|Home | About | Journals | Submit | Contact Us | Français|
Embryonic chick nerves encircle the cornea in pericorneal tissue until embryonic day (E)9, then penetrate the anterior corneal stroma, invade the epithelium, and branch over the corneal surface through E20. Adult corneal nerves, cut during transplantation or LASIK, never fully regenerate. Schwann cells (SCs) protect nerve fibers and augment nerve repair. This study evaluates SC differentiation in embryonic chick corneas.
Fertile chicken eggs were incubated from E0 at 38°C, 45% humidity. Dissected permeabilized corneas plus pericorneal tissue were immunostained for SC marker proteins. Other corneas were paraffin embedded, sectioned, and processed by in situ hybridization for corneal-, nerve-related, and SC marker gene expression. E9 to E20 corneas, dissected from pericorneal tissue, were assessed by real-time PCR (QPCR) for mRNA expression.
QPCR revealed unchanging low to moderate SLIT2/ROBO and NTN/UNC5 family, BACE1, and CADM3/CADM4 expressions, but high NEO1 expression. EGR2 and POU3F1 expressions never surpassed PAX3 expression. ITGNA6/IT-GNB4 expressions increased 20-fold; ITGNB1 expression was high. SC marker S100 and MBP expressions increased; MAG, GFAP, and SCMP expressions were very low. Antibodies against the MPZ, MAG, S100, and SCMP proteins immunostained along pericorneal nerves, but not along corneal nerves. In the cornea, SLIT2 and SOX10 mRNAs were expressed in anterior stroma and epithelium, whereas PAX3, S100, MBP, and MPZL1 mRNAs were expressed only in corneal epithelium.
Embryonic chick corneas contain SCs, as defined by SOX10 and PAX3 transcription, which remain immature, at least in part because of stromal transcriptional and epithelial translational regulation of some SC marker gene expression.
The cornea is one of the most highly innervated tissues on the surface of the body. During chick development, corneal nerves, part of the peripheral nervous system, are derived from neural crest cells, which reside together with ectodermal placode– derived nerve cells in the ophthalmic lobe of the trigeminal ganglion.1 Hamburger-Hamilton stage 13 to 15,2 embryonic day (E)4, trigeminal ganglion ophthalmic lobe placode– derived postmitotic nerves express high levels of transmembrane axon-guidance EphA3 receptor kinase.3 These nerves extend from the trigeminal ganglion into the ophthalmic periocular head mesenchyme, which does not express ephrin-A2 and -A5.4 Trigeminal ganglion ophthalmic postmitotic neural crest– derived nerves follow the placode-derived nerves to the periocular mesenchyme, where, from E5 to E8, these periocular nerve bundles subdivide and extend dorsally and ventrally around the cornea, forming a limbal pericorneal nerve ring, but are repelled from entering the cornea.5,6 Soluble neurorepellant semaphorin 3A (SEMA 3A), produced by developing lens and diffusing through the cornea and into the adjacent periocular mesenchyme, is responsible for corneal exclusion of neuropilin receptor-expressing periocular nerves during this period.7 In addition, secreted neuroguidance SLIT ligands are synthesized by embryonic lens epithelia, and could interact with axonal transmembrane roundabout (ROBO) receptors expressed on outer surfaces of growing axons to repel corneal nerve growth.8 On E9 neural crest– derived limbal ring nerves defasciculate, and sensory nerves invade the anterior corneal stroma simultaneously from all around its perimeter,1,5,6,9 then branch and extend anteriocentrally, penetrating the epithelium by E12 and reaching the cornea center by E14.6 It has been suggested6 that extracellular highly sulfated keratan sulfate proteoglycan (KSPG) accumulation, beginning in the posterior corneal stroma by E9,10,11 blocks the diffusion of lens SEMA 3A and probably also lens SLIT2, thus allowing the neural crest– derived sensory periocular nerves to grow into the anterior corneal stroma. Subsequent progressive accumulation of highly sulfated KSPG anteriorly across the stroma from E9 to E1610 guides the corneal nerves toward the epithelium as they seek to avoid highly sulfated KSPG.12,13
Peripheral nervous system nerves are accompanied by neural crest– derived Schwann cells (SCs),14 which undergo three main developmental transitions: from migrating neural crest cells to SC precursors, SC precursors to immature SCs, and immature SCs to mature myelinating SCs or mature nonmyelinating SCs.15 The final transition requires intimate contact between nerves and SCs and is accompanied by withdrawal of SCs from the cell cycle.13 As SCs differentiate, they express SC-related proteins in transition-stage characteristic patterns, as summarized in Table 1. Transcription factors SOX10, PAX3, POU3F1, and EGR2 are expressed in SCs in a temporally orchestrated pattern: SOX10 is expressed in neural crest cells and all subsequent SC developmental stages including myelination16; PAX3 is expressed in neural crest cells, SC precursors, immature SCs, and mature nonmyelinating SCs, but not in mature myelinating SCs17; POU3F1 is expressed in immature SCs and induces the transition of immature SCs to mature myelinating SCs, but later is reduced in expression in mature myelinating SCs18; expressions of EGR2, myelin-associated glycoprotein (MAG), myelin basic protein (MBP), and P0 (MPZ) genes are strongly upregulated at the onset of myelination in mature myelinating SCs and remain high throughout myelination.15 MPZ, with expression directly regulated by Sox10,19 is expressed by SC precursors, immature SCs, and mature myelinating SCs; S100 protein (S100) is expressed by immature SCs and by mature myelinating and nonmyelinating SCs, but not by SC precursors; and glial fibrillary acid protein (GFAP) is expressed by immature SCs and mature nonmyelinating SCs, but not by mature myelinating SCs.15 In addition, integrin beta 1 (ITGNB1) is expressed only in mature nonmyelinating SCs.15
