Innervation of the Chick Cornea Analyzed in Vitro

Innervation of the Chick Cornea Analyzed In Vitro

Neil D. Clarke and James A. Bee

Purpose. During the early stages (embryonic day 3 [E3]) of avian corneal development, nerve fibers extend from the trigeminal ganglion to the corneal limbus. On Ell, these nerve fibers enter the cornea and extend through the secondary stroma to begin innervation of the epithelium on El 3. This process of innervation is concomitant with the cornea's dehydration and transition from opacity to transparency; thus, suggesting a link between innervation and the attainment of corneal function. This investigation attempts to ascertain whether the developing cornea can support its innervation in vitro and whether there is a possible developmental interrelationship between corneal innervation and dehydration, with the associated transition from opacity to transparency. Methods. Isolated corneas from either E8 or E14 chicks were co-cultured with E8 dorsal root ganglia. After 4 days of culture, innervation was visualized by silver staining and immunohistochemistry. Changes in corneal composition and organization associated with this innervation in vitro were analyzed by measuring changes in specific hydration, thickness and compaction, and incorporation of [35S] sulfate into glycosaminoglycans during co-culture. Results. The E8 and E14 corneas support extensive innervation in vitro. Developing nerve fibers extend through the secondary stroma to innervate the epithelium. In vitro innervation of E8, but not E14, corneas was associated with a decrease in corneal specific hydration, whereas control corneas (without dorsal root ganglia) failed to show any such changes. E8 corneas also showed a significant increase in compaction when innervated in vitro. Corneal innervation in vitro did not significantly change the overall incorporation of [35S] sulfate into glycosaminoglycans. Furthermore, incorporation of [35S] sulfate into corneal sulfated glycosaminoglycans (sGAG) is not influenced by either the number of nerve fibers innervating the cornea or nerve growth factor (NGF). In addition, the distribution of staining of the corneal glycosaminoglycans, chondroitin sulfate and keratan sulfate, and peanut agglutinin- binding epitopes, suggests that these molecules are not associated with inhibition of axonal development. Conclusions. The in vitro system described here is a useful model to understand the process of corneal development. Co-culture has shown that corneal innervation promotes the process of dehydration, which is dependent on the age of the cornea. However, other functionally related refinements necessary for transparency—notably proteoglycan synthesis—may not be linked to innervation or NGF production. The authors conclude that the development of transparency is dependent on corneal innervation, though not exclusively, and that other controlling factors also are required. Invest Ophthalmol Vis Sci. 1996;37:1761-1771.

.During avian embryogenesis, corneal development serves as the substrate onto which cells derived from begins when the prospective corneal epithelium syn- the neural crest migrate to establish the corneal endo- thesizes and assembles the primary stroma on embry- thelium on E4.4 Thereafter, the corneal endothelium onic day 3 (E3). The primary stroma is composed of synthesizes hyaluronate,5 causing the primary stroma collagen types I, II,1 and IX2 and tenascin,3 and it to swell.6 Later, this is invaded by a second population of neural crest-derived cells that differentiate into cor- 4 From the Department of Veterinary Basic Sciences, The Royal Veterinary College, neal fibroblasts (E5). The cornea increases in thick- Royal College Street, London, NW1 OTU, United Kingdom. Submitted for publication September 13, 1995; revised February 21, 1996; accepted ness as the fibroblasts synthesize and assemble the April 12, 1996. secondary corneal stroma, an elaborate and very Profmetnry interest category: N. highly organized matrix composed of collagen types Reprint requests: Neil D. Clarke, Department of Veterinary Basic Sciences, The Royal 7 8>9 10 Veterinary College, Royal College Street, London, NW1 OTU, United Kingdom. I, V, and VI, " together with a range of proteogly-

Downloaded from iovs.arvojournals.org on 09/30/2021 1762 Investigative Ophthalmology 8c Visual Science, August 1996, Vol. 37, No. 9

