The effects of light regimes and hormones on corneal growth in vivo and in organ culture

Abstract

When chicks are exposed to constant light (CL) during growth, their corneas become flatter and lighter in weight, and their anterior segments become shallower than those of chicks exposed to cyclical periods of light and dark. These effects have been correlated with CL suppression of cyclical changes in melatonin levels. The question of whether light directly influences corneal growth (e.g. via cryptochromes in the cornea) or acts remotely via the suppression of the melatonin rhythm has not yet been answered. Retinoic acid (RA), an ubiquitous morphogen, also causes non-functional flattening during corneal growth, but its effect in vivo has not been correlated with light regimes. We wished to characterize and distinguish between hormonal and light effects on corneal growth. We used organ culture to study the direct effects of light regimes, melatonin, and RA, and compared these results with those of parallel in vivo experiments. In this study, eye drops containing melatonin or RA were applied to corneas exposed to CL in vivo or in organ culture, and effects on corneal mass and hydration were measured. We applied a melatonin blocker, luzindole, to chick corneas in normal light/dark conditions to confirm that the observed melatonin effects are mediated at the cell membrane. Anterior chamber depth and refraction in vivo were measured. We found that, during CL exposure, combined application of melatonin and RA eye drops increased the depth of the anterior segment in vivo, (P = 0.003) and interestingly, both also reduced the hyperopia of CL exposure after 2 weeks (P = 0.002), thus partially reversing the effects of CL. RA increased corneal hydration in vivo (P = 0.030) but not in organ culture. Melatonin had no effect on corneal hydration in vivo, but in organ culture, melatonin significantly decreased hydration (P < 0.001). We found no evidence for a direct effect of light on corneal hydration in growing chick corneas in culture. Melatonin is required for normal corneal growth in vivo, and together melatonin and RA, or RA alone, affects the regulation of water content within the chick cornea. Melatonin also affects corneal hydration in vitro, but RA does not.

Introduction

Internal regulation of the eye's growth and proper shaping is adversely affected by continuous light, as first demonstrated in growing chickens (Harrison & McGinnis, 1967). Constant light (CL) causes increased ocular size and weight, decreased corneal diameter, and decreased corneal mass (Wahl et al. 2009). Chicks raised in CL exhibit significant hyperopia after as few as 10 days as a result of corneal flattening, and this is independent of visual experience (Li et al. 1995). Form deprivation has been shown to alter the production of scleral matrix in both chickens and mammals, by increasing production of matrix proteoglycans and causing ocular elongation in the predominantly cartilaginous sclera of the chick (Rada et al. 1992) but by decreasing matrix production in the dense connective tissue sclera of mammals (Norton & Rada, 1995). Adult ocular pathologies, such as retinal tearing, hyperopia, and glaucoma are reported after many months’ exposure to CL in chickens (Lauber et al. 1961; Li et al. 1995).

While many adult tissues and organs exhibit hyper- or hypoplasia and reshaping as a result of environment and usage (e.g. muscles enlarge with exercise, skin thickens with friction, etc.), the adult chicken eye shows little response to environmental influences. Melatonin and continuous light only affect corneal shape during growth in the chicken (Schaeffel et al. 1995; Li & Howland, 2000). Among chicks exposed to CL, melatonin levels are not rhythmic, leveling off between the higher serum concentrations present during normal night and the lower levels during normal day (Li & Howland, 2000). When CL chicks are given melatonin injections or eye drops containing melatonin, CL-effected hyperopia is reduced. When injections or eye drops containing the nonspecific melatonin receptor blocker luzindole are administered to chicks raised in normal cycles of 12 h light and 12 h dark (condition ‘N’), they develop hyperopia (Li & Howland, 2002). These experiments have shown that cycling melatonin levels are required for normal corneal growth.

The tissue response of the cornea to light cycles is not as well studied as that of the sclera, although the change in shape of the cornea in response to light has a dramatic effect on the eye's refractive properties (Li et al. 1995). There is also evidence that peripheral clocks exist in connective tissues (Yagita et al. 2001).