If nerves are damaged such that close contacts with SCs are lost, mature myelinating SCs dedifferentiate into immature SCs and actively contribute to nerve regrowth.20 Neuregulin (NRG)-1 type III,21 (NRG1-III), a sensory and motor nerve-derived factor (SMDF) neuregulin isoform,22 is expressed by both peripheral nervous system axons21 and SCs23 and determines both whether adjacent SCs will myelinate nerves or just ensheathe them24 and how thick the myelin sheath will be.25 BACE1, a β-secretase that is expressed in peripheral nerves, processes SMDF neuregulin so that it can interact with its receptors; BACE1 is required for proper myelination in mice.26 CADMs, cell adhesion molecules, mediate close membrane interactions between adjacent cells. The interaction of axon-expressed CADM3 and SC-expressed CADM4 has been shown to be critical for myelination.27 Secreted netrins (NTNs) function as positive axon guidance factors when they bind nerve transmembrane deleted-in-colorectal-cancer (DCC)28,29 or neogenin (NEO)30 receptors, as repulsive factors when they interact with transmembrane UNC5 receptors29,31 or with UNC5-DCC heterodimers,32 and as neuronal survival factors.33 Alternatively, NTNs can be diverted from their role in axon guidance to regulate epithelial cell adhesion and migration by interacting with epithelial cell membrane α6β4 and α3β1 integrins.34 SCs interact with extracellular matrix molecules to stimulate peripheral nerve regeneration.35 Indeed, SCs accompanying stromal and subepithelial corneal nerves have been hypothesized to facilitate stromal and subepithelial nerve regeneration after rabbit corneal injury.36
When corneas are damaged during LASIK surgery, corneal transplantation, or accidental injury, nerve regeneration is slow and often incomplete.37,38 Evidence for the presence of SCs in the cornea rests on ultrastructural observations in adult mammalian corneas,39– 41 but an attempt to demonstrate their presence by staining for SC marker GFAP and S100 proteins immunocytologically was unsuccessful.39 To examine in more detail the expression of SC-related genes in the cornea, we used quantitative real-time PCR (QPCR) and in situ hybridization to reveal expressions of SC marker mRNAs, and immunocytochemical staining to detect SC marker protein expressions. Our results show that around the periphery of the cornea mature myelinating SCs associate closely with and myelinate pericorneal nerves. In contrast, within the cornea proper, many SC marker– expressing cells are not closely associated with corneal nerves. Moreover, several SC marker mRNAs are expressed in the corneal epithelium, but only the basal SC marker SOX10 is expressed in the corneal stroma, suggesting that corneal SCs are restricted to the immature SC stage and may be transcriptionally regulated in the stroma and translationally regulated in the epithelium.
Fertile White Leghorn chicken eggs were transferred to a 38°C incubator on E0 and incubated at 45% humidity. Corneas from embryos of the desired age were dissected in sterile saline G (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4, 6.1 mM glucose, 0.6 mM MgSO4, and 0.1 mM CaCl2 [pH 7.4]). For RNA isolation, the corneas were dissected from the pericorneal tissue, and then quick frozen in liquid nitrogen and stored at −70°C until used. These corneas include both central, mid-central, and lateral corneal tissue, as defined by Müller et al.40
RNA isolation, cDNA synthesis, and QPCR were performed as described previously.6,42 Sequences for genes of interest were obtained from GenBank (gi; http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). PCR primers for QPCR were designed (Designer 31 Molecular Beacons Design; Sigma-Aldrich, St. Louis, MO) to amplify fragments between 80 and 150 base pairs in length and are listed in Table 2. Gene names conform to guidelines established by the Second International Workshop on Poultry Genome Mapping, 1994 (described at http://www.chicken-genome.org). Each primer set generates only one amplified band with chick corneal cDNA. For QPCR, cDNA dilutions (X) were chosen so that the cycle threshold (Ct, the number of PCR cycles required for the fluorescent signal that measures product accumulation for a given primer set to exceed background levels) for housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was between 13 and 16. All comparative QPCR reaction series consisted of duplicates for 1× and ×/10 cDNA dilutions for each PCR primer pair. Primer pair efficiencies for the 10-fold cDNA dilution were between 90% and 110%. GAPDH expression was chosen for normalization of all gene expression. Because corneal GAPDH expression was greater than the expression of any of the other genes included in this study, graphic representation of the normalized data employs a negative y-axis.