cans, the most abundant of which is keratan sulfate.12 whether innervation is associated temporarily with By E10, the cornea has achieved its maximum thick- functionally appropriate changes in corneal structure. ness and is relatively opaque. At this stage of development, hyaluronidase is synthesized and released into the corneal stroma.5 Subsequendy, between E10 and MATERIALS AND METHODS E16, corneal thickness and specific hydration de- Tissue Preparation and Culture crease, and relative transparency increases to reach adult levels by day 18.l3 Fertilized White Leghorn (Gallus domesticus) chicken Concomitandy with diese significant changes in eggs, (Poyndon Egg Farm, Waltham Cross, UK) were corneal structure (E10 to E18), there is extensive sen- incubated at 38°C ± 0.5°C. Embryos were aged by 19 sory innervation by the trigeminal ganglion. However, reference to Hamburger and Hamilton, and all ex- the role of corneal innervation in the processes of periments were conducted according to the ARVO corneal structural development remains unknown. Statement for the Use of Animals in Ophthalmic and Corneal innervation by the trigeminal ganglion oc- Vision Research. After decapitation, E8 and E14 em- curs in a pattern that is highly conserved. Initially (phase bryos were dissected free of extraembryonic mem- 1, from E5 to E10), numerous nerve fibers extend toward branes in cold, calcium-free, and magnesium-free Tyr- the ventro-temporal aspect of the cornea. These prospec- ode's solution, pH 7.4. The anterior segment of die tive corneal nerves do not enter the cornea directly; they eye was excised, and the iris, retina, and lens were extend dorsally and ventrally around it to establish the peeled away from the cornea and its contiguous tis- perilimbal ring.14 This "deflection" of nerve fibers has sues. The cornea was dissected free of surrounding been proposed to occur because of the associated high tissues widi fine iridectomy scissors. rate of corneal oppositional growth.15 On completion of Dorsal root ganglia (DRG) were dissected from the perilimbal ring on E10 and the associated reduction E8 chick embryos. Embryos were decapitated, eviscer- in corneal growth rate, numerous nerve fibers extend ated, and divided lengthwise along die spinal column. radially into the mid-stroma of the cornea (beginning The DRG were then dissected from the exposed sur- phase 2 of innervation).14 Intracorneal nerve fibers ex- face with watchmakers forceps and cleaned of adher- tend through the secondary stroma to begin innervation ent tissue and nerve fibers with tungsten needles. Be- of the overlying epithelium on El 2, which progresses and fore the establishment of cultures, isolated corneas subsequently is completed by E18.14 During this second and DRG were stored in Tyrode's solution at 37°C, in phase, (as oudined above) the cornea concomitantly 5% QOZ in air, for as long as 2 to 3 hours. changes its composition and organization; initially Subsequendy, all tissues were cultured on semi- opaque, the cornea begins to dehydrate on Ell and solid gels composed of 0.5% agarose (Sigma Chemi- achieves adult levels of transparency by El 8. cal, Poole, UK) containing Ham's F12 nutrient me- Thus, innervation of the embryonic cornea occurs dium supplemented with 10% fetal calf serum and over the same developmental period as its composi- antibiotics, (Flow Laboratories, Thames, UK). Cor- tion and organization are refined, ultimately to confer neas were placed with the endothelium facing the gel its mature function. The temporal correlation be- and a single ganglion pipetted direcdy adjacent to the tween innervation and structural change suggests an cornea in triple-vented 30 mm tissue culture plastic important developmental role for corneal nerves, in dishes (Flow Laboratories), each containing 1 ml of addition to their obvious sensory role in later life. semisolid 0.5% agarose:medium. Excess fluid was aspi- The possible interrelationship between innerva- rated to facilitate close apposition between the gan- tion and changes in extracellular matrix structural glion and the cornea. Once established, cultures of components has been alluded to previously.lb It has E8 and E14 corneas, with or without E8 DRG, were been suggested that certain extracellular matrix mole- maintained for as long as 6 days, and the analyses cules act as spatial barriers to nerve fiber extension in oudined below were carried out at daily intervals. the peripheral nervous system.1718 These components Histologic Procedures include the glycosaminoglycans, keratan and chondroitin sulfate, as well as molecules that bind to the Wholemount Staining. Nerve fiber extension from lectin, peanut agglutinin. DRG into the cornea was visualized by the modified To investigate the developmental interrelation- Bodian staining method described by Lewis.20 Briefly, ship between innervation and functionally related corneas were fixed in absolute ethanol:water:37% changes in corneal structure, we have developed a formaldehyde:glacial acetic acid, 75:15:5:5 vol/vol, for system that reproduces corneal innervation in vitro. 48 hours and incubated overnight at 37°C in the dark Using this system, we also examined whether nerve in a 1% aqueous solution of Protargol-S (Sterwin fiber extension appears to be limited by components Chemicals, New York, NY) containing copper. There- of the proposed extracellular matrix barriers and after, corneas were reduced in 1% hydroqui-

