Article Text

Download PDFPDF

Nerve terminals at the human corneoscleral limbus
  1. Mouhamed A Al-Aqaba,
  2. Fady S Anis,
  3. Imran Mohammed,
  4. Harminder S Dua
  1. Larry A Donoso Laboratory for Eye Research, Section of Academic Ophthalmology, Division of Clinical Neuroscience, University of Nottingham, Nottingham, UK
  1. Correspondence to Professor Harminder S Dua, Larry A Donoso Laboratory for Eye Research, Section of Academic Ophthalmology, Division of Clinical Neuroscience, University of Nottingham, Nottingham, UK; harminder.dua{at}


Aims To demonstrate and characterise distinct subepithelial compact nerve endings (CNE) at the human corneoscleral limbus.

Methods Ten fresh human donor corneoscleral discs (mean age, 67 years) and 26 organ-cultured corneoscleral rims (mean age, 59 years) were studied. All samples were subjected to enzyme histochemical staining related to endogenous acetylcholinesterase present in nerve tissue and H&E staining. Whole-mount en face imaging with NanoZoomer digital pathology microscope and serial cross-section imaging with light microscope were undertaken.

Results Nerves entering the corneoscleral limbus and peripheral cornea terminate under the epithelium as enlarged multiloculated and multinucleated ovoid structures within a 2 mm zone. They are closely associated with the rete pegs of the limbal palisades and the limbal epithelial crypts, often located within characteristic stromal invaginations of these structures. Their numbers ranged from 70 to 300 per corneoscleral rim. The size ranged from 20 to 100 µm. They had one or more nerve connections and were interconnected to other similar endings and to the limbal nerve plexus.

Conclusion Human corneoscleral limbus demonstrates a population of nerve terminals resembling CNE with distinct morphological features. They are closely associated with the limbal stem cell niches, suggesting a potential contribution to the niche environment.

  • cornea
  • conjunctiva
  • anatomy
  • experimental– laboratory
View Full Text

Statistics from


The specialised sensory nerve endings in cutaneous and subcutaneous tissues are diverse. These receptors are regarded as selective peripheral encoding devices and can be divided into three groups based on their function: mechanoreceptors, nociceptors and thermoreceptors. On the basis of the morphology, sensory nerve fibres terminate in the peripheral tissues of the body either as free nerve endings or encapsulated compact structures also referred to as compact nerve endings (CNE).1 The latter are often called ‘receptors’ or ‘corpuscles’. Free nerve endings lack myelin and often serve nociception and thermoception functions.

The sensory ‘encapsulated’ receptors are specialised structures formed by a central axon coated by differentiated Schwann and endoneurial/perineurial-related cells.2 They are located mainly in glabrous skin such as the ventral portion of the fingers, palms, soles of feet, lips, labia minora and glans penis.3 Functionally they are rapidly adapting low-threshold mechanoceptors, which are responsible for light-touch sensation in mammals.4 Based on their structure and tissue organisation, different types of skin receptors have been identified. These include Meissner corpuscles, Pacinian corpuscles, Merkel cells and Ruffini receptors.

The human cornea is a richly innervated structure. It is densely supplied by sensory and, to a lesser extent, autonomic nerve fibres.5 The sensory nerves, predominantly derived from the ophthalmic division of the trigeminal nerve, constitute the majority of corneal nerves.6 7 Corneal nerves enter the anterior two-thirds of the stroma from the corneoscleral limbus and give rise to a highly branched axonal network. They perforate Bowman’s zone to terminate in bulb-like structures form, in which a leash of neurites emerge to spread as the sub-basal nerve plexus. The nerves eventually enter the corneal epithelium and, after additional branching, terminate as free nerve endings in very close contact with the epithelial cells.8

Corneal nerves exert a variety of sensory (afferent) and efferent functions. Mechanical, thermal and chemical stimulations of the corneal nerves produce predominantly a sensation of pain or irritation in humans.9 In contrast, conjunctiva and facial skin perceive equivalent stimuli as touch, warm or cold feelings.10

Little is known about limbal innervation, although this is a highly specialised area of the ocular surface in the context of limbal stem cells, vasculature, antigen-presenting cells (Langerhans and dendritic cells) and the trabecular meshwork. We undertook a study to explore the nerve terminals at the limbus and report the presence of acetylcholinesterase (AchE)-positive CNEs at the limbus that are distinct from the nerve endings seen in the cornea.