Whole cornea staining was performed as described previously.6,43 The corneas were dissected to include pericorneal tissue as well as peripheral and central corneal tissue. Each cornea was slit radially through the peripheral two thirds of the tissue three times around the cornea to allow antibodies to diffuse laterally into the corneal stroma and epithelium.43 Primary antibodies included: corneal nerves: anti-neuronal class III β-tubulin (TUJI; CRP Inc., Denver, PA), diluted 1:1000; anti-GFAP (Abcam Inc., Cambridge, MA), diluted 1:500; anti-MAG (United States Biological, Swampscott, MA), diluted 1:500; anti-MPZ (1E8; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), diluted 1:500; anti-S-100 (Vector Laboratories, Inc., Burlingame, CA), diluted 1:500; and anti-SCMP (anti-SMP; Developmental Studies Hybridoma Bank), diluted 1:500. The secondary antibody was horse-radish peroxidase– conjugated goat anti-mouse IgG [H+L] (Invitrogen, Carlsbad, CA) diluted 1:400. Stained corneas were stored in 100% glycerol at 4°C. They were photographed with a dissecting microscope (model M5; Wild Heerbrugg, Heerbrugg, Switzerland) with a digital camera (Coolpix 995; Nikon, Tokyo, Japan) and transmitted light from a light source placed below the glass stage.
PCR primers used to generate cloned gene fragments for Kera, 833 bp; Lum, 944 bp; Slit2, 647 bp; MPZL1, 662 bp; Sox10, 1091 bp; PAX3, 810 bp; S100, 970 bp; and MBP, 909 bp probes are listed in Table 3. The cDNAs, obtained by reverse transcription PCR from E10 to 14 embryo corneal RNA, were cloned into pGEM-T vectors (Promega, Madison, WI), and sense and antisense riboprobes were synthesized with DIG RNA labeling kit (Roche, Indianapolis, IN) as described previously,44,45 digested with DNAse I, precipitated, re- dissolved in hybridization buffer, quantitated by dot blot, and stored at −20°C.
E14 eye fronts were dissected into PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4-7H2O, 1.4 mM KH2PO4 [pH 7.3]) containing 3.7% formaldehyde, freed of adherent lens and retina, fixed overnight at 4°C in modified Carnoy’s,46 washed three times in PBS to remove all fixative, dehydrated through an EtOH series, washed two times in xylenes, and embedded in paraffin. Paraffin block serial sections of 10 μm were mounted on slides (SuperFrost Plus; Fisher, Pittsburgh, PA), dried overnight at 40°C, and stored in desiccated boxes at −20°C until hybridization. Hybridization was performed as described previously,46,47 except that dewaxing was done with xylenes, prehybridization, and hybridization solutions were made according to Moorman et al.,48 and incubations were conducted in a gasket-sealed metal tray (InSlide Out; Boekel, Feastenville, PA) containing wipes soaked in 50% formamide/50% one time SSC (Kimwipes; Kimberly Clark, Neenah, WI) to prevent dehydration. Controls were performed using the sense digoxigenin-labeled RNA transcript of the corresponding antisense RNA probe, and final probe concentrations were 1 ng labeled probe/μL hybridization buffer. Hybridization solution (15–30 μL) was placed on each set of three sections; prehybridization and hybridization were conducted at 65°C. High-stringency washes were performed at 65°C. Antibody reaction and stain development were conducted in the same trays with 50 to 100 μL solution per section set and 50% PBS/50% H2O as the humidifying solution. Anti-digoxigenin-AP Fab fragments (Roche) were diluted 1:300 and probes were detected (BM Purple AP Substrate; Roche). Staining was stopped by washing in pH 5.5 PBS, and the slides were rinsed in pH 7.5 PBS, mounted in 70% glycerol/30%PBS, viewed with a microscope (Diaphot 300; Nikon), and photographed with a digital camera (Coolpix 995; Nikon).