Blue-light sensitivity is universal among most animal and plant tissues, and generates responses such as phototaxis, phototropism, and entrainment of circadian rhythms (Sancar, 2000). Although the superior chiasmatic nucleus (SCN) of vertebrates overrides all local oscillators and coordinates peripheral oscillators via a ‘pacemaker clock’, in the absence of the influence of the SCN, peripheral, cell-autonomous photoresponses have been demonstrated by employing photosensitive proteins known as cryptochromes (Cashmore, 2003). Cryptochromes are related to the photolyase/blue-light photoreceptor family, and when expressed as recombinant proteins they have a 420-nm absorption peak (Sancar, 2000). Because cryptochromes have been demonstrated in the chick retina (Bailey et al. 2004) we considered the possibility that the cornea may be directly sensitive to light. We tested this hypothesis by culturing corneas under different light regimens, as described below.

It has been demonstrated that corneal flattening among CL chicks is prevented by covering either the pineal or one eye for intervals of as little as 4 h during the 24-h day (Li & Howland, 2003). This suggests that melatonin is involved in the regulation of corneal growth, since it is produced by the pineal and by the retina in response to light and dark cycles.

Retinoic acid (RA), an important growth factor, also influences corneal shape as demonstrated in RA receptor knockout experiments in mice (Mark et al. 2006). RA administered by eye drop or feeding produces corneal flattening among chicks raised in normal cycles of light and dark (Napier & Mertz, 2003; McFadden et al. 2006). RA is a signaling molecule that induces morphological and functional differentiation of embryonic stem cells (Marshall et al. 1992) and causes alteration of the embryonic extracellular matrix in the hindbrain (Moro Balbas et al. 1993).

Two ways that melatonin and RA might affect the shaping of the cornea include (i) cyclical stimulation and (or) coordination of stromal cell mitosis and apoptosis, or (ii) alteration of the water content of the corneal stromal matrix. Circulating melatonin levels do influence the circadian rhythm of cells (Cassone, 1998), including epithelial tissues of the cornea. Cell clock mechanisms are found within connective tissue cells and indeed are now thought to exist within all cells (Simmons, 1992; Roenneberg & Merrow, 2003).

However, water is the most significant portion of the cornea's mass and thus it might be expected that alteration in water content of the cornea would affect its shape more significantly than changes in the numbers of stromal cells. The water content within the stroma has been described as either ‘total’ or ‘nonfreezable’ (bound) water. Total water decreases linearly across the stroma from the endothelium to the epithelium, and the concentration of nonfreezable water increases along the same path (Castoro et al. 1988). Water distribution is mirrored by two major stromal glycosaminoglycans (GAGs); keratan sulfate concentration increases with the increase in total water, whereas dermatan sulfate concentration increases with bound water. Alteration in the quantity or distribution of these two proteoglycans will dramatically affect the hydration state of the cornea. The production of proteoglycans in organ culture has been shown to be similar to that found in vivo (Cöster et al. 1983). The production and turnover of dermatan and keratan sulfate is controlled by the corneal keratocytes and can be inhibited in culture by RA (Dahl & Axelsson, 1980).

Because the effect of melatonin and RA on the growth of the cornea is profound, we predicted that melatonin and RA would have an influence on corneal hydration in organ culture. We cultured corneas with melatonin or RA and compared our results with similarly treated corneas in vivo. To eliminate the possibility that light regimen may regulate corneal hydration state in vitro, fellow corneas were cultured in CL or in 12-h cycles of light and dark without hormone supplements, and compared to corneas cultured with hormones. Here we establish a correlation between corneal water content and direct corneal exposure to the hormones melatonin and RA.

Materials and methods

Animal husbandry

Two hundred and sixty-seven White Leghorn chicks (Cornell-K strain) were used in this study. The illumination level in the aviary averaged 700 Lux during the light-on period. Illumination was supplied by fluorescent lamps (Sylvania 40 W, Cool White). Chicks were raised in temperature-controlled brooders (30 °C). Food (Agway), crop gravel, and water were provided ad libitum.