Downloaded from iovs.arvojournals.org on 09/30/2021 Innervation In Vitro 1763

none:7.5% anhydrous sodium sulfite, washed in three after removing excess fluid with filter paper. They changes of water, incubated in acidified 1% gold chlo- were then incubated at 140°C for 1 hour. Specific hy- ride, and rinsed in two changes of water (each for 1 dration was calculated according to the equation (wet hour at 4°C with agitation). Corneas were incubated weight — dry weight/dry weight)M with n = 5. in fresh alkaline 5% sodium thiosulfate, rinsed in water, and dehydrated. Finally, corneas were cleared in Incorporation of [35S] Sulfate Into Sulfated toluene, mounted in DPX (Merck, Lutternorth, UK), Glycosaminoglycans and photographed using an inverted Olympus (Lon- Incorporation of [35S] sulfate into glycosaminoglycans don, UK) photomicroscope. (sGAG) was determined in corneas that had been cul- Corneal Thickness-Compaction Analysis. Corneal tured in the presence or absence of DRG or NGF thickness after culture was analyzed by width measure- at 0 to 100 ng/ml (Sera Lab, Crawley Down, UK). ment of paraffin wax-embedded corneas. Cultured Triplicate cultures were used, and, in all experiments, corneas were fixed in formalin buffered saline over- the culture gel contained 5 //Ci/ml carrier-free [3r>S] night, then slowly were dehydrated through a graded sulfuric acid (Amersham International, Amersham, series of alcohols. Corneas were prepared for wax em- UK). bedding after a 4-hour incubation hours in a 50:50 At various times (from 24 to 144 hours), tissues mix of CNP30:hot wax. After they were mounted in were removed, and 250 fig of whale- and shark-mixed wax, corneas were cut serially at 8 fxm, mounted on isomer chondroitin sulfate (Sigma) and 250 /ig hu- slides, and subsequently stained with hematoxylin and man umbilical cord hyaluronan (Sigma) were added eosin. Two measurements were taken—actual corneal to each sample. Concentrated Trizma HC1 was added thickness (mm) and corneal fibroblast number. For to give a final concentration of 0.2 M. Subsequently, thickness, the mean of at least five sections per cornea samples were heated (100°C for 5 minutes), cooled was determined. Corneal fibroblast number per cor- (on ice), and incubated (for 48 hours at 50°C) with nea was determined from a mean of at least 10 counts preincubated Streptomyces griseus pronase (0.18 mg/ml; of a defined region anterior to the endothelium. Boehringer Mannheim, Lewes, UK) in the presence Immunostaining Against GAP 43. Nerve fibers were 23 of CaCl2 and ethanol. Fresh enzyme was added after visualized in sections of cornea using immunostaining 24 hours. After digestion with pronase, diluted sam- with a polyclonal antibody directed against growth- ples were applied to freshly regenerated columns of associated protein (GAP43). This antibody was pre- Dowex 1-Cl" resin (0.5 X 6 cm; AG1-X2 [200-400 pared by Dr. Diana Moss and coworkers, and full de- mesh; Bio-Rad Laboratories, Watford, UK]) and subse- 2122 tails of its preparation can be found. In brief, rab- quently were eluted with an increasing concentration bit antisera to the 3D5 antigen (a neuronal membrane of NaCl.24"26 Unincorporated isotope was eluted with skeleton protein) were shown to precipitate the trans- six 10-ml volumes of 0.2 M NaCl, followed by 10 ml lation product of chick GAP-43 cDNA. After overnight each of 0.3, 0.4, 0.5, and 1 M NaCl, and by three 10- fixation at 4°C in 4% paraformaldehyde in phosphate- ml volumes of 2 M NaCl. Fractions were collected, buffered saline (PBS), pH 7.4, corneas were incubated and 0.5 ml aliquots were taken and analyzed for [35S] at 4°C for 24 hours each in 5% and then 15% sucrose radioactivity using an LKB 1214 Rackbeta Liquid Scin- in PBS, immersed in OCT compound, frozen in Arc- tillation Counter and Aqueous Counting Solution ton fluid over liquid nitrogen, and stored at — 70°C (Amersham International). Total incorporation was until used. After a 1-hour incubation with anti-GAP- expressed as cpm per cornea or explant. For each 43 antibody, sections were washed twice with PBS and experiment, results are expressed as means ± stan- incubated with a rhodamine-conjugated swine anti- dard errors (w = 3). rabbit antibody (1:40; Dako, High Wycombe, UK). Extracellular Matrix Epitopes Double Immunostaining. Sections were immunostained with GAP-43 as above RESULTS and subsequently by the schedule shown in Table 1. All immunostained sections were finally washed with Innervation of Isolated Corneas In Vitro 2 X PBS and mounted under coverslips with glycerol: After 4 days of co-culture, numerous nerve fibers ex- PBS (9:1, vol/vol) containing 2.5% DABCO (diazobi- tended from the E8 DRG (*) into and throughout the cyclo-(2,2,2)-octane; Sigma) and 0.1% sodium azide cornea (Fig. 1A). In wholemount, nerve fibers ex- at pH 8 to pH 9. Stained sections were viewed and tended extensively and randomly throughout the cen- photographed on an Olympus BH-T photomicro- tral region of the cornea. In addition, large fascicles scope equipped for epifluorescence. frequently were observed around the periphery of the Specific Hydration cornea (Fig. 1A, arrows). Similarly, but perhaps not Freshly isolated corneas and corneas cultured alone so extensively, nerve fibers extended from E8 DRG or with a DRG for 24, 48, 72, or 96 hours were weighed throughout the E14 cornea (Fig. IB). In these sam-

Downloaded from iovs.arvojournals.org on 09/30/2021 1764 Investigative Ophthalmology & Visual Science, August 1996, Vol. 37, No. 9