The research was conducted at The University of Nottingham, Nottingham University Hospital NHS Trust, Queens Medical Centre, UK.

En face examination of flat mounts and of cross sections of human corneas and corneoscleral rims was carried out.

Study of flat mounts

Ten fresh (within 72 hours) normal human corneas from five deceased patients with a mean age of 67 years (range 55–73 years) and 17 eye bank corneoscleral rims maintained in organ culture (mean age=57.6 years, range 20–71 years) were used in this experiment. The corneoscleral rims used remained in organ culture for an average of 8 days before being processed (range 3–15 days). Due to the method of retrieving the corneoscleral rim samples through an eye bank, the specific orientation of the rim could not be identified, and therefore it was not possible to determine if the structures identified were unique to a specific quadrant or clock hours.

AchE method for corneal nerve demonstration

The cholinesterase enzymes, found along the corneal sensory nerve axons, are believed to be responsible for the maintenance of the ionic gradient along the axons during propagation of the nerve impulse.11 This method has been described in detail previously.12 Briefly, the corneal buttons were fixed in cold 4% formaldehyde (pH 7) for 4 hours and then rinsed overnight in phosphate buffered saline. Specimens were incubated in the stock solution containing acetylthiocholine iodide as a substrate for 16–24 hours at 37°C. The AchE enzyme in the nerves reacted with acetylthiocholine iodide in the substrate to produce a brown colouration of the nerves. The colour was then intensified with a dilute solution of ammonium sulfide. Specimens were dehydrated by immersion in alcohol and were cleared in xylene. The specimens finally were mounted between a slide and coverslip and were prepared for image analysis.

Study of cross sections

Nine human corneoscleral rims from nine donors with a mean age of 64 years (range 48–83 years) were processed for histological examination. Sixteen blocks of tissue corresponding to the entire circumference of the corneoscleral rim were cut and stained by the AchE method as described above with the exception of the dehydration and clearing stages. Areas of CNE were identified, and the tissue blocks or smaller blocks cut from the main block corresponding to abundant location of the CNE were embedded in optimal cutting temperature compound and snap-frozen for cryomicrotomy. Seven-micron cryosections were stained with H&E and mounted on glass slides with a coverslip.

Image analysis

The stained specimens were photographed using a light microscope (Leica DM4000B; Leica Microsystems, Nussloch, Germany) and a Hamamatsu NanoZoomer digital pathology microscope system (Hamamatsu, Hamamatsu City, Japan). Oil immersion technique was used to examine the corpuscular terminations at a higher magnification (×1000). The areas of interest were serially imaged in the Z-axis starting from the most superficial layer in order to study the relationship of the CNE to the corneal nerves and other structures. The images were then stacked and merged to give a single, holistic, detailed anatomical view of the area. Merging the images was done through a Z-stacking software, AllFocus (AllFocus—Extended Depth of Field Software; Saphicon, Palo Alto, California, USA). Adobe Photoshop CS4 Extended (Adobe Systems, San Jose, California, USA) was required for additional image processing. For the measurement of the size of CNE, the widest diameter was considered.

The Pearson correlation coefficient was used to determine the correlations between the age of the donors and the number of the CNE.


Study of flat mounts

Nerves within the limbal area were found to terminate under the epithelium as enlarged ovoid structures resembling CNE. The majority of these CNEs were located in the peripheral 2 mm zone of the cornea (figure 1). They were found in all samples. Their numbers were variable, ranging from 70 up to 300 per corneoscleral rim. The size varied from 20 to 50 µm.

Figure 1

Photomicrographs of a whole-mount acetylcholinesterase-stained corneoscleral rim. (A) Three clusters of compact nerve endings (CNE), white arrows and black box, are seen. (B) A magnified photomicrograph of the selected area in (A) showing one cluster of 5 CNEs. (C) A large ovoid CNE is seen under the epithelium. (D) A close-up view of the CNEs and their tortuous nerve fibres. (E) En face identification of CNEs grouped together in clusters in a different corneoscleral rim. (F) Magnification of box seen in (E). Bar=1000 µm (A), 250 µm (B), 50 µm (C) and 120 µm (D), 500 µm (E) and 200 µm (F).

