<5 and C-nervefibers measure 1 to 5 //m and 0.2 to 2 //m, respectively.They are known to act as nociceptors, which monitormechanical, thermal, and chemical signals in periph-eral nerves including the cornea.1013 In the rabbit, thedensity of corneal nerve endings is 20 to 40 times thatof tooth pulp and 300 to 600 times that of skin.14

Although the human cornea is known to be extremelysensitive, the density of free nerve endings has not

From the Department of Morphology, The Netherlands Ophthalmic ResearchInstitute, Amsterdam, The Netherlands.Submitted for publication July 29, 1996; revised October 7, 1996; acceptedNovember 11, 1996.Proprietary interest category: N.Reprint requests: Linda Muller, Department of Morphology, The NetherlandsOphthalmic Research Institute, P. 0. Box 12141, 1100 AC Amsterdam, TheNetherlands.

been estimated. This is mainly because corneal nervesdegenerate rapidly after death. This is a disadvantagebecause discussions on corneal pain perception andspreading of neurotrophic viruses (Herpes) are moremeaningful if the distribution of human cornealnerves would be known. In the past decades, numer-ous studies have been performed on transmitters oc-curring in corneal nerve bundles. They appeared tostain positive with antibodies against classical neuro-transmitters, for instance acetylcholine,15'16 catechol-amines,17"19 and antibodies, against peptidergic trans-mitters such as substance p20"27 and calcitonin gene-related peptide.15'28"30 The presence of classical andpeptidergic transmitters may represent specific cor-neal nerve fiber subpopulations or may point at colo-calization of different transmitters in the same fibers.

Using different approaches, it became clear thatin the central cornea, nerve fiber bundles located atthe anterior border of Bowman's layer run parallel toeach other.8'9'31 Because the position of the corneasin the eye was not marked in these studies, it remainedinconclusive whether nerve fiber bundles run tempo-rally-nasally (9-3 hours), superiorly-inferiorly (12-6hours), or obliquely. Recently, it has been shown ele-gantly by in vivo confocal microscopy that human cor-

Investigative Ophthalmology & Visual ScCopyright © Association for Research in

e, April 1997, Vol. 38, No. 5on and Ophthalmology 985

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933200/ on 06/01/2018

986 Investigative Ophthalmology & Visual Science, April 1997, Vol. 38, No. 5

FIGURE l. (A) En face section of the central cornea (13.5 hours after death) at the level ofthe basal epithelial cells (*) and Bowman's layer (BL). Degenerating nerve fibers run parallelin the upper lower direction (arroius). (B) En face section of the fellow cornea (35.5 hoursafter death) at the same level as in (A). Notice the halos surrounding the epithelial cells(*). (C) Detail of (A) showing that the nerve fibers (arroius) have changed into strings ofvacuoles. (D) Enlargement of B in which epithelial edema is visualized as halos surroundingthe cells (*). The nerve fiber bundles (arrows) hardly can be distinguished among theepithelial cells. (E) Fixation within 1 hour after removal of the corneal button results inintact, straight running parallel organized nerve fiber bundles (arrows) in the central-peripheral cornea. (F) Enlargement of (E) shows connections (arrows) between the parallelrunning nerve fibers at the basis of the basal cells (*). BL = Bowman's layer; Ep = epithelium;N = nucleus. Bar = 0.2 mm (A,B), 0.1 mm (C,D), 0.2 mm (E), 0.05 mm (F).


Architecture of Human Corneal Nerves 987

& [im

FIGURE 2. (A) Reduced collage of electron micrographs and (B) a drawing made from thiscollage showing bifurcations of a nerve bundle containing four to six straight nerve fibers.On the drawing, the irregular outlines of the basal cells clearly are visible (arrows in A). Ep= epithelium; NFB = nerve fiber bundle. Bar = 8 jj,m (A,B).

neal nerve fiber bundles run temporally-nasally andbeaded fibers obliquely.32 Furthermore, there is noinformation regarding their size and postmortem de-generation in the human cornea. The current studyaims to determine the orientation, the size distribu-tion, and the postmortem changes of the nerve fibersin the subbasal nerve plexus.

