J. Anat. (1991), 177, pp. 75-84
With 13 figures
Printed in Great Britain
The structure of corpuscular nerve endings in the limbal
conjunctiva of the human eye
JOHN G. LAWRENSON AND GORDON L. RUSKELL
Vision Research Centre, Department of Optometry and Visual Science, Dame Alice
Owen Building, 311-321 Goswell Road, London ECl V 7DD
(Accepted 14 January 1991)
Corpuscular nerve endings were first described in the human conjunctiva by Krause
in 1859. These subepithelial structures, which he termed Endkolben, were observed at
the termination of myelinated nerve fibres throughout the conjunctiva. The Krause
corpuscle or endbulb has been confirmed subsequently in the human conjunctiva by
many authors, employing improved techniques, although descriptions differed with
regard to their relative incidence and precise morphology. The Krause corpuscle was
typically described as a round or oval encapsulated structure, served by one to three
myelinated fibres that lose their myelin sheath before entering the body of the
corpuscle. The fibres then form a tightly coiled network before terminating. Most
workers were able to observe a capsule, consisting of a thin connective tissue layer
completely investing the structure.
A significant development came with the introduction of methylene blue as a means
of staining nerve fibres intravitally (Ehrlich, 1886). Knusel and Vonwiller (1922) used
this stain on the living conjunctiva of the human eye and demonstrated corpuscular
nerve endings varying in size and distribution. Generally, the greatest number of
stained endings were seen in areas ofbulbar conjunctiva covered by the upper lid, their
incidence decreasing in the interpalpebral zone, with fewest in areas covered by the
Oppenheimer, Palmer & Weddell (1958) found that the majority of conjunctival
nerves terminated as free nerve endings, often in relation to blood vessels but,
particularly in the human eye, a number of more complex terminals were seen. These
displayed a variety of morphological forms, many resembling the descriptions of
earlier studies, but in addition they saw a number of structures which they considered
'intermediate' in form. This diversity, together with their irregular distribution, led
Oppenheimer and his coworkers to conclude that, rather than representing specialised
sensory terminals, conjunctival 'endbulbs' correspond to stages in the natural cycle of
growth and decay of certain peripheral nerve fibres. A reported increase in their
number with age is consistent with this conclusion (Riisager, 1962).
The stimulus for the present study was the chance finding ofcomplex sensory nerve
endings in concentrated numbers in the limbal conjunctiva of the human eye. They
appeared to show a close association with the limbal palisades of Vogt, a series of
connective tissue ridges in the conjunctiva which form a radial pattern adjacent to the
cornea (Vogt, 1921).
Earlier workers were aware of the presence ofencapsulated or complex terminals at
the human limbus (Lightbody, 1867; Key & Retzius, 1876; Dogiel, 1891). Knusel &
J. G. LAWRENSON AND G. L. RUSKELL
Vonwiller (1922) noted that their intravitally stained terminals were smaller and more
numerous at the limbus. The histological appearance of complex limbal terminals led
Wolter (1964) to identify them as Krause corpuscles, which were noted to have an
accessory nerve supply of possible autonomic origin.
In this report we describe the histology ofcorpuscular nerve endings with the limbal
conjunctiva, and give the first account of their ultrastructure.
MATERIALS AND METHODS
The material used in this study consisted of the anterior halves of cadaver eyes
following the removal of the corneas for penetrating keratoplasty. Each specimen was
cut into a series of approximately equal radial segments, from which most of the
cornea and uvea was trimmed away. All segments from 2 eyes, and two or more
segments from 6 further eyes were studied.
Tissues were prepared for light and electron microscopy by immersion in 2%
glutaraldehyde and 3% paraformaldehyde buffered to pH 7-4 with sodium cacodylate.
Segments were then postfixed in 1% osmium tetroxide, dehydrated through graded
ethanols, and cleared in xylene, before embedding in Araldite. Semithin sections were
cut from each segment, transversely to the palisades, beginning at the cornea. These
included full serial and interrupted serial sections taken at intervals ranging from 10
to 30 ,um. Sections were mounted on glass slides and stained with 1% toluidine blue
in 2-5% sodium carbonate.