The CNEs were predominantly arranged in clusters and were randomly distributed along the circumference without any predilection for any specific quadrant (see online supplementary figure 1). These clusters were often found in a close proximity to limbal palisades of Vogt (figure 2A,B).

Supplementary file 1

Figure 2

Photomicrographs of whole-mount acetylcholinesterase-stained cornea. (A,B) Compact nerve endings (CNE) (arrows) distributed around limbal palisades of Vogt (arrowheads). (C) En face identification of nerve fibres in the deep stromal plexus (arrowheads) and superficial stromal plexus (in black box). White arrows point to CNE in the peripheral cornea. (D) A parent straight nerve trunk in the superficial plexus is seen giving rise to tortuous nerve branches (arrows). The tortuous nerves terminate in CNE as seen in ‘C’ above. (E) Two tortuous nerve fibres are seen terminating in an ovoid CNE. (F) A group of three interconnected CNEs of different sizes and shapes. (G) Another group of relatively larger three interconnected CNEs. The deep plexus of thicker nerves (out of focus) is clearly visible. (H) A cluster of CNEs, which are all connected to fibres (arrowheads) arising from the same parent trunk (arrow) in the superficial plexus. A large thick nerve of the deep plexus is also seen. Bar=100 µm (A,B), 500 µm (C), 100 µm (D), 10 µm (E), 25 µm (F), 120 µm (G) and 100 µm (H).

CNEs were associated with a perilimbal plexus of superficial and deep nerves. In the superficial plexus, a parent nerve, which showed a straight course along the limbus, gave rise to very tortuous nerves, arising almost at right angles from the main trunk and took a course either towards the peripheral cornea or towards the sclera (figure 2C,D). The tortuous nerves terminated in CNE, at times two of them in one CNE (figure 2E,F). CNEs in turn were seen to be connected to adjacent CNEs by nerve fibres in the superficial plexus on a one-to-one or one-to-two basis (figure 2B,E–H).

The deep stromal nerves appeared as thick nerve trunks, which did not appear to have any direct connection with the CNE but took a course towards the corneal periphery to enter the stroma, providing innervation to the cornea (figure 2C,D).

The staining of CNE was not homogeneous, showing areas of light and dark staining delineating small loculi (figure 2E,F). At a higher magnification, CNEs were seen to be made of many layers of convoluted lamellar tissue with apparent loculated spaces (figure 3 and see online supplementary figure 2).

Supplementary file 2

Figure 3

Sequential photomicrographs of whole-mount acetylcholinesterase-stained cornea taken at 5 µm intervals through a compact nerve ending (CNE) of a different sample, from the surface (A) to the deeper aspect (F). The internal lamellar structure is visible along with loculations, which are more obvious in the deeper sections (E,F), suggesting that the more superficial part of the CNE is relatively compact. The nerve fibre connection is again visible in the deeper part of the CNE. Epithelial cells are seen in the superficial sections (A,B). Bar=25 µm.

Experiment 2 (cross-section analysis)

AchE prestaining of corneoscleral rims prior to cryosectioning allowed accurate localisation of tissue to obtain cross section of the CNE. The prestain gave these structures a brownish colour. This method enabled visualisation of nerves within different levels of the stroma. Nerves identified within the mid-stroma could be traced in serial 7 µm sections of the corneoscleral rim as they approached the epithelium.

The cross sections showed that, in most instances, CNEs were located within the anterior loose limbal stroma residing just beneath the limbal epithelium. However, a few of the structures identified were noted to reside deeper within the underlying more compact stroma. Nerve connection, only in part, could be clearly seen in many sections.

CNEs were often found isolated with no indication of their terminal nerves. However, in a few samples, fine terminal nerve branches were identified at the point of entry of the structure. The complex (CNE and its terminal branch) can only be visualised when the orientation of the complex is parallel to that of the cross section (figure 4A,B).

Figure 4

(A,B) Photomicrographs of acetylcholinesterase prestained cross section displaying a subepithelial compact nerve ending (CNE) (arrows) and the corresponding nerve connection (arrowheads). (C–F) Cross section of four different CNEs showing the compartmentalised/loculated structure with associated nuclei. Each loculum has a nucleus giving the whole structure a multinucleated appearance. Bar=20 µm (A–F).