METHODS

Transmission Electron Microscopy

For measuring the size of the nerve fibers, seven cor-neas derived from eyes of people with melanoma (sixapproximately 1 hour after enucleation and one less

than 15 minutes after enucleation) were used. Thedegree of anterior extension of the melanoma wasrather small, indicating hardly any or no effect on thecorneal nerves. Donor age varied between 49 and 87years. Postmortem effects as well as orientation of thenerve fiber bundles were studied by comparison oftwo corneas derived from eyes of people with mela-noma (one less than 1 hour after enucleation and oneless than 15 minutes after enucleation) and two pairsof postmortem corneas, of which one was processedimmediately. The fellow eye was left for 24 hours in aphosphate-buffered saline before fixation (first pair,72 years old, fixed after 13.5 versus 37.5 hours afterdeath and the second pair, 77 years old, fixed after16.5 versus 40.5 hours after death) was used.



Because the cornea was 12.6 mm across, 4 mm ofthe central third was used as the central, the sur-rounding 2 mm as the central-peripheral, and theremainder as the peripheral part. Samples of approxi-mately 4 mm2 were dissected. Because of the curvatureof the cornea, only 1 to 2 mm2 of each sample couldbe used for analysis in "en face" sections. After fixa-tion in 1.25% glutaraldehyde-1% paraformaldehydein 0.08 M cacodylate buffer for several days, the tissuewas postfixed in 1% OsO4 supplemented with 1.5%ferrocyanide in 0.1 M sodium cacodylate buffer. Theosmolality of the aldehyde fixatives was determinedand kept within the range of 475 to 500 mOsmol.After being dehydrated in a graded series of ethanol,the pieces were flat embedded in epoxy resin. Foren face sections, pieces of tissue were sawed off andmounted on epoxy resin blanks.

Postmortem changes and orientation of the nervefiber bundles were analyzed in serial, Toluidine bluestained, 1 /xm en face sections reaching from the ante-rior epithelium to the anterior stroma. At levels wherestructures of interest appeared in the light micro-scope, ultrathin sections (60 to 80 nm) were cut fortransmission electron microscopy. To measure the sizeof the nerve fibers, photographs of randomly selectedultrathin cross-sections were used. Ultrathin sectionswere stained with uranyl acetate and lead citrate andinspected in Philips EM 201 and CM 12 electron mi-croscopes (Philips Industries, Eindhoven, The Nether-lands). Methods for securing human tissue were hu-mane, and the tenets of the Declaration of Helsinkiwere followed.

Reconstruction

Nerve bundles in samples of the central, central-pe-ripheral, and peripheral part of five corneas were re-constructed. At low magnification (X100), outlines ofconsecutive semithin sections were drawn on transpar-ent paper using a drawing tube. At higher magnifica-tion (X260), nerve fibers were drawn and reduced tofit in the drawings obtained at low magnification. Theywere copied on transparent sheets and superimposed,thus reflecting the three-dimensional organization ofthe nerves. In addition, reconstructions of nerve fiberbundles at the ultrastructural level were made. Theoutlines of the profiles of the nerve fibers and thebasal cells were drawn on paper, overlaying collages ofphotographs taken at a final magnification of X7500.

QuantificationNerve fiber profiles present in cross-sections of thedifferent samples were photographed at a final magni-fication of X 31,000. Fifteen photographs of each spec-imen were made. With the Videoplan (Vidas Kontron,Munich, Germany), the minimal and maximal diame-ters of individual nerve fibers in the central (N = 714),

FIGURE 3. (A) A photograph derived from a marked andrapidly fixed fresh cornea shows that in the center, mainnerve fiber bundles (*) run in the 9-3 hours direction andthat intermediate nerve fiber bundles bifurcate and run inthe 12-6 hours direction. (B) Enlargement of the inset in(A) illustrates smaller side branches (single arrows) derivedfrom the intermediate nerve fiber bundles {double arrows),which run in the same direction as the main nerve fiberbundles. Bar = 0.1 mm (A), 0.05 mm (B).

central-peripheral (N = 640), and peripheral (N =126) cornea were determined. In addition, the diame-ter of the nerve fiber bundles (N = 113) was mea-sured. Nerve fibers, being considered as long tubes,will appear as circular profiles in cross-sections and asellipsoid profiles in oblique sections. The small diame-ter of the ellipsoid profiles approximates the real di-ameter of the nerve fibers.33 Therefore, the minimaldiameter was used as an estimate of the size of thenerve fibers.