At intervals in each series, ultrathin sections were cut, mounted on copper grids, and
stained with a saturated solution of uranyl acetate in 70% ethanol, followed by 0 4%
lead citrate in 01 M sodium hydroxide.
A whole mount gold chloride impregnation method was used on some segments.
These were placed in fresh filtered lemon juice for 15 min without prior fixation,
washed in distilled water, and transferred to a 1% solution of gold chloride at room
temperature for 20 min. The tissues were then placed in a solution containing 50 ml of
distilled water, to which 5 drops of glacial acetic acid had been added, and left for
14-18 h. Because of the density of staining, in order to visualise conjunctival nerve
fibres and their terminals it was necessary to remove the majority of the underlying
sclera and conjunctival epithelium from each segment. Finally the tissues were fixed in
10% formal saline, dehydrated with alcohol and cleared in xylene before mounting on
slides for examination.
Fig. 1. A pigmented example of the palisades of Vogt at the conjunctival limbus. The crests of the
palisades are represented by the clear interval (arrow) between each pair ofpigment lines. The wider
clear intervals separating pairs of pigment lines indicate the interpalisade zones. c, cornea. Bar,
Fig. 2. Transverse section ofthe palisades cut parallel to the corneoconjunctival junction. Each ofthe
3 palisades shown contains corpuscular nerve endings (arrowed). Bar, 50/sm.
Fig. 3. Corpuscular nerve ending located towards the apex ofa palisade, close to the basal epithelial
layer. The serving myelinated nerve fibre (arrow) is sectioned obliquely as it ascends the palisade. Bar,
Fig. 4. A limbal corpuscle showing detail within the body of the structure. Densely stained round or
oval neural profiles are surrounded by the lightly stained cytoplasm of Schwann cells. A delicate
connective tissue capsule (arrow) is visible around the perimeter of the corpuscle. Pigment is present
within the basal cells lining the slopes of the palisade (asterisks). Bar, 10,m.
Fig. 5. Two corpuscular nerve endings underlying the epithelium within an area oflimbal conjunctiva
lacking palisades. Bar, 10 Fm.
Fig. 6. Whole mount gold chloride preparation showing the afferent nerve fibre (n) and the capsule
(c) which appears complete. Bar, 10 um.
Conjunctival nerve endings
Figs 1-6. For legend see opposite.
J. G. LAWRENSON AND G. L. RUSKELL
Compared with our experience of samples of sectioned bulbar conjunctiva, the
limbal conjunctiva was noted to have a higher concentration of corpuscular nerve
endings. Most of them lay in a narrow, 1-00 mm wide, annular zone of conjunctiva
beginning approximately 0 5 mm from the corneoscleral margin, which was taken as
the termination of Bowman's layer. Beyond this zone corpuscles were less frequently
encountered. In radial segments usually at least 1, but as many as 16 were observed,
and based on eyes examined so far we estimate a mean incidence of 2-3 limbal
corpuscles per mm of circumference.
Sectioning the conjunctival palisades transversely (Fig. 1) revealed a characteristic
series of stromal elevations projecting into the overlying epithelium (Fig.
Corpuscular nerve endings were often located within the elevations, at which points
the epithelium was often reduced to a thickness of only 2-3 cells (Figs 2-4). Others
were found between palisades and in areas of limbal conjunctiva lacking palisades
Corpuscular nerve endings were round to oval in shape, and variable in size, with
a mean maximum diameter of 30 ,um (range 20-60 ,im). They were most often served
by a single myelinated nerve fibre, approximately 4,um in diameter, derived from
nerve fibre bundles deeper in the conjunctiva or episclera (Fig. 6). In specimens stained
for whole mount observation, additional fine calibre axons were occasionally seen
which appeared to terminate close to the corpuscle. The whole structure was
surrounded by a capsule which demarcated the nerve ending from the surrounding
connective tissue. The capsule typically consisted of a single layer of thin elongated
cells which joined to invest the corpuscle. In semithin sections the capsule frequently
failed to make a complete investment (Fig. 4), whereas in whole mount gold chloride
preparations the capsule usually appeared complete (Fig. 6).
Corpuscular nerve endings showed some variability in their internal organisation.