Higher magnification images taken using the oil immersion technique allowed for the internal structure of the CNE to be examined closely. The loculated structure of the CNE was confirmed and a nucleus was seen to be associated with each loculum or compartment, giving the whole structure a multinucleated appearance (figure 4C–F).

The mean cross-sectional dimension of CNE, calculated from 26 different CNEs in nine different samples, was 60 µm (ranging from 20 to 100 µm). CNEs were also associated with other defining features within the corneal limbus. The 7 µm serial sections allowed us to track changes in the structure of the limbal epithelium in relation to the CNE. CNEs were often located within characteristic focal projections of the rete pegs of the limbal palisades (figure 5 and see online supplementary figure 3) and the limbal epithelial crypts (LECs) (figure 6), which are identified stem cell niches. In some instances CNEs in association with LECs were found to be completely surrounded by a nest of epithelial cells (figure 7).

Supplementary file 3

Figure 5

Photomicrographs of serial acetylcholinesterase prestained sections displaying a compact nerve ending (arrows) developing into a cavity within the limbal epithelium. Direction of the arrows also displays entry point of the cavity. Bar=100 µm (A–C).

Figure 6

Photomicrographs of serial acetylcholinesterase prestained sections of a compact nerve ending (arrows) invaginating into the limbal epithelium and closely associated limbal epithelial crypts (arrowheads). B–F represent sequential sections of the region marked by the black box in ‘A’. Bar=100 µm (A), 50 µm (B–F).

Figure 7

Photomicrographs of serial acetylcholinesterase prestained sections of limbus demonstrating compact nerve endings (arrows and arrowheads) completely wrapped by focal projections of limbal epithelium in to the stroma, as described with limbal epithelial crypts. Bar=50 µm (A–D).

Finally, a summary of donor demographics and histological findings is shown in table 1.

Table 1

Summary of demographic and histological donor data


The human skin has a variety of nerve terminations ranging from free endings to specialised sensory structures termed corpuscles or receptors, which respond to specific stimuli such as pain nociceptors (free nerve endings), touch and vibration (Meissner and Pacinian), and thermal sensation.13 14 The ocular surface, which represents a highly specialised part of the integument, presents similar specialised sensory structures and free endings.1 The human cornea contains exclusively free nerve endings, which densely populate the superficial layers of the epithelium.8 Corneal nerves, once believed to serve merely as sensory receptors conveying innocuous and noxious stimuli, have emerged as pivotal players in a complex regulatory system involved in immune, inflammatory and trophic functions of the cornea.15

The conjunctiva has numerous CNEs, which are well known to morphologists as sensory receptors for cold ‘thermoceptors’. Krause first described subepithelial CNEs in the human conjunctiva and named them Endkolben.16 They represent the termination of myelinated nerve fibres throughout the conjunctiva. Subsequent studies confirmed the morphological heterogenecity and the distribution of these structures.1 Riisager, using a methylene blue dye instilled into the conjunctival sac of 31 living human eyes, showed that approximately half CNEs in each case were found in the lateral, superior quadrant of the bulbar conjunctiva, none of which were seen in the limbal region. The number of CNE has been shown to increase by about 1.25 for each year of life. The morphology of conjunctival CNE varies from small elongated structures to large ovoid bodies.17 Furthermore, previous observations revealed that in cattle, CNEs of the conjunctiva were absent in calves and first appear at 2–3 years of age.1 In human conjunctival CNEs are absent before puberty. This has raised concerns over the proposed specialised sensory function of CNE as they are absent in young members of the species and most numerous in the elderly.17 These observations have led to the proposal that the presence of conjunctival CNE is a tissue response to environmental exposure.1 Other workers disputed previous theories and suggested that these structures could be the result of continuous regenerative and degenerative process, such as those reported in other parts of the peripheral nervous system.18

Similar CNEs have also been found in the human scleral spur and showed lack of reactivity to AchE stain.19

Limbal CNEs have not been well studied or characterised. They appear to have a distinct morphological appearance and anatomical location. They are large ovoid structures that are located in the 2 mm annular zone of the corneal limbus, in the peripheral cornea and adjacent limbal conjunctiva. We found them to be located under the epithelium and connected to tortuous nerve branches. They were identified in all samples of different ages, the youngest being from a patient in the third decade of life. They were predominantly located within invaginations in the limbal rete pegs and LECs, a region thought to be the niche for limbal stem cells. Occasionally they were completely wrapped by focal projections of limbal epithelium in to the stroma, as described with LEC.