Statistical Analysis

A one-way analysis of variance test34 was used to deter-mine differences in frequency distributions of thenerve fiber diameter. Differences were considered sig-nificant when P < 0.05.



to each other in the zone between Bowman's layerand the basis of the basal cell layer. It became clearthat nerve fiber bundles, sometimes forming loops,are interconnected (Figs. IE, IF). However, in lightmicroscopy, it is impossible to establish the exact posi-

FIGURE 4. (A) Low power electron micrograph of a nervefiber bundle in the subbasal plexus containing a bead atthe periphery (arrow). (B) Enlargement of the bead in (A)illustrating that the bead is located within the bundle ofstraight nerve fibers. It contains an accumulation of mito-chondria (M) and glycogen particles (arrow). All the nervefibers contain vesicles (arrowheads). Ep = epithelium. Bar =2.5 fj.m (A), 1 /j,m (B).

RESULTS

light Microscopy

Analysis of central, central-peripheral, and periph-eral en face sections showed that subbasal nerve bun-dles, closely aligned to Bowman's layer, show signs ofdegeneration within 13.5 hours after death, as shownin Figure 1A. At higher magnifications, these nervebundles have changed into strings of vacuoles (Fig.1C). In the fellow cornea (37.5 hours after death),degeneration was even more pronounced; at low mag-nifications, it was difficult to distinguish the nerve fi-ber bundles among the swollen epithelial cells, whichare surrounded by lucent halos (Fig. IB). However,at higher magnification, the vacuolized nerve bundleswere identifiable (Fig. ID). Postmortem alterations inthe epithelium were more pronounced in the secondpair of eyes (16.5 versus 40.5 hours after death, notillustrated).

In corneas fixed within less than 1 hour after enu-cleation, nerve fiber bundles proved to run parallel

BLFIGURE 5. (A) A cross-sectioned nerve bundle containingseven profiles with circular profiles. Notice the mitochon-dria (*) and glycogen particles (arrows). (B) A single oval-shaped nerve fiber profile in the subbasal plexus filled withmitochondria. (C) This cross-sectioned nerve bundle showsprofiles with varying diameters and shapes. Lines on theprofiles in (A,B) and (C) represent the minimal diameter.BL = Bowman's layer; Ep = epithelium. Bar: 1 fim (A,B,C).



. .

t1

peripheral Q

(

Ill| j |lllLi.li... ...

yperipheral p

. . . :

... . . . „....

FIGURE 6. Frequency distri-butions of the diameters (inmicrometers) of the nervefibers in the subbasal plexusof (A) central, (B) central-peripheral, and (C) periph-eral corneal samples. (D)The number of large pe-ripheral fibers between 1.3and 2.6 /um is shown sepa-rately. The majority of thefibers range from 0.2 to 0.5fj,m.

tion of the nerve plexus within the cornea. Recon-structions of electron micrographs (Fig. 2A) showedthat the outlines of the basal cells are irregular in formand that the nerve fiber bundles, consisting of four tosix nerve fibers, run between the basal cell layer andBowman's layer. A drawing of the reconstruction inFigure 2A illustrates the highly undulated membranes.of the basal cells (Fig. 2B).

Reconstructions of nerve fiber bundles in post-mortem and fresh tissue at the light microscopic levelshowed that the main bundles run in parallel. Despitethe presence of many degenerated fibers in postmor-.tern corneas, this pattern was not abolished. Part ofone of the reconstructions derived from a cornea ofwhich the position was marked in the eye is shown inFigure 3. This figure illustrates that the main largebundles run in the 9-3 hours direction (Fig. 3A). Fromthese nerve bundles, smaller intermediate bundles bi-furcate, which run at almost right angles in the 12-6hours direction. From the intermediate bundles,smaller bundles bifurcate and run again in the samedirection as do the large main bundles (9-3 hours)(Fig. 3B). Nerve fiber bundles consist of straight fibersand beaded fibers (Fig. 4A). The beads are located atthe periphery of the bundle and are characterized byan accumulation of mitochondria and glycogen parti-cles (Fig. 4B).