The body of the structure contained nerve fibres surrounded by non-neural accessory
cells which were later shown to have the characteristics of Schwann cells, and will be
referred to by this name. The large oval nuclei of these cells varied in number and
location and were often situated peripherally (Fig. 4). In the majority of corpuscles
neural elements showed up as pale circular profiles, whereas in others, similarly shaped
profiles, evidently neural, were darkly stained.
The density of the Schwann cells showed some variation; although often tightly
packed, some larger corpuscles had a looser arrangement, and sometimes connective
tissue channels partitioned the endings into two or more lobes.
At the ultrastructural level, features of limbal corpuscular nerve endings common
to similar endings found elsewhere in the body were recognised (Fig. 7).
The afferent myelinated axon of the limbal terminal (Fig. 8), having lost its
Fig. 7. Electron micrograph of a limbal corpuscular nerve ending. The interior of the corpuscle
comprises predominantly axon terminals (t) with Schwann-like accessory cells (s). Fibrocytes (I) with
thin processes are present which form septa within the corpuscle. A thin fibrocyte capsule (c)
demarcates the periphery of the structure. Bar, 5 4um.
Fig. 8. Myelinated nerve fibre (n) at the periphery of the corpuscle. c, capsular fibrocyte. Bar, I ,um.
Fig. 9. Branching of an axon terminal within a corpuscle. Bar, 1 ,um.
Conjunctival nerve endings
Figs 7-9. For legend see opposite.
J. G. LAWRENSON AND G. L. RUSKELL
Figs 10-13. For legend seeopposite.
perineurial sheath, shed itsmyelinsheathimmediatelybefore entering the corpuscle,
and then divided several times(Fig. 9) togiverise to a variable number of axon
terminals. Axon terminals consisted of a series of varicosities which were pre-
dominantlysectionedtransversely, presentinground or oval profiles in cross-section.
The varicosities contained neurofilaments, scattered microtubules,
aggregatesofmitochondria;the latter accounted for the increase in local volume (Fig.
10).The inter-varicoseregionof the nervefibres,incontrast, contained neurofilaments
and microtubules withsparsemitochondria (Fig.11). A particular morphological
variantdisplayedaxon terminals in which theaxoplasmcontained an electron-dense
material which hindered organelle visibility; however, within varicosities mito-
chondrialprofilescouldusuallybe made out, apparentlyin good order (Fig. 12).
Smallunmyelinatednerve fibres wereoccasionally seen outside the capsule, and
also near the base of thecorpusclewhere they were enclosed by the perineurium
of theserving myelinated axon, but we were unable to follow these fibres to their
Schwann cellcytoplasmcontained microfilaments with microtubules and sparse
mitochondria(Fig. 13).The cell border was characterised by numerous pinocytotic
vesicles which were visible as membraneinvaginationsor as cytoplasmic enclosures.
A basal lamina waspresentwhichdistinguished Schwann cells from neighbouring
connective tissue orcapsularcells.Awayfrom the main body of the cell the cytoplasm
was drawn into a number of thinprocessesorlamellae, and 2 or more layers of these
lamellae invested bothpreterminaland terminal axons (Fig. 7). The innermost layer
lacked a basal lamina at the interface with its axon, with a narrow space of
membrane structures wereoccasionallyseen between axons and lamellae, consisting
ofpaireddense membraneplaqueswith an interposed lighter material within the
Extracellular material consisted of a fine filamentous substance, similar to that
found in basal laminae, which wasinterspersed with collagen fibres.
The delicatecapsulewasfrequently incompleteand comprised fibroblast-like cells.
These cellsgaverise to fine extendedprocesseswhich enclosed the greater part of the
perimeterof thecorpuscle. Apartfrom limited overlapping only a single layer was
usually present, although3 or 4 wereoccasionallyseen. It was not uncommon for
fibroblastprocessesto extend into the main body of the structure. Particularly in
larger corpuscles, fibroblasts, which were located entirely within the body of the
corpuscleandappearednot to contribute to the capsule, sent out processes which
acted assepta, partitioningthe interior of the structure (Fig. 7). Both capsular and
non-capsularfibroblasts lacked a basal lamina, and the cytoplasm of these cells
contained pinocytotic vesicles, as well as rough endoplasmic reticulum and free
20 nm separating
the plasma membranes
Fig. 10. Detail of an axon terminal lying within an invagination of a Schwann cell. Accumulated
small mitochondria, neurofilaments, microtubules and a few agranular vesicles are present in the
axoplasm. Bar, 0-5,um.
apparent; mitochondria are largelyconfined to the varicose regions (between arrows) with the
intervaricoseregion containing predominantlyneurofilaments. Bar, 1,um.