Experiment 1: In vivo eye drop treatment with melatonin or RA

Experiment 1 was designed to examine the effects of hormone supplementation in vivo. Fifty-one hatchling chicks (2–3 days old) were treated with eye drops in both eyes for 2 weeks in CL. The daily eye drop treatment contained either a hormone supplement, or drops containing only vehicle and no hormone. Eye drops were applied every 4 h during the same daily 12-h period, for a total of four applications per day. Eye drop conditions for the experiments are summarized in Table 1.

Table 1.

Experiment 1 – eye drop treatments

Chick treatment group	Eye drop treatment	Treatment length (days)
SHAM (control) n = 7	CL: Hank's buffered saline eye drops	14
MEL, n = 8	CL: 50 μL of a [50 mg mL−1 solution of melatonin in ethanol) per mL Hank's buffered saline, final (ethanol) ≤ 5%]	14
RA, n = 18	CL: [retinoic acid; 2.7 mg mL−1 solution of RA in 95% ethanol was added 30 μL mL−1 into Hank's buffered saline, final (ethanol) ≤ 5%]	14
MEL/RA, n = 18	CL: (melatonin and RA, combined as above)	14

CL, constant light; RA, retinoic acid.

Throughout the experiment, the chicks were under veterinary supervision. Their corneas remained clear and their eyes did not develop signs of pathology during the treatments. After 2 weeks, the anterior chamber depth, lens thickness, vitreal chamber depth, refractive power of the eye, and corneal curvature were assessed using ultrasound and neutralizing infra-red retinoscopy (Li et al. 1995).

Birds from each in vivo group were euthanized after 2 weeks of eye drop treatment. One eye was sliced open just posterior to the ora serrata and immersed in 4% phosphate-buffered paraformaldehyde, pH 7.4. The cornea from the remaining eye was removed using iridectomy scissors, then weighed. Wet corneas were placed in a 60 °C oven for 2 days, then their dry weights were recorded. Weight and hydration data were compared with similar corneal data from a previous in vivo experiment using birds raised in either CL or 12/12 h of light/darkness (LD) under otherwise identical conditions (Wahl et al. 2009).

Experiment 2: melatonin receptor blocking in vivo

This experiment was designed to study the effect of reduced melatonin on growth of the anterior segment by blocking corneal melatonin receptors. Eighty-five newborn chicks were divided into six groups of 14–18 chicks. Half were raised under 12/12 LD with subconjunctival injections of 20 μg of luzindole (LI) or luzindole eye drops in 0.05 mL 8% alcohol saline solution. The rest were raised under CL with subconjunctival injections of 0.2 mg melatonin or melatonin eye drops, also in 0.05 mL 8% alcohol saline solution. The injections and the eye drops were given at the time equivalent to that of the beginning and middle of the normal dark phase. All the control groups received comparable saline and alcohol injections or eye drops. The treatments started at day one and terminated 18 days later. Refraction, radius of corneal curvature, and axial lengths of the chick eyes were measured at the end of the experiment.

Experiments 3–6: organ culture

These experiments examined the effects of hormones and light in vitro. Organ culture experiments are summarized in Table 2. In Experiment 3, cultured corneas were harvested from hatchling chicks (n = 36 birds). Fellow corneas were cultured in CL or N conditions for 11 days (n = 6 birds), 14 days (n = 18 birds) or 21 days (n = 12 birds). In Experiment 4, the corneas from 12 hatchling chicks were cultured with melatonin-supplemented Dulbecco's modified Eagle's medium (DMEM) in either N (6 birds) or CL (6 birds) conditions.

Table 2.