Quantification

Cross-sections through the central, central-periph-eral, and peripheral cornea show nerve fiber bundles

(Fig. 5A) and large single nerve fibers (Fig. 5B) at thebasal aspect of the basal cells. When a nerve bundleis cut at almost right angles, it contains more or lesscircular; profiles as illustrated in Figure 5A. However,when the bundle is hit obliquely, the nerve fiber pro-files appear as ovally shaped structures. In general,microtubules dominate in the small fibers and neuro-filaments in the large fibers (Fig. 5C). Numerous mito-chondria and clusters of glycogen particles are presentin large fibers (Figs. 5A and 5B).

The distributions of nerve fiber diameters areshown in Figure 6 for the central (Fig. 6A), the cen-tral-peripheral (Fig. 6B), and the peripheral part(Figs. 6C, 6D) of the cornea. The distributions areskewed at the right side. In the periphery, the tailappeared to be rather long, and therefore it is pre-sented in a separate graph (Fig. 6D). Nerve fibers varyin size between 0.05 and 2.5 fim, and the majority ofthe fibers is between 0.1 and 0.5 fim. The average sizein the center is 0.36 /L/m ± 0.18 (mean ± standarddeviation), in the center-periphery 0.34 /im ± 0.2,and in the periphery 0.67 fim ± 0.64. Equal numbersof photographs showed that the number of nerve fiberprofiles encountered in die center (N = 714) and inthe center-periphery (N = 640) is approximately fiveto six times higher than the number encountered inthe periphery (N = 126). No significant differenceswere observed between the distributions in the centerand in the center-periphery (P< 0.01), whereas bothdistributions differed significantly from the distribu-tion in the periphery (P < 0.01). This mainly was



because approximately one fourth (32 fibers) of theperipheral fibers are larger than 1.3 /xm (Fig. 6D).The distributions of the nerve bundle diameters alsowere skewed and had an average size of 1.0 ± 0.3 /xm(data not shown).

DISCUSSION

Architecture of Human Corneal Nerves

The current study shows that 13.5 hours after death,a great part of the human corneal nerves has degener-ated or gone. As analysis of the nerve fiber distributionin the subbasal plexus was carried out with corneasobtained from persons with melanoma, the questionarises whether this distribution reflects the real archi-tecture of human corneas. It has been stated that cor-neal nerves become more prominent in corneas withan intrinsic corneal disease. Examples are Fuch's dys-trophy, bullous keratopathy, keratoconus, posteriorpolymorphous dystrophy, and herpes zoster-herpessimplex keratitis.35 For corneas obtained from patientswith melanoma, such a prominent increase has notbeen reported. Analysis of these corneas showed thatthe parallel organization of the main nerve bundleswas similar to that observed in in vivo confocal micros-copy.32 Therefore, the conclusion seems justified thatthe pattern of the nerve tree in corneas of patientswith melanoma reflects the real architecture of humancorneas.

Light and electron microscopic observations gavea precise description of the penetration of nerve fiberbundles into the corneal domain from the limbal area.Because of quantitative measurements, it became clearthat the branching pattern in rabbits, as described byRozsa and Beuerman,14 is different from that in hu-man. In rabbit, the number of intersections decreasesgradually from the center toward the limbus, whereasthe current study does not provide evidence for sucha gradual decrease in the number of nerve fibers. Thisnumber, given as an arbitrary measure of profiles per15 photographs, is five to six times less in the periph-ery than in the center and center-periphery. Compar-ison of our data with those of the rabbit suggests thatthe apex of the rabbit cornea is innervated heavily,whereas the central two thirds of the human corneaequally is dense innervated. These data have led tothe following architecture of the corneal subbasalplexus in humans. Radially oriented thick stromalnerve bundles, running parallel to the corneal surface,enter the corneal domain from the limbus. Beforepenetration of Bowman's layer, stromal nerve bundlesbend 90°. Subsequently, the nerve fiber bundles bendagain 90° and run between the basal cell layer andBowman's layer in a 9-3 hours direction. From thesebundles, a rather regular meshwork of finer branches