Fig. 12. Two axon terminals whichdisplay darklystained axoplasm. Bar, 1 1um.
Fig. 13. A Schwann-likeaccessory cell showing numerous processes, each with a basal lamina and
pinocytotic vesicles. Bar, 05,m.
11. Longitudinal section through a terminal axon. The varicose nature of the terminal is
J. G. LAWRENSON AND G. L. RUSKELL
To overcome theproblemsassociated with an eponymous classification of a large
number ofcutaneoussensory corpuscles,based on their light microscopic appearance,
severalattemptshave been made to substitute a reclassification according to their
ultrastructure(Halata, 1975; Chouchkov, 1978; Malinovsky, 1988). Opinion differs
on the relativeimportanceof aperineurial capsuleas a distinguishing feature, whereas
thesignificanceof therelationshipbetween Schwann or accessory cell lamellae and
terminal axons isgenerally acknowledged. Corpusclescan thus be divided into those
where lamellae arearrangedin anasymmetricalmanner (the Meissner corpuscle fits
into thiscategory),and those wheretheyform one or more'inner cores' consisting
of a symmetric, concentric arrangement of lamellae (for example, the pacinian
The term 'Krause endbulb orcorpuscle', following its first description in the
conjunctiva, has been used to describe a number of morphologically different
encapsulatednerveendingswithin skin and various mucosa. Classically they have
been divided into two distinct groups; cylindrical (simple) endbulbs, which are
consideredtypicaloflowermammals,and the more complex spherical endbulbs found
in primates. Ultrastructural studies have been carried out on both types. The
cylindricalendbulb of Krause has been found in the lip, tongue and nasal mucosa of
the cat(Spassova, 1965, 1973, 1974).It consists of a single straight terminal neurite,
surrounded by concentrically arranged cytoplasmic lamellae, resembling a small
pacinian corpuscle. Spherical endbulbs have been described in primate mucosa,
namelythe oralcavity,rectum(Chouchkov, 1973)and vocal cord (Nagai, 1982). In
contrast, thesecorpusclesconsist of a coiled terminal neurite. Each axon terminal is
enveloped by Schwann cell lamellae which lack a symmetrical pattern.
Followingthehistologicalstudies of the late nineteenth century, corpuscular nerve
endingsfound in the humanconjunctivawereregardedas similar to those found in the
genitalia,termedgenital corpuscles (Krause, 1881; Suchard, 1884). The fine structure
of these terminals has been described inman, for example in the clitoris (Polacek &
Malinovsky, 1971; Malinovsky, 1979),and theglans penis (MacDonald & Schmidt,
1979; Halata &Munger, 1986), confirmingtheir identity as Krause corpuscles.
Thesensory corpuscleswhich we have observed in the limbal conjunctiva similarly
displaythe form of thesphericalKrausetypeas defined in primate skin and mucosa.
Inparticular, they possessanasymmetrical patternof cytoplasmic lamellae derived
fromcomparableSchwann-likeaccessorycells. The axon terminals themselves, with
their accumulation of small mitochondria, are typical of those found in sensory
corpuscles. However, the varicose nature of the axon terminals present in limbal
corpuscleshas not beenemphasisedinequivalentstructures although, as described
earlier, Dogiel (1891)illustrated varicose terminalprofilesamong encapsulated nerve
endings at the limbus.
The finding of axon terminals containing an electron-dense axoplasm was
unexpected,and inconsistent withdescriptionsof other sensory corpuscles. It must be
emphasisedthat the availableeye-bankmaterial was subject to a variable amount of
delay priortofixation, and thismayhave influenced terminal morphology.