Secreted NTNs are a complex family of nerve guidance molecules that can be attractive guidance factors when they interact with nerve transmembrane NEO receptors, or repulsive guidance factors when they interact with nerve transmembrane UNC5 receptors. In the developing chick cornea, NTN2L expression was low and unchanging from E9 to E20, whereas NTN1 and NTN 4 expressions were 10- and 200-fold lower, respectively, and increased only slightly from E9 to E20 (Fig. 1). Expressions of genes for repulsive NTN receptors UNC5A-C were marginally higher than NTN2L expression, with UNC5A and UNC5B expressions virtually equivalent to, and UNC5C expression consistently approximately 7.5-fold higher than, NTN2L expression (Fig. 1). In contrast, expression of the positive transmembrane NTN receptor NEO1 was 100-fold higher than that of NTN2L, was 10-fold higher than the expression of the most highly expressed negative receptor, UNC5C, and remained constant from E9 to E20 (Fig. 1). The SLIT/ROBO neuroguidance family members, on the other hand, were more highly expressed than any of the NTNs or their negative receptors, especially at the earlier ages, when the nerves are just entering the cornea,6 with SLIT2 expression 20-fold higher and ROBO1 expression 100-fold higher than NTN2L expression from E9 to E14 (Fig. 1), but then expressions of both SLIT2 and ROBO1 declined 10-fold from E14 to E20, when nerves are continuing to grow in the corneal epithelium6 (Fig. 1). In contrast, expression of ROBO2 remained constant at approximately 0.5-fold that of SLIT2 from E9 to E14, did not decline from E14 to E20, and was approximately 7.5-fold higher than SLIT2 and ROBO1 by E20. Expression of BACE1, whose enzymatic protein product is essential for processing neuronal neuregulin SMDF before its interaction with SC receptors,26 was moderate throughout corneal development, and did not change as the cornea became highly innervated (Fig. 1).
Expressions of the CADMs most closely associated with the axon–SC interaction necessary for myelination, CADM3 and −4, were moderate for neuronally expressed CADM3 and approximately 7.5-fold lower for SC-specific CADM4 in the cornea throughout development, and rose less than 2-fold from E9 to E20 (Fig. 2). CADM1 and −2 were 7.5-fold more highly expressed than CADM3 from E9 to E12, while nerves were penetrating the stroma and extending into the epithelium, but then declined in expression from E12 to E20 as CADM3 expression rose, although they remained more highly expressed than CADM4 (Fig. 2).
The expressions of the SC transcription factors PAX3, POU3F1, and EGR2 are regulated independently of one another, as they in turn regulate migration and stage transition of SCs from SC precursors to mature myelinating SCs in the peripheral nervous system.17,18 In the developing chick cornea, PAX3 was expressed at a moderate level, which did not change appreciably from E9 to E20 (Fig. 2). In contrast, POU3F1 and EGR2 expressions began 10-fold lower than that of PAX3, and rose 2-fold from E9 to E14. Then from E14 to E20, POU3F1 expression declined to 7.5-fold lower than PAX3, whereas EGR2 expression rose to 0.5-fold that of PAX3 (Fig. 2). However, PAX3 expression remained greater than that of both POU3F1 and EGR2 throughout corneal development.
In addition to their functions as nerve guidance factors, extracellular NTNs have been found in the basement membranes of epithelial tissues where they bind integrins α3β1 and α6β4 and become adhesive/migratory regulators for epithelial cells.34 In the developing chick cornea, the most highly expressed integrin was ITGNB1, with expression constant from E9 through E14, and then declining slightly from E14 to E20 (Fig. 3). In contrast, expression of ITGNA3 began 100-fold lower than ITGNB1 expression, rose approximately 7.5-fold from E9 to E14, and then declined back to the E9 level by E20 (Fig. 3). Although their ratios remained the same during E9 to E20, the significant difference in their expression levels suggests that ITGNB1 and ITGNA3 may not always function together in the developing chick cornea. On the other hand, ITGNB4 and ITGNA6 expressions were almost 100-fold lower than that of ITGNB1 at E9, but rose coordinately 20-fold to a level similar to that of ITGNB1 by E20. Their more equivalent expressions suggest that ITGNA6 and ITGNB4 function together in the developing chick cornea.
Various marker proteins are made by SCs as they differentiate from SC precursors to mature myelinating or nonmyelinating SCs15 (Table 1). MBP mRNA and protein expressions were strongly upregulated as the SCs transitioned from immature SCs to mature myelinating SCs.15 In the embryonic chick cornea, MBP mRNA expression was moderate on E9, increased approximately twofold from E9 to E12, but then did not increase beyond that level from E12 to E20 (Fig. 4). Similarly, the expression of S100, observed in all SCs in the peripheral nervous system,15 began at the same level as that of MBP in the cornea, declined slightly by E12, and then rose approximately 7.5-fold from E12 to E20, having equivalent expression with MBP on E14 and E20 (Fig. 4). In contrast, expressions of GFAP, expressed in mature nonmyelinating SCs,15 and of SCMP and MAG, upregulated in mature myelinating SCs,15 were 50- to 100-fold lower than the expression of MBP and S100 on E9, and did not increase from E9 to E20 (Fig. 4). No MPZ gene has been cloned in chicks, but genes for P0-like protein 1 (MPZL1) and P0-like protein 3 (MPZL3), which share significant homology with MPZ in their extracellular and transmembrane epitopes, have been identified.49,50 In the chick cornea, MPZL1 expression was 10-fold higher than MBP expression at E9 and did not change from E9 to E20 (Fig. 4). Expression of MPZL3 was the same as that of MBP and S100 on E9, but then increased 10-fold from E9 to E20, such that the expressions of MPZL1 and MPZL3 were almost equivalent by E20 (Fig. 4).