FIGURE 7. Schematic drawing of the distribution of nervefiber bundles across the central and central-peripheral cor-nea. The large entering nerve fiber bundles run in the 9-3hours direction. After the first bifurcations, they run in the12-6 hours and after the second bifurcations, they run againin the 9-3 hours direction. The interwoven beaded fibersrun singly and obliquely after branching.

is formed, as illustrated in Figure 7. As described in aprevious article, in cross-sections, alternating nervefiber bundles and single fibers are located at more orless equal distances.9 The current study, based on enface sections, supports the previous observations andshows connections of main nerve bundles withobliquely running single-beaded fibers. A three-di-mensional artist's impression of the distribution of thenerve fibers is given in Figure 8. The large nerve fiberbundles entering the subbasal plexus consist ofstraight fibers and beaded fibers. The straight nervefiber bundles splice and also bifurcate several timesto become gradually smaller nerve bundles. From themain large bundles, single-beaded fibers bifurcate andrun obliquely. At certain locations of the meshwork,these beaded fibers turn 90° in an upward direction(Fig. 8). This conclusion is based on observations9 thatlarge profiles (widi many mitochondria), approxi-mately two to four times larger than those of straightnerve fibers, were present above the basal cell layerin en face sections. The reason for such a specificorganization is not known, but one can imagine that inthis way, a homogeneous distribution of nerve endingsover the central and central-peripheral cornea isguaranteed.

To provide us with the resolution of this mesh-work, it is important to know the distances betweenthe intermediate and the smaller nerve fiber bundles.



Epithelium

Bowmans layer

Stroma

FIGURE 8. Three-dimensional drawing of the penetration and distribution of stromal bundlesinto the subbasal plexus. The unmyelinated nerve fiber bundles (blue) are sometimes splic-ing, have bifurcations at almost right angles, and consist of several straight (red) and single-beaded (green) fibers. Single-beaded fibers bifurcate obliquely and turn upward betweenthe basal cells to reach the wing cells.

However, measuring distances in a significant samplerequires many more fresh human corneas and is atedious, almost impossible, task. In vivo confocal mi-croscopy could give a rapid impression, but the resolu-tion of images in whole-mount tissue in comparisonto those in thin sections still is poor, so that intermedi-ate and small nerve fiber bundles will not be observedeasily. What can the meshwork tell us about the localsensitivity of pain perception? The density in the cor-nea is known to be rather high in comparison to toothpulp (20 to 40 times) and extremely high in compari-son to the skin (300 to 600 times).14 The distancebetween the first-order branches is in the order ofmagnitude of 20 to 50 /im, and between the secondand third branches, it certainly will be smaller, indicat-ing that injuries to individual epithelial cells may besufficient to give a pain perception.

The parallel organization of the nerve fiber bun-dles described here seems to be specific for the humancornea because it also was observed in formaldehydesolution-fixed tissue stained with hematoxylin,31 ingold chloride-impregnated tissue,8 and in in vivo con-focal microscopy.32 In gold chloride-impregnated epi-thelial tissue, Schimmelpfennig8 has described fusionof nerve fibers. A fusion in Schimmelpfennig's study

is described as a branching point in in vivo confocalmicroscopy.32 These fusions or branching points alsowere observed by Martinez in formaldehyde-fixed tis-sue.31 It is tempting to speculate on fusion of nervefiber bundles, because single fibers fuse with smallbundles to become gradually thicker bundles. Becauseseveral figures in the forementioned studies and inthe current study show that thick bundles divide intotwo smaller ones of the same size and then fuse againto become a bundle of the original thickness (Fig.IF), it seems rather unlikely to describe the branchingpoints as fusions. The bundles most likely splice toincrease the surface area of the sensory receptors. Inaddition, in this respect, most light microscopic de-scriptions of nerve fibers28'11'29'31 are nerve fiber bun-dles when studied with the electron microscope.9 Onlyat the location of the beads are single fibers clearlyvisible in light microscopy.11'32