Encapsulationofmorphologicallysimilar sensory corpuscles is variable. Halata &
Munger (1986)found that thedegreeofencapsulation of genital corpuscles varied
withposition in the dermis, such that deeper-lying corpuscles possessed a more
pronounced capsule.The thincapsuleof thesuperficiallylocated limbal corpuscles is
consistent with this observation.
Firm evidence of a secondary nerve supply as described originally by Lightbody
(1867) and subsequently by Wolter (1964) was not obtained in this study. Although
fine calibre axons appeared occasionally to terminate close to an encapsulated nerve
ending in whole mount preparations, it was not possible to decide with the electron
microscope whether or not these unmyelinated nerve fibres entered the corpuscle.
The function ofcorpuscular nerve endings in the human conjunctiva generally, and
more specifically at the limbus, is open to question. Krause (1881) considered
conjunctival endbulbs to be touch receptors, and this view persisted until von Frey
(1906) introduced his popular punctate theory of cutaneous sensibility. According to
this theory each sensory modality is subserved by a particular morphologically distinct
receptor. Von Frey proposed that the Krause corpuscle was responsible for the
perception of cold, and further evidence came from Strughold & Karbe (1925), who
claimed to show a correlation between the position of Krause corpuscles within the
conjunctiva and points of enhanced cold sense. The existence of such specificity was
questioned by Oppenheimer et al. (1958), whose morphological observations led them
to conclude that the complex nerve endings within the conjunctiva were the result of
neural degeneration and regeneration.
At the ultrastructural level limbal conjunctival corpuscles exhibit a high degree of
structural complexity, sharing many features common to sensory corpuscles found in
other tissues. Physiological investigations of corpuscles found within glabrous skin
have suggested that they are rapidly adapting mechanoreceptors, based on their
response to perpendicular ramp indentation (Iggo & Ogawa, 1977; Johannson &
Valbo, 1983). If corpuscular nerve endings within the conjunctiva act as extero-
receptors then a concentration of specific receptors at the limbus, so close to the highly
sensitive cornea, would seem surprising. However, if their particular responsiveness
demands aggregation, together with a corpuscular form, then there would be no place
for them in the cornea, in the interests of vision. Since the maintenance of corneal
integrity is of utmost importance, a mechanism which could signal a potentially
damaging stimulus would be conveniently sited within the pericorneal zone.
Corpuscular nerve endings were found to be numerous within a narrow, 1-00 mm
wide, annular zone of limbal conjunctiva, located approximately 0 5 mm from the
corneoscleral margin. A light and electron microscopic study was carried out on these
nerve endings found within samples of human eye-bank material.
Corpuscular endings were found immediately under the epithelium, often within the
stromal elevations which make up the limbal palisades of Vogt. They were round to
oval in shape, and varied in size, with a mean maximum diameter of 30/tm. The
afferent nerve fibre lost its myelin sheath soon after entry, and subsequently branched
to give rise to a variable number ofaxon terminal varicosities, which were characterised
by an accumulation of mitochondria. Neural elements within the nerve ending were
invested by the cytoplasmic lamellae of Schwann-like accessory cells. The corpuscle
was demarcated from the surrounding connective tissue by a delicate fibrocyte
The corpuscular nerve endings described here in the conjunctiva share features
common to corpuscles found in other mucosae. The function of such complex sensory
nerve endings is as yet unknown, but the possibility that they represent receptors for
particular sensory modalities should be explored.
J. G. LAWRENSON AND G. L. RUSKELL
CHOUCHKOV, C. (1973). On the fine structure of Krause bulbs in the human skin, oral cavity, and rectum.
Archivum histologicum japonicum 35, 365-375.
CHOUCHKOV, C. (1978). Cutaneous receptors. Advances in Anatomy, Embryology and Cell Biology 54, 7-61.
DoGIEL, A. S. (1891). Die Nervenendk6rperchen (Endkolben, W. Krause) in der Cornea und Conjunctiva
bulbi des Menschen. Archivfur mikroskopische Anatomie 37, 602-619.
EHRLICH, P. (1886). Uber die Methylenblaureaction der lebenden Nervensubstanz. Deutsche medizinische
Wochenschrift 12, 49-52.