TABLE l. Immunostaining Protocol Antibody A - Lectin Name Epitope Recognized Raised in: Dilution: PBS Secondary Antibody, 1:40 GAP-43* GAP-43 neuronal membrane Rabbit (polyclonal) 1:250 Swine anti-rabbit TRITC protein conjugated! MZ15J Keratan sulfate Mouse (monoclonal) 1:200 Goat anti-mouse FITC conjugated§ CS-56§ Chondroitin-4 and -6 sulfate Mouse (monoclonal) 1:200 Goat anti-mouse FITC conjugated§ Peanut agglutinin Galactose-N- Plant species Arachis 10 /ig/ml Goat anti-PNA (PNA)§ acetylgalactosamine hypogaea (peanut) residues Anti-goat FITC Goat antibody bound to PNA Rabbit (polyclonal) 1:40 conjugated|| All incubations were for 30 minutes, and after each stage sections were washed twice with phosphate-buffered saline (PBS). * Kindly donated by Dr. Diana Moss, University of Liverpool. t Dako Laboratories. X Kindly donated by Fiona Watt, ICRF, London. This antibody was prepared by Dr. Watt, and its preparation is described in reference 23. § Sigma Chemical.™ || Vector Labs. Sections first were blocked with bovine serum albumin and nonimmune rabbit serum in HEPES-buffered saline (10 mM HEPES, 0.15 M NaCI, and 0.1 M CaCl;.). PNA-treated sections subsequently were visualized with a fluorescein isothiocyanate (FITC) antibody described in row 5.

pies, a larger number of delicate fibers extended di- more superficial levels of the secondary stroma and of- rectly from the ganglion across the cornea, and thick- ten was oriented toward the epithelium (Fig. 2). Nerve ened fascicles were not observed. fiber extension through the posterior secondary stroma In sections of co-cultured E8 cornea and E8 DRG toward the endothelium was not observed. maintained for 24 hours and stained immunohistochem- ically with GAP-43 antibody, nerve fibers were observed Specific Hydration of the Cultured Cornea within the secondary stroma and epithelium (Fig. 2). Results from Figure 3A show that when E8 corneas Nerve fiber extension was restricted to the middle and were cultured in isolation, there was no overall change

FIGURE l. Wholemounts of (A) an E8 cornea and (B) an E14 cornea, both after 4 days in culture with an E8 dorsal root ganglion. In both, axons extend from the DRG (*) into the corneal stroma. In E8 corneas, axonal streaming around the corneal periphery is prominent (arrows). Bars = 0.5 mm.

Downloaded from iovs.arvojournals.org on 09/30/2021 Innervation In Vitro 1765

between corneas cultured alone or in combination with DRG. In parallel experiments (Fig. 4B), E14 corneas were cultured alone and similarly showed a gradual but consistent rise in 35S incorporation into glycosami- noglycan—from 5535 cpm to 22097 cpm after 96 hours, representing a 4-fold increase (P < 0.01, n — 3). When cultured in combination with E8 DRG, incorporation by the E14 cornea again increased, from 6744 cpm to 22097 cpm after 96 hours, representing a 3.3-fold increase (P< 0.01, n = 3). Over the culture period, there was no significant difference between E14 corneas cultured alone or in combination with E8 DRG. However, incorporation by the El4 cornea FIGURE 2. Section of E8 cornea cultured with E8 dorsal root ganglia for 24 hours, immunostained with GAP-43 and visu- did appear to be significantly elevated by co-culture alized with a secondary rhodamine-conjugated antibody. with an E8 DRG after 96 hours (P < 0.05, n = 3). Nerves (white specks) extend from the DRG (not visible on All data represented in Figure 4 are per explant (an this section) through the mid-stromal region toward the explant is a pair of corneas or recombinants, or six epithelium (e). Nerves innervating the epithelium are also DRG); all explants were taken in triplicate. evident (arrows). Bar =100 fim. Isolated E8 and E14 corneas were cultured either

30 in specific hydration (SH), which was the ratio of cor- p<0 02 025 neal water content to corneal dry weight (no units); 25 - analysis of variance indicated that from an SH value

of 14.5 ± 0.445 (SE) at day 0, there was a significant 20 - sj decrease in the first daily analysis of SH (P < 0.01, n s ~~~- I 1 = 5), and subsequently there were no differences from rr i2 / Q "c \ / the starting day 0 value. Conversely, when co-cultured 1 5 " with an E8 DRG, the corneal SH decreased to V4.5 its u. *= 10 • r" \ initial value over the 96-hour culture period (14.5 ± o -9 LLJ < 0.445 to 3.21 ± 0.58; P = 0.01; n = 5). Notably, the 0- — CO c decrease in SH between control corneas and corneas 3 —o--- Conirol cultured with DRG was statistically significant by 3 days t Cornea+DRG (n=5) (P < 0.02) and 4 days (P < 0.025) (unpaired t-test, n = 5). 24 46 72 9G The E14 corneas cultured either alone or together CULTURE DURATION (Hours) with E8 DRG tended to swell with water (Fig. 3B). B 1 6 From a starting value of 5.83 ± 0.78, this increased after 96 hours to X2.3 to 13.48 ± 1.11 (P< 0.01) and to X2.2 to 12.89 ± 1.26 (P < 0.01) in the absence or presence of DRG, respectively (n = 4). There were no 1 2 significant differences between the control and exper- O imental SH values over the culture period. is- 1 0 Q c 39 > 3 Incorporation of [ S] Sulfate Into Sulfated X >, Glycosaminoglycans by the Cultured Cornea O ^ 3F a € 6 When isolated E8 corneas were cultured alone, 'S LU < Q_ "— (Si — D--- Conirol incorporation increased during the culture period; 4 •— Cornea+DRG 1840 cpm to 3182 cpm after 144 hours, representing (n = 4) a 1.7-fold increase (P < 0.05, n = 3) (Fig. 4A). The incorporation by E8 corneas cultured in combination ?A 48 72 96 with E8 DRG also rises during the culture period; 1915 CULTURE DURATION (Hours) cpm to 3211 cpm after 144 hours, presenting a 1.7- FIGURE 3. Graphs show the effect of innervation in vitro on fold increase (P < 0.05, n = 3). However, at each of corneal specific hydration. E8 (A) and E14 (B) were cul- the times analyzed, there was no significant difference tured either alone or together with an E8 DRG.