A quantitative analysis as performed in the currentstudy has not been carried out previously. In mice, anattempt has been made to measure the size of theindividual fibers, but because of variations in size, itwas difficult to draw conclusions. Despite the greatvariation and poor quality of die tissue, it seems thatin mice, most of the nerve fibers measured are 0.4 //m



or less.36 This size is in the range of the physiologicallyclassified C-fibers (0.2 to 2 fJ,m) and not in the rangeof the A<5 nerve fibers (1 to 5 fxm) in the skin. Thecurrent study suggests that the majority of the nervefibers in the human cornea also are C-fibers. In theperiphery, the size of the nerve fibers does not allowa discrimination between C and A<5 fibers becausemany are equal to or larger than 1 t̂m across, whichmeans that they can belong to both classes. In addi-tion, single-beaded fibers can not be classified simplybecause they have the same size as that of large straightfibers at the location of the beads. Although no quanti-tative data on the size distribution of the stromal nervefibers are available, both small (0.4 fim) and largefibers (1 fj,m), as observed in the subbasal plexus, alsoare present in the corneal stroma.9 When the size ofthe nerve fiber bundles is taken into account, the num-ber of fibers larger than 1 [im (32+ 113 fim) is fourtimes higher. This figure may be in favor of the A6fibers, but still this number is not enough to explainthe distribution of type I (thermal and chemical) andtype II (mechanoreceptive) endings as described byMaclver and Tanelian.11 The skewed frequency distri-butions of the minimal diameters with a tail at theside of the larger diameters indicate two populationsof fibers. However, the tail is too small to distillatetwo overlapping distributions. The small tail might berelated to the relatively low number of beads hit in thesections. Therefore, it is concluded that the skeweddistribution in the center and center-periphery prob-ably is caused most by beads and in the periphery bylarge fibers.

Density of Nerve Fibers

The density of nerve fibers in human corneas afterpeptidergic and classical neurotransmitter stainings asfound in light microscopic studies15'18'20'28'37 is difficultto compare with the density of nerve fibers in thecurrent study and that of Schimmelpfennig.8 Analysisof rapidly fixed fresh corneal tissue and of gold chlo-ride-treated corneal tissue shows larger numbers ofcorneal nerves. Differences in fiber density in humancorneal tissue only can be ascribed partly to the speci-ficity of the antibodies or to penetration of the anti-bodies used in immunocytochemical stainings. An ad-ditional factor influencing the density of immuno-stained nerve fibers might be the quality of the tissue.It is known that postmortem eyes loose their epithe-lium easily after death (Cornea Bank Amsterdam, per-sonal communication, 1996). In addition, this crite-rion does not explain the differences in density be-cause Uusitalo et al28 have used corneas of enucleatedeyes and Ueda et al15 have used postmortem corneas.Uusitalo et al28 found hardly any staining, whereasUeda et al15 found between 5 and 9 hours postmor-tem, dense staining with nonspecific markers for neu-

ral elements. These observations are in line with ourfindings on tissue from enucleated eyes, but in con-trast with our findings on tissue of postmortem eyes,which show that 13.5 hours after death, nerve fiberbundles have become strings of vacuoles. It is, ofcourse, possible that nerve fibers degenerate mainlyafter 9 hours. A verification is rather difficult becausepostmortem corneas of less than 9 hours are rare andwill almost always be used for transplantation pur-poses. Furthermore, Tervo et al20 and Toivanen18 haveused corneas from eyes of people with keratoconus.Keratoconus corneas are easier obtainable but farfrom healthy, and their nerves are more pronouncedthan in those of control corneas.35 Nevertheless, whenhuman corneas are compared with those of othermammalians, it is obvious that, for example in rats,many more nerve fibers are stained with calcitoningene-related peptide29 and with cholinesterase.38

These data suggest that corneas of rat are innervatedmore heavily than are corneas of humans. This cer-tainly is not the case, because the number of nervefiber profiles in cross-sections of rats, rabbits, andmonkeys perfused transcardially is lower than that ob-served in humans in electron microscopy (unpub-lished observations, 1995). Thus, it is not possible tocompare the density of nerve fibers in human tissueafter serial section analysis with immunocytochemicalstaining techniques.