Lawrenson and Ruskell described similar CNE in the limbal conjunctiva using gold chloride impregnation method. The CNEs were found immediately under the epithelium and within a narrow, 1.00 mm wide, annular zone of limbal conjunctiva, located approximately 0.5 mm from the corneoscleral margin.20 They however did not mention if conjunctival CNEs were extending into the peripheral cornea even though the samples they studied were of scleracorneal rims left over from corneal transplantation. They also used a different stain, which may not have picked up the peripheral corneal CNE.

The presence of limbal CNE raises questions about their role. In an effort to determine the functional role of CNE, the same authors investigated the touch sensitivity of the limbal conjunctiva using a Cochet-Bonnet aesthesiometer.21 Touch sensitivity was found to be significantly higher in the palisade zone compared with the adjacent conjunctiva. The temporal limbus showed greater sensitivity than the inferior limbus. As a result, the authors have proposed a mechanoreceptive role for these CNEs. However, qualitative histological data on limbal CNE presented in this study indicate a rather random distribution around the limbus and question the basis for their conclusion.

It is worth noting that CNEs have been found in the region of the iridocorneal angle of cetaceans.22 They are frequently seen, as clusters or singles, in the stroma of the ciliary body, within the trabecular meshwork and sclera in the region of the iridocorneal angle. The intraocular pressure (IOP) in dolphins can vary from 33.32 to 33.56 mm Hg in natural horizontal position to 57.33–62.8 when beached.23 Therefore, it has been suggested that the CNE within the cetacean eye may function as pressure receptors, possibly to regulate IOP.

The limbus has unique attributes of which an important one is its role as a repository of epithelial stem cells. The palisades of Vogt as a whole and the LEC specifically are considered to represent the corneal epithelial stem cell niche.24 Our finding of the close proximity of limbal CNE to LECs is a novel observation. The crosstalk between nerve endings and epidermal cells is important for morphogenesis and homeostasis.25 A recent study showed that the nerve endings/Schwann cells control Lgr6 expression in skin, suggesting that they play a role in the regulation of dermal epithelial cells.26 Lgr6, a stem cell marker, closely localised with the surrounding nerve endings and their corresponding Schwann cells throughout the hair cycle. Ablation of cutaneous nerves results in degeneration of Schwann cells and diminished expression of Lgr6. The LEC also expresses stem cell proteins and, like the skin stem cell niche, is shown here to be very closely related to limbal CNE, suggesting a similar role for these CNEs in relation to maintenance of the niche and cell proliferation and differentiation.

Based on the findings of this study, the consistent morphology, the attachments to one or more nerve fibres, the location in the peripheral cornea and the close association with LEC, we feel that the term CNE used for similar structures in the skin and conjunctival is not appropriate for limbal CNE and propose that the term ‘limbal nerve corpuscles (LNC)’ be used instead. Table 2 summarised the main differences between LNC and the conjunctival and skin CNE, indicating that the LNCs are not the same as the latter.

Table 2

Comparison of histological features of limbal nerve corpuscles (LNC) and conjunctival compact nerve endings (CNE)

The localisation and characterisation of these structures in the limbus allow some deductions on the possible role they might play. Their precise role is unknown and will require functional studies on animal models. The potential interaction with limbal stem cells will be an important area for investigation.


View Abstract


  • Contributors Concept and research design: MAAl-A and HSD. Data collection: MAAl-A and FSA. Data analysis and interpretation: MAAl-A, FSA, IM and HSD. Supervision: HSD. All authors participated in manuscript preparation.

  • Funding Elizabeth C. King Trust, Pittsburgh, PA, and the Philadelphia Eye Research Foundation.

  • Competing interests None declared.

  • Ethics approval The study was approved by the East Midlands Research Ethics Committee 2 (REC no 06/Q2403/46).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.