Downloaded from iovs.arvojournals.org on 09/30/2021 1766 Investigative Ophthalmology 8c Visual Science, August 1996, Vol. 37, No. 9

6000 endothelium is analyzed, it can be seen that after 2

H Cornea days of culture, the difference in the cell density is 5000 - 0 Ganglion significantly greater in the corneas cultured with a m Recombinanl DRG than in those cultured alone. 4000 - Distribution of Chondroitin Sulfate, Keratan 1 Sulfate, and Peanut Agglutinin Binding in the 3000 . I Cultured Cornea CO The presence of chondroitin sulfate was indicated us- CO 200° ing immunohistochemistry. There was staining throughout the secondary stroma, whereas neither the corneal epithelium nor endothelium exhibited CS-56 Q. d 3000 48 72 96 -/ h 0 ng/ml NGF CULTURE DURATION < 10 ng/ml NGF (Hours) LLJ • 50 ng/ml NGF

DC O B o 2000- DC 111 Q_ 111 30000 - z en en 1000- to CO 20000 - D 00 a. d

00 10000 -

144

CL CULTURE DURATION (Hours) d 48 72 CULTURE DURATION B 10000 (Hours) UJ 0 ng/ml NGF FIGURE 4. Graphs show the incorporation of 35S sulfate into 10 ng/ml NGF 8000 • 50 ng/ml NGF

sulfated glycosaminoglycans of (A) E8 and (B) El4 corneas CORr< cultured alone or with dorsal root ganglia (recombinant). 100 ng/ml NGF PER UJ 6000 i— < alone or in the presence of increasing concentrations LL of NGF (Fig. 5). Overall, compared with controls, when present at a final concentration of 10, 50, or 4000 100 ng/ml, NGF has no significant effect on the incor-

35 [35S] SUL poration of [ S] sulfate into sulfated glycosaminogly- . 2000

cans by either the E8 or the E14 cornea. P.M 6 Corneal Thickness and Compaction Within the Cultured Cornea Figure 6 shows the thickness of corneas and the total 96 144 cell number within a small, denned area anterior to CULTURE DURATION the endothelium of corneas cultured alone or with a (Hours) 35 DRG for 2 days. In Figure 6A, it can be seen that FIGURE 5. Graphs show the incorporation of S sulfate into although corneas cultured with a DRG are thinner, it sulfated glycosaminoglycans of (A) E8 and (B) E14 cultured is just outside significance with P = 0.07. However, in either the absence or increasing concentrations of nerve when cell number in a region just anterior to the growth factor.

Downloaded from iovs.arvojournals.org on 09/30/2021 Innervation In Vitro 1767

B 40 p<0.005 i.o- T E, 30 CD ^ O w 0.8- CO s i. - z111 — o cr o 0.6- 20 3 FIGURE 6. Graphs show (A) X 0) 2 thickness and (B) cell num- • • Cornea+DRG I § I ber in a denned region just < 0.4- • Control above the endothelium in E8 zLU ^ o corneas cultured in the pres- cr 10 O ence or absence of a DRG o 0.2- for 48 hours. Measurements were taken from at least four • corneas after dehydration n o- 0 and subsequent 8 /*m paraf- 48 48 fin wax sectioning. CULTURE DURATION (Hours)

staining (Fig. 7A). Co-localization of GAP-43 demon- through the keratan sulfate-rich secondary stroma. strates that numerous nerve fibers have extended These nerve fibers are restricted to the anterior sec- through this chondroitin sulfate-rich environment ondary stroma and are oriented toward and within (Fig. 7B). An elaborate and extensive pattern of nerve the overlying epithelium. fibers is revealed within the corneal epithelium. The distribution of peanut agglutinin binding epi- Similarly, keratan sulfate staining was present topes is shown in an adjacent section (Fig. 8B). Neither throughout the secondary stroma of the cornea after die epithelium nor the endothelium of the cornea is 1 day in culture (Fig. 8A), whereas neither epithelium stained with this lectin. In contrast, peanut agglutinin nor endothelium exhibited staining. Co-localization stains the secondary stroma; staining is most prominent of GAP-43 with keratan sulfate staining demonstrates beneath the epithelium and progressively decreases in that numerous nerve fibers extended into and intensity toward the endothelium. Bowman's and Des-

FIGURE7. A section of an E8 cornea cultured with an E8 dorsal root ganglion for 3 days, stained for (A) chondroitin sulfate (CS-56) and (B) axons (GAP-43). Co-localization of axons (white specks) and high chondroitin sulfate expression is prevalent (B) Further evidence of extensive epithe- lial innervation is seen {arrows). Bars = 50 /j,m.