In conclusion, the nerve bundles in the subbasalplexus of the human cornea form a regular densemeshwork with equal density over a large central andcentral-peripheral area. Because of their size, the ma-jority of the fibers can be classified as C-fibers.

KeyWords

epithelium, human corneal nerve fibers, quantification, sub-basal plexus, ultrastructure

Acknowledgments

The authors thank Drs. Jan Klooster and Jan Ruijter fordiscussions; Niko Bakker, Ton Put, and Marina Danzmanfor the drawings and photographs; Drs. M. Kliffen and C. M.Mooy at the Department of Ophthalmology, Erasmus Uni-versity, Rotterdam; and L. Koornneef at the AcademicalMedical Center, Amsterdam, for their efforts in obtainingfresh corneal tissue derived from eyes of persons with mela-noma.

References

1. Schornstein T. Ein beitrag zur hornhautinnervationdes menschlichen auges. Arch fur Augenheilkunde.1934; 108:601-609.

2. Zander E, Weddell G. Observation on the innervationof the cornea. J Anat. 85:68-99.

3. Sakamoto K, Histological study on the innervation ofthe human cornea. Tohoku J Exp Medicine. 1951;54:105-114.



4. Ruskell GL. Ocular fibres of the maxillary nerve inmonkeys. JAnat. 1974:118:195-203.

5. Ten Tusscher MPM, Klooster J, Vrensen GFJM. Theinnervation of the rabbit's anterior eye segment: Aretrograde tracing study. Exp Eye Res. 1988;46:717-730.

6. Ten Tusscher MPM, Klooster J, Van der Want JJL, etal. The allocation of nerve fibers to the anterior eyesegment and peripheral ganglia of rats. I. The sensoryinnervation. Brain Res. 1989;494:95-104.

7. Beuerman RW, Schimmelpfennig B. Sensory denerva-tion of the rabbit cornea affects epithelial properties.ExpNeurol 1980;69:196-201.

8. Schimmelpfennig B. Nerve structures in human cen-tral corneal epithelium. Graefes Arch Clin Exp Ophthal-mol. 1982:218:14-20.

9. Muller LJ, Pels E, Vrensen GFJM. Ultrastructural or-ganisation of human corneal nerves. Invest OphthalmolVisSci. 1996; 37:476-488.

10. Belmonte C, Gallar J, Pozo MA, Rebello I. Excitationby irritant chemical substances of sensory afferentunits in the cat's cornea. JPhysiol. 1991;437:709-725.

11. Maclver MB, Tanelian DL. Free nerve ending terminalmorphology is fiber type specific for AS and G-fibersinnervating rabbit corneal epithelium. / Neuraphysiol.1993;69:1779-1783.

12. Tanelian DL, Monroe S. Altered thermal respon-siveness during regeneration of corneal cold fibers. /Neurophysiol. 1995; 73:1568-1573.

13. Giraldez F, Geijo E, Belmonte C. Response character-istics of corneal sensory fibers to mechanical and ther-mal stimulation. Brain Res. 1979; 177:571-576.

14. Rozsa AJ, Beuerman RW. Density and organisation offree nerve endings in the corneal epithelium of therabbit. Pain. 1982;14:105-120.

15. Ueda S, del Cerro M, LoCascio JA, AquavellaJV. Pep-tidergic and catecholaminergic fibers in the humancorneal epithelium. Ada Ophthalmol. 1989; 67:80-89.

16. Tervo K, Latvala TM, Tervo TMT. Recovery of cornealinnervation following photorefractive keratoablation.Arch Ophthalmol. 1994; 112:1466-1470.

17. Tervo T, Palkama A. Ultrastructure of die cornealnerves after fixation with potassium permanganate.AnatRec. 1978; 190:851-860.

18. Toivanen M, Tervo T, Partanen M, et al. Histochemi-cal demonstration of adrenergic nerves in the stromaof human cornea. Invest Ophthalmol Vis Sci. 1987;28:398-400.