In the E14 chick cornea, in situ hybridization revealed that the gene for one of the corneal KSPG core proteins, keratocan (KERA; Figs. 5A–C), was expressed by stromal keratocytes and by endothelial cells lining the posterior surface of the cornea, but was not expressed in cells of the epithelium that cover the corneal anterior surface. SLIT2 (Figs. 5D–F), whose protein product is a secreted molecule that interacts with ROBO receptors on nerve growth cones,8 was also expressed by anterior stromal cells. Unlike KERA, however, SLIT2 was also expressed by the epithelial cells that cover the anterior surface of the E14 cornea and was not expressed by posterior corneal stroma cells or corneal endothelium. A third expression pattern was displayed by MPZL1, a transmembrane cell adhesion glycoprotein with an intracellular domain very different from that of SC-specific MPZ.49,50 MPZL1 was expressed only in the corneal epithelium (Figs. 5G–I), where its expression appeared stronger in the peripheral and mid-corneal epithelium than in the central corneal epithelium (Fig. 5G, arrowheads). MPZL1 was expressed primarily in the basal layer of the corneal epithelial cells (Fig. 5I).
From E14 through E21, pericorneal nerve bundles, as visualized by neuronal-specific class III β-tubulin immunostaining, formed a ring around the cornea (Fig. 6A, white arrow), defasciculated and extended through the limbus to the peripheral cornea (Fig. 6A, white arrowhead), and continued extension and significant branching radially toward the center of the cornea (Fig. 6A, black arrowhead). In contrast, immunostaining revealed that the SC marker proteins MPZ, S100, MAG, and SCMP, products of genes known to be upregulated in mature myelinating SCs,15 accumulated in the presumptive SCs accompanying the limbal pericorneal ring of nerve bundles (Figs. 6B–E, white arrows) and along defasciculating nerve trunks as they branched away from the limbal ring toward the peripheral cornea (Figs. 6B–E, white arrowheads), but then ceased to be detected from the peripheral to the central cornea (Figs. 6B–E, black arrowheads) where corneal nerves continued to extend (Figs. 6A, black arrowhead). This pericornea-restricted staining pattern was identical for all these proteins characteristic of mature myelinating SCs, even though MPZ and S100 are also known to be expressed in immature SCs,15 and there was a 100-fold greater expression of S100 mRNA in the developing cornea than of MAG and SCMP (see Fig. 4). Immunostaining for GFAP in the chick pericorneal tissue and peripheral and central cornea was not detected (data not shown).
The expressions of corneal keratocyte mRNA and SC marker mRNAs in pericorneal nerve bundles are shown in Figure 7. An overview of pericorneal nerve bundle location relative to the cornea is shown in Figures 8B, 8E, 8H, 8K, 8N, and 8Q. Stromal KSPG core protein lumican LUM mRNA was not expressed in the pericorneal nerve bundles (Fig. 7A), whereas immunocytochemically detected nerve-specific class III β-tubulin protein was heavily expressed in the many individual nerve fibers that composed the pericorneal nerve bundles (Fig. 7B). Transcription factor SOX10 mRNA, expressed by all peripheral nervous system neural-crest– derived SCs at all stages of differentiation,15 was highly expressed by many cells within this pericorneal nerve bundle (Fig. 7C), as was MBP mRNA, an SC marker known to be significantly upregulated in myelinating SCs15 (Fig. 7E). In contrast, S100 mRNA (Fig. 7F), which is expressed in immature, mature myelinating, and mature nonmyelinating SCs,15 was not as highly expressed by as many cells within the pericorneal nerve bundle, and transcription factor PAX3 mRNA (Fig. 7D), which is not expressed in mature myelinating SCs and has been shown to repress MBP expression in primary mouse SC cultures,17 was expressed by few cells within the pericorneal nerve bundle.