Downloaded from iovs.arvojournals.org on 09/30/2021 1768 Investigative Ophthalmology 8c Visual Science, August 1996, Vol. 37, No. 9

FIGURE 8. Sections of an E8 cornea with an E8 DRG after 1 day in culture. (A) Axons (brighter specks within the corneal stroma, CAP43) extend through the keratan sulfate (MZ-15)-rich cornea! stroma (lighter background around the corneal fibroblasts) toward the overlying epithelium (e). Bar = 50 /xm. (B) Axons (brighter speck and lines, GAP43) extend into a region rich in peanut agglutinin binding (lighter background) beneath the corneal epidielium (e). Bowman's (b) and Descemet's (d) membranes exhibit staining. The molecule (s) that peanut agglutinin binds to appear (s) to have no inhibitory effect on axonal extension. Bar = 50 pm.

cemet's membranes exhibit relatively strong staining of DRG in combination culture. After co-culture, E8 and this lectin. Co-localization of GAP-43 demonstrates that El4 corneas exhibited large numbers of nerve fibers numerous nerve fibers have extended into and through extending direcdy from the DRG into and throughout the microenvironment stained with peanut agglutinin. the cornea. (Fig. 1). These nerve fibers appeared to Indeed, once within the secondary stroma, nerve fibers enter the cornea through the mid-level of the second- are oriented into and through areas in which staining ary stroma before extending toward and through the with this lectin is most intense. overlying corneal epithelium (Fig. 2). However, nerve fiber extension through the posterior secondary stroma toward the corneal endothelium was not ob- DISCUSSION served. We have demonstrated innervation of E8 cor- We have shown that isolated E8 and El4 chick corneas neas in vitro. This suggests that the developing cornea promote and support nerve fiber extension from E8 is permissive to nerve fiber extension at an earlier