19. Katz DM, Markey KA, Goldstein M, et al. Expression ofcatecholaminergic characteristics by primary sensoryneurons in the normal adult rat in vivo. Proc NatlAcadSci USA. 1983;80:3526-3530.

20. Tervo K, Tervo T, Eranko L, et al. Substance P-immu-noreactive nerves in the human cornea and iris. InvestOphthalmol Vis Sci. 1982;23:671-674.

21. Sasaoka A, Ishimoto I, Kuwayama Y, et al. Overall dis-tribution of substance P nerves in the rat cornea andtheir three-dimensional profiles. Invest Ophthalmol VisSci. 1984;25:351-356.

22. Lethosalo JI. Substance P-like immunoreactive trigem-inal ganglion cells supplying the cornea. Histochem J.1984; 80:273-276.

23. Stone RA, Kuwayama Y. Substance P-like immunoreac-tive nerves in the human eye. Arch Ophthalmol.1985;103:1207-1211.

24. Stone RA, McGlinn, AM. Calcitonin gene-related pep-tide immunoreactive nerves in human rhesus monkeyeyes. Invest Ophthalmol Vis Sci. 1988; 329:305-310.

25. Kuwayama Y, Stone RA. Distinct substance P and calci-tonin gene-related peptide immunoreactive nerves inthe guinea pig eye. Invest Ophthalmol Vis Sci.1987;28:1947-1954.

26. Bee, JA, Kuhl U, Edgar D, et al. Avian corneal nerves:Co-distribution with collagen type IV and acquisitionof substance P immunoreactivity. Invest Ophthalmol VisSci. 1988;29:101-107.

27. Beckers HJM, Klooster J, Vrensen GFJM, et al. Sub-stance P in rat corneal and iridal nerves: An ultrastruc-tural immunohistochemical study. Ophthalmic Res.1993;25:192-200.

28. Uusitalo H, Krootila K, Palkama A. Calcitonin gene-related peptide (CGRP) immunoreactive sensorynerves in the human and guinea pig uvea and cornea.Exp Eye Res. 1989;48:467-475.

29. Jones MA, Marfurt CF. Calcitonin gene-related pep-tide and corneal innervation: A developmental studyin the ra t . / Comp Neurol. 1991;313:132-150.

30. Beckers HJM, Klooster J, Vrensen GFJM, et al. Ultra-structural identification of trigeminal nerve endingsin the rat cornea and iris. Invest Ophthalmol Vis Sci.1992; 33:1979-1986.

31. Martinez R. Etude sur l'innervation de la cornee hu-maine. Invest Biol. 1940;32:75-109.

32. Auran JD, Koester CJ, Kleiman NJ, et al. Scanning slitconfocal microscopic observation of cell morphologyand movement within the normal human anterior cor-nea. Ophthalmology. 1995; 102:33-41.

33. Weibel ER. In: Stereological Methods: Practical Methods forBiological Morphometry. London: Academic Press; 1979.

34. Conover WJ. Statistics of the Kolmogorov-Smirnovtype. In: Shewhart A, Wilks SS, eds. Practical Non-Para-metric Statistics. New York: John Wiley & Sons;1980:344-384.

35. Marfurt CF, Ellis LC. Immunohistochemical localiza-tion of tyrosine hydroxylase in corneal nerves. / CompNeurol. 1993;336:517-531.

36. Dennehy PJ, Feldman GL, Kambouris M, et al. Rela-tionship of familial prominent corneal nerves and le-sions of the tongue resembling neuromas to multipleendocrine neoplasia Type 2B. Am J Ophthalmol. 1995;120:456-461.

37. Tervo T. Histochemical demonstration of cholinester-ase activity in the cornea of the rat and the effect ofvarious denervations on the corneal nerves. Histochem-istry. 1976;47:133-143.

38. Whitear M. An electron microscope study of the cor-nea in mice witfi special reference to the innervation.JAnat. 1960;94:387-409.


Architecture of human corneal nerves. - ARVO...

Documents

Transcript of Architecture of human corneal nerves. - ARVO...