In the E14 cornea itself, mRNA expression of the KSPG core protein LUM, like that of KERA, was high in keratocytes across the entire corneal stroma and along the anterior side of the iris (Fig. 8A), but ceased at the cornea–limbus border (Fig. 8B), and did not occur in the corneal epithelial cell layer (Fig. 8C). In contrast, nerve-specific class III β-tubulin protein immunocytochemical staining occurred in the anterior corneal stroma (Figs. 8D, 8F, black arrowheads), throughout the iris (Fig. 8D), in limbal pericorneal nerve bundles (Fig. 8E, ring with arrow) and other defasciculated nerve segments (Fig. 8E), in nerve segments penetrating from the corneal stroma into the epithelial cell layer (Fig. 8F, white arrowhead), and in both basal and anterior epithelial layers (Fig. 8F). The SC transcription factor SOX10 mRNA expression pattern was similar to nerve-specific protein expression, occurring in peripheral and central anterior corneal stroma (Figs. 8G, 8I, black arrowheads), throughout the iris (Fig. 8G), in the limbus in both pericorneal nerve ring bundles (Fig. 8H, ring with arrow) and individual limbal cells (Fig. 8H), in SCs penetrating from the anterior stroma into the corneal epithelial layer (Fig. 8I, white arrowheads), and in many cells throughout the basal and anterior corneal epithelium (Fig. 8I). SC transcription factor PAX3 mRNA (Figs. 8J–L) expression patterns were similar to those of SOX10, except that PAX3 mRNA was not expressed in the corneal stroma, was expressed in few SC cells within the limbal pericorneal nerve bundle (Fig. 8K, ring with arrow), and was more heavily expressed in anterior epithelial cells than in basal epithelial cells (Fig. 8L). Similarly, immature and mature myelinating and nonmyelinating SC marker S100 mRNA (Figs. 8M–O) was expressed most strongly in the corneal anterior epithelium, throughout the iris, and throughout the limbus, but less strongly within the pericorneal nerve bundle than SOX10, and not at all in the stroma. Finally, the expression pattern of myelinating SC marker MBP mRNA (Fig. 8P–R) was similar to PAX3 and SOX10 patterns, except that MBP expression was high in the pericorneal nerve ring bundle (Fig. 8Q, ring with arrow), but less robust in other limbal tissue and the iris, and largely restricted to the anterior corneal epithelium (Fig. 8R).
To examine the spatial relationships between SC marker mRNA-expressing cells and nerve class III β-tubulin protein-expressing nerve segments, we developed sections of E14 corneas, first by in situ hybridization techniques to reveal the SC markers (blue) and then by double staining with antibody to reveal neuronal tubulin (red). LUM mRNA expression was restricted to the anterior iris (Fig. 9A), whereas neuronal tubulin protein was expressed throughout the iris (Figs. 9A, 9C, 9E, 9G, 9I). LUM mRNA expression was so intense in the corneal stroma (Fig. 9B) that neuronal tubulin staining was difficult to record, but corneal epithelial nerve segments were seen in both the basal and anterior stroma (Fig. 9B). There were many SOX10 mRNA-expressing cells in the iris, some of which (Fig. 9C, small white arrows) were immediately adjacent to neuronal tubulin–stained cell segments (Fig. 9C, large white arrows), whereas other SOX10-expressing cells did not appear to be intimately associated with neuronal tubulin protein-stained cell segments (Fig. 9C, black arrows). The same relationships occurred between SOX10 mRNA-expressing SCs and neuronal tubulin–stained nerve segments in the anterior corneal stroma and corneal epithelium (Fig. 9D). Similarly, for iris PAX3 (Fig. 9E), S100 (Fig. 9G), and MBP (Fig. 9I) expressions, some SC marker– expressing cells were in immediate contact with neuronal tubulin–stained nerve segments (Figs. 9E, 9G, 9I, small and large white arrows), whereas others were not (Figs. 9E, 9G, 9I, black arrows). These same relationships between PAX3, S100, and MBP SC marker mRNA-expressing cells and neuronal tubulin–stained nerve segments occurred in the corneal epithelial layer (Figs. 9F, 9H, 9J). Notably, in the iris, PAX3, S100, and MBP mRNA-expressing cells occupied the anterior iris, where LUM mRNA-expressing cells also reside, whereas in the corneal stroma these SC marker mRNAs were not expressed in tissue where LUM mRNA was expressed. In addition, in all tissues shown, neuronal class III β-tubulin–stained nerve segments were visible that had no apparent contact with any SC marker– expressing cells.
Evidence of SCs in the cornea has previously been based on microscopic descriptions of cells flanking corneal nerves in adult mammalian corneas, but has not demonstrated any SC marker gene expression in the cornea.39–41 In this study, QPCR showed that genes for transcription factors characteristic of peripheral nervous system SCs, SOX10, PAX3, POU3F1, and EGR2, were expressed in the embryonic chick cornea. PAX3 expression remained higher than both POU3F1 and EGR2 expressions throughout development. In mouse SC development, expression of PAX3 declines after SC transition from immature to mature myelinating SCs,18 but forced elevated expression of PAX3 suppresses myelinating SC expression of MBP and returns SCs to a nonmyelinating state.17 Thus, it is possible that chick corneal SCs do not transition to mature myelinating SCs, because PAX3 expression keeps MBP and other myelin-related gene expression repressed.17 QPCR expression profiles also revealed moderate expression of MBP and S100, and low expression of GFAP, upregulated in mature nonmyelinating SCs,51 and SCMP and MAG, upregulated in mature myelinating SCs.15 These SC transcription factor and marker gene expression profiles support the conclusion that chick embryonic corneal SCs are in an immature SC stage throughout embryonic corneal development (Table 1). Whereas ITGNB1, a marker for mature nonmyelinating SCs,51 was highly expressed in the embryonic cornea, we currently have no evidence that it is expressed specifically in corneal SCs.