FREY, M. VON (1906). The distribution of afferent nerves in the skin. Journal of the American Medical
Association 47, 645-648.
HALATA, Z. (1975). The mechanoreceptors of the mammalian skin. Ultrastructure and morphological
classification. Advances in Anatomy, Embryology and Cell Biology 50, 7-75.
HALATA, Z. & MUNGER, B. L. (1986). The neuroanatomical basis for the protopathic sensibility of the human
glans penis. Brain Research 371, 205-230.
GGO, A. & OGAWA, H. (1977). Correlative physiological and morphological studies of rapidly adapting
mechanoreceptors in cat's glabrous skin. Journal ofPhysiology, London 266, 275-296.
JOHANSSON, R. S. & VALLBO, A. B. (1983). Tactile sensory coding in the glabrous skin of the human hand.
Trends in Neurosciences 6, 27-32.
KEY, A. & RETZIUS, G. (1876). Studien der Anatomie und des Nervensystems und des Bindegewebes, vol. ii, pp.
211-220. Stockholm: Samson and Wallin.
KNUSEL, 0. & VONWILLER, 0. (1922). Vitale Farbungen am menschlichen Auge. Zeitschriftfur Augenheilkunde
KRAUSE, W. (1859). Uber Nervenendigungen. Zeitschrift rationelle Medizin, 3rd Series, 5, 28-43.
KRAUsE, W. (1881). Die Nervenendigungen innerhalb der terminalen Tastk6rperchen. Archiv far mikro-
skopische Anatomie 19, 53-136.
LIGHTBODY (1867). Observations on the comparative microscopic anatomy of the cornea of vertebrates.
Journal ofAnatomy 1, 15-43.
MACDONALD, D. M. & SCHMIDT, D. (1979). Ultrastructure of the human mucocutaneous end organ. Journal
ofInvestigative Dermatology 72, 181-186.
MALINOVSKY, L. (1979). Lamellar complexes in the human female genital nerve endings. Folia morphologica
MALINOVSKY, L. (1988). What is a sensory corpuscle? In Mechanoreceptors. Development, Structure and
Function (ed. P. Hnik, T. Soukup, R. Vejsada and J. Zelena), pp. 283-286. New York: Plenum Press.
NAGAI, T. (1982). Encapsulated sensory corpuscle in the mucosa ofhuman vocal cord. Archivum histologicum
japonicum 45, 145-153.
OPPENHEIMER, D. R., PALMER, E. & WEDDELL, G. (1958). Nerve endings in the conjunctiva. Journal ofAnatomy
POLACEK, P. & MALINOVSKY, L. (1971). Die Ultrastruktur der Genitalkorperchen in der Clitoris. Zeitschrift
far mikroskopisch-anatomische Forschung 84, 293-3 10.
RIISAGER, P. M. (1962). A clinical study of the numerical variation with age of compact nerve endings in the
human conjunctiva. Journal ofNeurology, Neurosurgery and Psychiatry 25, 134-136.
SPASSOVA, I. (1965). On the structure of encapsulated nerve endings in the cat and its functional significance.
Zeitschrift far mikroskopisch-anatomische Forschung 72, 366-382.
SPASSOVA, I. (1973). Ultrastructure of Krause endbulbs in the nasal skin of the cat. Acta anatomica 84,
SPASSOVA, I. (1974). Ultrastructure of the simple encapsulated nerve endings (simple endbulbs of Krause) in
the tongue of the cat. Journal ofAnatomy 118, 1-9.
STRUGHOLD, H. & KARBE, M. (1925). Vitale Farbung des Auges und experimentelle Untersuchung der
gefarbten Nervenelemente. Zeitschrift far Biologie 83, 297-308.
SUCHARD, E. (1884). Recherches sur la structure des corpuscles nerveux de la conjonctive et des organes
genitaux. Archives de Physiologie 4, 337-347.
VOGT, A. (1921). Atlas of Slitlamp Microscopy of the Living Eye. Berlin: Springer.
WOLTER, R. (1964). The double nerve supply of Krause corpuscles at the human corneal limbus. Archives of
Ophthalmology 71, 409-411.