Downloaded from iovs.arvojournals.org on 09/30/2021 Innervation In Vitro 1769

stage of its development in vivo, that is, at least day 8 neas after 4 days of culture is likely to be an in vitro of development. The reasons such innervation does phenomenon that has no relevance to the in vivo situa- not occur at this earlier stage in vivo are not clear; tion. In this case, the innervation that occurs in vitro however, it is possible that innervation is limited by takes place 8 days after the corneas would have been the fast corneal growth rate observed in vivo, between innervated in vivo. E8 and E10,lf) but not in vitro. Hence, axons in vivo Preliminary experiments indicate that the cornea are unable to penetrate the matrix and cells of the produces a number of neurotrophins, including NGF, limbus at this stage of development and stream dor- during its development; hence, any observed effects sally and ventrally around the corneal periphery. in vitro may be mediated by NGF (nerve fiber exten- We have observed significant differences in the sion from the isolated E8 DRG is dependent on the levels of hydration between corneas cultured alone presence of these neurotrophic factors, most notably and those with DRG. Previous work27 has shown that NGF). However, our results suggest that this may not corneas tend to swell in culture with any extraneous be the case, as can be seen in Figure 5. Both E8 and water present. We have shown that there is a signifi- E14 corneas cultured with NGF show no overall sig- cant decrease in specific hydration of the E8 cornea nificant changes in GAG synthesis, suggesting that compared to the El4 (Fig. 3) when cultured with a NGF plays no direct role in the normal course of in DRG. Therefore, innervation has a functional role in vivo transparency formation. It is possible that endoge- vitro by promoting corneal dehydration. For El4 cor- nous NGF already was eliciting a maximal response. neas, innervation showed no effect on SH in vitro. Similarly, in data not presented here, an increase in This result reinforces the argument that innervation the number of DRG cultured with individual corneas has a direct influence on the normal course of dehy- also showed no effect on [35S] sulfate incorporation dration in vivo. For the E14 corneas in culture, the into GAGs. normal developmental dehydration process already Results of the analysis of the thickness of corneas has been switched on by innervation that previously cultured alone or with a DRG are seen in Figure 6A. occurred in vivo; hence, subsequent experimental in- After 2 days, it can be seen that corneas cultured with nervation showed no effect, and the corneas swelled a DRG are thinner than control corneas alone. How- 28 with water in agreement with previous studies. Cor- ever, it is just outside significance (P = 0.07). When neal dehydration, which occurs much more slowly and cell density is analyzed (Fig. 6B) in a small, defined 13 over a longer period of time in vivo, is probably more region anterior to the endothelium, it can be seen rapid in vitro because of the earlier arrival of nerves. that corneas cultured with a DRG for 2 days show a However, other influences may affect dehydration significant increase in cell density within the region (and the differences observed in vivo and in vitro). above the endothelium. This indicates that the cornea For example, intraocular pressure is absent as are nor- becomes more compact in this region once corneal mal in vivo corneal growth rates. water within the extracellular matrix is expelled We have examined the effect of innervation on through the endothelium.30 It should be noted that sulfated GAG synthesis and find there is no significant the total cell number through the full depth of the change in the total corneal sGAG amounts synthesized corneas was not different in innervated versus nonin- between controls and recombinants (Fig. 4). The nervated corneas, indicating that the increase in cell method we have used is a good index of overall glycos- density was indeed caused by compaction rather than aminoglycan production; in previous studies, (Abra- proliferation. ham L, et al, manuscript submitted, 1996) ,29 in which Finally, we have shown that these proteoglycans both [3H] N-acetyl glucosamine and [35S] sulfuric acid within the cornea, with keratan sulfate and chondroi- were used to measure GAG synthesis, it was shown tin sulfate side chains, did not act as barriers to the that the N-acetyl glucosamine incorporation into GAG axonal extension in our in vitro model (Figs. 7, 8). chains is proportional to the incorporation of radiola- Previous work1718 has suggested that axonal extension beled sulfate molecules into the sulfate epitopes of from the dorsal root ganglia is oriented according to the GAG chains. The production of proteoglycans is barrier molecules along the axonal pathways; these another aspect of corneal transparency development, molecules are keratan sulfate, chondroitin-6-sulfate, of which sGAGs are a requirement. Unpublished data and peanut agglutinin binding epitopes. These mole- (Clarke and Bee, 1994) have shown that disruption of cules are clearly present in the developing chicken the synthesis of specific proteoglycans affects transpar- cornea, and none present themselves as barriers to ency formation and corneal ultrastructure.29 However, axonal advance seen by the co-localization of the ex- unlike the process of dehydration, there appears to be tracellular matrix epitopes and axons labeled by the no relationship between innervation and this aspect antibody to GAP-43. This experimental result is sup- of transparency formation. The significant difference ported by the evidence of corneal innervation in vivo. observed between controls and recombinant E14 cor- It is known that innervation in vivo occurs at the time

Downloaded from iovs.arvojournals.org on 09/30/2021 1770 Investigative Ophthalmology & Visual Science, August 1996, Vol. 37, No. 9