The in situ expression profiles of SOX10, PAX3, S100, and MBP coupled with immunochemical profiles of neuronal-specific class III β-tubulin in embryonic chick pericorneal nerve bundles revealed that many cells in pericorneal nerve bundles express SOX10 mRNA, characteristic of all SCs; and/or MBP mRNA, characteristic of mature myelinating SCs.15 Some cells express S100, characteristic of immature as well as mature myelinating and nonmyelinating SCs,17 and few cells express PAX3, characteristic of immature or mature nonmyelinating SCs. Colocalization of these SC marker mRNAs in a pericorneal nerve bundle in patterns where many myelinating and a few immature SCs are expected supports the conclusion that our SOX10, PAX3, S100, and MBP probes detected SCs in corneal and pericorneal tissues. These four mRNAs and neuronal tubulin also colocalized in the corneal epithelium, although often not with such close approximation between SC marker-expressing cells and neuronal tubulin–identified nerve segments as was seen in pericorneal nerve bundles. Outside the central nervous system, SOX10 expression is restricted to neural crest– derived SCs and melanocytes,52 and there are no melanocytes in the cornea. During embryogenesis, PAX3 expression is found in trunk skeletal muscle progenitor cells,53 ectodermal placode cells that form trigeminal ganglion ophthalmic lobe ectodermal placode– derived nerves,54 optic cup retinal pigment epithelium and optic stalk cells,55 and melanocyte stem cells,55 as well as in craniofacial nonmyelinating SCs cells.17 Again, there are no melanocytes or skeletal muscle cells in the cornea, and ectodermal placode– derived nerves do not contribute to corneal nerves.1 Thus SOX10- and PAX3-expressing cells in the cornea could only be SCs.
On the other hand, in addition to their expression in SCs, S100B is expressed in developing skeletal muscle cells,56 chondrocytes,57 adipocytes,58 melanocytes,59 some brain lymphocytes,59 and Langerhans cells,60 and MBP is expressed in thymocytes61 and other hemopoietic progenitors62 in Golli forms that are splice variants of the same gene that encodes myelin-specific MBP mRNA.63 Golli mRNAs contain three exons unique to Golli-MBP, spliced in-frame to exon 1b plus part of intron 1 of the myelin-MBP sequence (HMBP), or to exons 1b-7 of the myelin-MBP sequence (MBP2). No complete chick Golli-MBP protein coding sequences are currently in the GenBank database, but there is an EST (expressed sequence tab) containing a Golli-specific sequence and a few 3′ bases identical with the most 5′ bases of chick myelin-MBP 5′ UTR. If chick Golli-MBPs are similar to mouse Golli-MBPs, our chick MBP QPCR- and in situ primers exclude all HMBP sequences, but could be detecting MBP2 mRNA, since mouse MBP2 exon 1b-7 mRNA is alternatively spiced in ways identical with mouse myelin-MBP mRNA,62,63 and shares the same 3′ UTR region. Whereas skeletal muscle cells, chondrocytes, adipocytes, and melanocytes are not present in the cornea, the human cornea has recently been shown to have some Langerhans cells in its epithelium, some potential antigen-presenting dendritic cells in its anterior stroma, and some macrophages in its posterior stroma.64 Our in situ studies showed no S100 or MBP expression in the anterior or posterior stroma, but did show expression of S100 and MBP mRNAs in the epithelial cell layer where Langerhans cells have been identified. Thus, it is possible that in the corneal epithelium some S100- and MBP2-expressing cells could be Langerhans cells. Notably, however, another lymphoid cell– expressed gene observed in this study, MPZL1, was expressed primarily in the basal epithelium (Fig. 5I), whereas S100 and MBP mRNAs, as well as SOX10 and PAX3 mRNAs, were expressed strongly in the anterior epithelium (Figs. 8I, 8L, 8O, 8R), suggesting that S100 and MBP probes also detect SCs in the corneal epithelium.
S100, SCMP, and MAG mRNAs were expressed in the cornea, but were not translated into enough protein to be detected immunocytochemically in the central cornea, whereas their protein products were detected in the pericorneal nerves that defasciculate into corneal nerves. These observations suggest that there may be translational regulation of the expression of these mRNAs, as well as that of MBP, in embryonic corneal epithelial SCs, where S100 and MBP mRNAs have been localized. Mechanisms for SC posttranslational regulation have yet to be delineated. Intriguingly, S100-, MBP-, and PAX3-expressing cells were identified in the iris, too, where nerves become myelinated and contain MBP protein.65 Notably, SOX10-, PAX3-, S100-, and MBP-expressing cells cohabited in anterior iris tissue where LUM mRNA was expressed, but where lumican core protein is not posttranslationally modified by addition of sulfated KS chains.10 In contrast, SCs within the sulfated KS-rich E14 corneal stroma expressed SOX10, but not PAX3, S100, or MBP mRNAs. It is possible that extracellular matrix sulfated KS chains somehow repress downstream SC marker transcription in SOX10-expressing SCs in the corneal stroma. Mechanisms for glycosaminoglycan regulation of SC transcription remain to be resolved, but chondroitin/dermatan sulfate66 and heparan sulfate67 proteoglycans have been shown to influence SC migration, myelination, and peripheral sensory axon regeneration.