of the appearance of large increases in the amounts of ization of type V collagen in the chick embryo with extracellular keratan sulfate.12 Thus, if keratan sulfate monoclonal antibodies. Collagen Rel Res. 1982; 2:541- were acting as a barrier to axonal advance, innerva- 555. tion—both in vivo and in vitro—would be unlikely to 9. Linsenmayer TF, Fitch JM, Schmid TM, et al. Mono- occur. However, this is not the case. clonal antibodies against chicken type V collagen: Pro- duction, specificity, and use of immunocytochemical In conclusion, these experiments have shown how localization in embryonic cornea and other organs. / the isolated embryonic cornea supports extensive in- Cell Biol. 1983;96:124-132. nervation by dorsal root ganglia from E8 chicks in 10. Linsenmayer TF, Bruns RR, Mentzer A, Mayne R. Type vitro. We have established a useful model for corneal VI collagen: Immunohistochemical identification as a development. However, it should be noted that DRG filamentous component of the extracellular matrix of are purely of neural crest origin, whereas the trigemi- the developing avian corneal stroma. Deu Biol. nal ganglion that innervates the cornea in ovo is of 1986;118:425-431. approximately a 50:50 mix of neural crest and epider- 11. Linsenmayer TF, Fitch JM, Birk DE. Heterotypic colla- mal placode-derived neurons.31 We have shown (un- gen fibrils and stabilizing collagens: Controlling ele- published data) that in our culture setup, E4 trigemi- ments in corneal morphogenesis? Ann NY Acad Sri. nal ganglia innervate the E8 and El4 cornea in vitro, 1990;580:143-160. but further work is necessary for an analysis of the 12. Funderburgh JL, Caterson B, Conrad GW. Keratan sulfate proteoglycan during embryonic development exact origin of the neurons that normally innervate of the chicken cornea. Deu Biol. 1986; 116:267-277. the cornea. 13. Coulombre AJ, Coulombre JL. Corneal develop- We have shown that depending on corneal age, ment—the role of the thyroid in dehydration and the innervation plays an important role in corneal trans- development of transparency. Exp Eye Res. 1964; formation, predominantly in the control of dehydra- 3:105-114. tion, which is fundamental to the subsequent appear- 14. Bee JA. The development and pattern of innervation ance of the cornea's mature functional property, of the avian cornea. Deu Biol. 1982;92:5-15. transparency. 15. Neath P, Roche S, Bee JA. Intraocular pressure-dependent and -independent phases of growth of the embry- Key Wards onic chick eye and cornea. 1991; 32:2483-2491. 16. Bee JA, Unruh NC, Sommerfield DC, Conrad GW. corneal dehydration, corneal organ culture, development, Avian corneal innervation—inhibition of ring forma- extracellular matrix, innervation tion by 6-diazo-5-oxo-l-norleucine. Deu Biol. 1982; Acknowledgments 92:123-132. 17. Snow DM, Lemmon V, Carrino DA, Caplan AI, Silver The authors thank Dr. Andrew Pitsillides and Dr. Eleanor J. Sulfated proteoglycans in astroglial barriers inhibit Mackie for their constructive criticism of the manuscript. neurite extension in vitro. Exp Neurol. 1990; 109:111- 130. References 18. Oakley RA, Tosney KW. Peanut agglutinin and chon- 1. von der Mark K, von der Mark H, Timpl R, Trelstad droitin-6-sulfate are molecular markers for tissues that RL. Immunofluorescent localization of collagen types act as barriers to axon advance in the avian embryo. I, II, and III in the embryonic chick eye. Deu Biol. Deu Biol. 1991; 147:187-206. 1977;59:75-85. 19. Hamburger V, Hamilton HL. A series of normal stages 2. Fitch JM, Mentzer A, Mayne R, Linsenmayer TF. Ac- in the development of the chick embryo. / Morphol. quisition of type IX collagen by the developing avian 1951;88:49-92 primary corneal stroma and vitreous. Deu Biol. 20. Lewis J. Pathways of axons in the developing chick 1988; 128:396-405. wing: Evidence against chemo-specific guidance. Zoon. 3. Tucker RP. The distribution of Jl/tenascin and its 1978; 6:175-179. transcript during the development of the avian cor- 21. Allsopp TE, Moss DJ. A developmentally regulated nea. Differentiation. 1991; 48:59-66. chicken neuronal protein associated with cortical cy- 4. Johnston MC, Noden DM, Hazelton JL, Coulombre toskeleton. / Neurosri. 1989; 9:13-24. JL, Coulombre AJ. Origins of avian ocular and perioc- 22. Moss DJ, Fernyhough P, Chapman K, Baizer L, Bray ular tissues. Exp Eye Res. 1979; 29:27-43. D, Allsopp TE. Chicken growth-associated protein 5. Toole BP, Trelstad RL. Hyaluronate production and GAP-43 is tightly bound to the actin-rich neuronal removal during corneal development in the chick. Deu membrane skeleton. /Neurochem. 1990;54:729-736. Biol. 197l;26:28-35. 23. Zanetti M, Ratcliffe A, Watt FM. Two subpopulations 6. Hay ED, Revel JP. Fine Structure of the Developing Avian of differentiated chondrocytes identified with a mono- Cornea. Basel: Karger; 1969. clonal antibody to keratan sulfate. / Cell Biol. 1985; 7. Hay ED, Linsenmayer TF, Trelstad RL, von der Mark 101:53-59. K. Origin and distribution of collagen in the devel- 24. de la Haba G, Holtzer H. Chondroitin sulfate: Inhibi- oping avian cornea. Curr Top Eye Res. 1979; 1:1-35. tion of synthesis by puromycin. Srience. 1965; 8. von der Mark K, Ocalan M. Immunofluorescent local- 149:1263-1265

Downloaded from iovs.arvojournals.org on 09/30/2021 Innervation In Vitro 1771

25. Schiller S, Slover GA, Dorfman A. A method for the 29. Hahn, RA, Birk DE. B-d-Xyloside alters dermatan sul- separation of acid mucopolysaccharides: its applica- fate proteoglycan synthesis and the organization of tion to heparin from the skin of rats. / Biol Chem. the developing avian corneal stroma. Development. 1961; 236:983-987. 1992; 115:383-393. 26. Conrad GW, Dorfman A. Synthesis of sulfated muco- 30. Hodson S, Guggenheim J, Kaila D, Wigham C. Anion polysaccharides by chick corneal fibroblasts in vitro. pumps in ocular tissues. Biochem Soc Trans. 1991; ExpEyeRes. 1974; 18:421-433. 19:849-852. 27. Conrad GW, Hamilton C, Haynes E. Differences in 31. D'Amico-Martel A, Noden DM. An autoradiographic glycosaminoglycans synthesized by fibroblast-like cells analysis of the development of the chick trigeminal from chick cornea, heart and skin. J Biol Chem. ganglion. JEmbryolExp Morphol 1980;55:167-182. 1977; 19:6861-6870. 32. Avnur Z, Geiger B. Immunocytochemical localization 28. Coulombre AJ, Coulombre JL. Corneal development- of native chondrotin-sulfate in tissues and cultured transparency changes during rapid hydration. Am J cells using a specific monoclonal antibody. Cell. Ophthalmol. 1958; 46:276-281. 1984; 38:811-822.

Downloaded from iovs.arvojournals.org on 09/30/2021