In both the E14 iris and the E14 cornea, some SC marker expressing cells appeared to be in contact with nerve segments, whereas others appeared not to be, although it is possible that the nerves they contacted were just not exposed in the corneal section viewed. Nevertheless, in the iris, close nerve-SC interaction developed sufficiently to allow nerves to become myelinated, whereas in the corneal stroma and epithelium nerve myelination did not occur. It is possible that something in the cornea prevents the necessary close adhesion between SCs and nerves that myelination demands. CADMs are a family of cell-adhesion molecules.68 Typically CADM1 and -3 are expressed by nerve cells, CADM1 and -4 are expressed by SCs, CADM4 expression is upregulated in SCs when they begin to myelinate, and heterophilic interaction between nerve axon-CADM3 and SC-CADM4 mediates an axon-SC interaction that is essential for peripheral nervous system myelination.27,69 In the developing cornea, CADM4 expression remained the lowest of the CADM expressions throughout corneal development, apparently not high enough to support myelination. Similarly, in the developing chick cornea moderate BACE1 expression did not increase during development, supporting further the conclusion that developing chick corneal SCs remain in an immature SC stage.
SLIT2 expression in embryonic chick anterior corneal stroma and epithelium places it in the region of the cornea where nerves extend and branch during chick embryonic development.5,6 Expression of SLIT2 has also been reported in mouse embryo corneal epithelium.70 Although classically thought to act as chemorepellant axon guidance factors, in vertebrates SLIT/ROBO interactions are also positive regulators for peripheral nervous system sensory axon extension and branching, especially in trigeminal ganglion axon branching in the embryonic head,71 depending on which extracellular matrix component binds them.72 In Drosophila, SLITs bound by heparan sulfate sidechain-bearing core protein Glypican-1 are transformed into nerve elongation/branching–stimulating forms.73 SLIT2 and KSPG core protein mRNAs42 are highly expressed in E9 to E14 anterior embryonic chick corneal stromal cells, before accumulation of highly sulfated KS in the extracellular matrix advances from the posterior into the anterior corneal stroma on E14 to E16.10 It is attractive to imagine that KSPG core proteins act as positive corneal nerve extension/branching chemoregulators during early corneal nerve development by binding with SLIT2 protein, since each corneal sensory nerve that enters the anterior cornea supports from 200 to 3000 epithelial nerve endings.56 Further studies on the binding properties of corneal KSPG core proteins with SLITs will be necessary to better understand SLIT function in corneal nerve branching.
Corneal expression of the NTN chemoattractant receptor gene NEO1, the most ubiquitously expressed NTN receptor outside the central nervous system,74 was high throughout corneal development, consistent with the possibility that corneal NTNs also may be acting through the NEO1 receptor to attract corneal nerve extension. Alternatively, NTNs are expressed in developing epithelia,34 where they function in epithelial cell adhesion and migration by interacting with the integrins α6β4 and α3β1.34 In the developing chick cornea, expressions of ITGNA6 and ITGNB4 rose coordinately from E9 to E20, suggesting that α6β4 is present in sufficient concentration to interact with corneal NTNs to play a role in corneal epithelial cell adhesion and migration. Although ITGNB1 expression was approximately 100-fold higher than ITGNA3 expression, their expression patterns also paralleled one another during corneal development, suggesting that α3β1 could also function in corneal epithelial cell adhesion and migration, with the additional ITGNB1 participating in other as yet undefined corneal roles. In addition, adhesion/migration of lymphocytes or epithelial cells in corneal epithelium may also be facilitated by MPZL1, expressed in mesenchymal and hemopoietic cells in mice,49 and MPZL3, expressed in epidermal keratinocytes in mice,50 as these genes have not been shown to be expressed by SCs. MPZL1, the more highly expressed in embryonic chick cornea, was expressed chiefly in the peripheral corneal basal epithelial layer, where it could facilitate cell migration on the basement membrane underlying the epithelial layer.49
SCs are thought to stimulate nerve regeneration by positive signals that are independent of providing a tract for the nerves to regrow along.15,36 Human stromal nerves are ensheathed by thin SC extensions suggestive of immature SCs,75 and chick embryonic SCs show gene expression profiles characteristic of immature SCs, suggesting that they are in an appropriate developmental stage to stimulate corneal nerve regeneration after hatching. Mechanisms to temporarily stimulate SC development in posttransplantation anterior corneas or post-LASIK flaps may improve posttransplantation or post-LASIK nerve regeneration.
Supported by National Institutes of Health EY000952 (GWC); the Howard Hughes Medical Institute Biological Sciences Education Program HHMI 52003734 (MA); and the Terry C. Johnson Center for Basic Cancer Research at Kansas State University (MA, MP-S).
Disclosure: A.H. Conrad, None; M. Albrecht, None; M. Pettit-Scott, None; G.W. Conrad, None