Experiments 3–6: conditions in culture

Initial exposure prior to harvest of corneas	Control cornea: culture conditions	Treated cornea: culture conditions	Days in culture (days)
Experiment 3
N, 1–3 days (n = 6 birds)	N	CL	11
N, 1–3 days (n = 18 birds)	N	CL	14
N, 1–3 days (n = 12 birds)	N	CL	21
Experiment 4
N, 1–3 days (n = 6 birds)	N	N + melatonin	14
N, 1–3 days (n = 6 birds)	CL	CL + melatonin	14
Experiment 5
CL, 14 days (n = 18 birds)	CL	N	14
N, 14 days (n = 18 birds)	N	CL	14
Experiment 6
CL, 14 days (n = 12 birds)	CL	CL + melatonin	14
CL, 14 days (n = 12 birds)	CL	CL + RA	14
CL, 14 days (n = 12 birds)	CL	CL + melatonin + RA	14

Corneas for Experiment 5 were harvested after 36 chicks had been raised for 14 days in CL (n = 18); or 14 days in N (n = 18 birds). The corneas from each bird were separated, one into N conditions, and its fellow into CL. These corneas did not receive hormone treatment but were assessed for the effects of light and the organ culture system on corneal growth. In Experiment 6, corneas from 47 birds raised in CL for 2 weeks were harvested and cultured with hormone-supplemented DMEM, with fellow (control) corneas cultured in plain DMEM.

Each cornea was removed with iridectomy scissors. Care was taken to include a thin scleral rim, ensuring that stroma was not exposed. Corneas were weighed, collected into fresh DMEM, rinsed for 30 s in 1% povidine-iodine, rinsed twice more in fresh changes of DMEM, and placed epithelial-side-up on 0.8% agar cushions in DMEM plus gentamycin (80 μg mL⁻¹), adapted from the technique employed by Liminga & Oliw (2000). The agar cushions were placed on Corning organotypic culture well inserts in either 6- or 12-well Corning tissue culture dishes. A 1-mL aliquot of fresh medium was supplied to each well and replaced every 3 days. The epithelial surfaces of the corneas were continuously exposed to the air. Nutritional support from the medium was delivered via diffusion through the agar cushions (Fig. 1).

Fig. 1.

(Experiments 3–6) Photograph of culture conditions. The cornea is resting in the culture well on an agar ‘pillow’ and is not in direct contact with the culture medium. The medium was removed and replaced from the space between the culture well insert and the wall of the culture well. It diffuses through the bottom of the insert and the agar to reach the cornea. This cornea has been in culture for 2 weeks. It has contracted significantly during this time and has nearly tripled in thickness (see Fig. 6). Originally, the borders of this cornea extended almost to the edge of the agar ‘pillow’. Contraction was typical for all corneas cultured in these experiments and did not affect the health of the tissue.

Corneas were cultured using high-glucose formulated DMEM (Gibco, with Gluta-Max). Groups in culture were: CL (control, no humoral factor added); CLM (melatonin Sigma # M5250) added with each fresh change of media to a concentration of 60 ng mL⁻¹); CLRA (retinoic acid; Sigma #R2625) diluted as follows: 2.7 mg mL⁻¹ solution of RA in 95% ethanol was added at 1 μL mL⁻¹ to fresh DMEM; CLMRA (melatonin and RA) concentrations as above.

Culture plates were placed into humidified Billups–Rothenburg modular incubators filled with 5% CO₂ in air. The modular incubators were housed at 37 °C in commercially available Hov-a-Bator picture window egg incubators. Light was provided by 50 W halogen lamps placed directly over the incubator windows. The overall luminance inside the incubator averaged 800 Lux. The growth in organ culture was assessed by measuring changes in wet and dry weight. After each experiment, corneas were weighed and dried as described above. We defined hydration state as [(final wet weight − dry weight)/final wet weight] × 100, or % water weight of the cornea. Random samples from each group were collected into 4% paraformaldehyde at pH 7.4 for histological assessment. Corneas were prepared for paraffin embedding and serial sectioning at 6 μm followed by conventional hematoxylin and eosin staining.

Statistics

To compute probability estimates of Type I errors that the variation in measured factors resulting from various treatments could have been obtained by chance, we wrote a Monte Carlo program in basic. We chose this method because various post-hoc statistical techniques are known to be more or less conservative and to have various not-well-known sensitivities to differences in sample numbers and standard deviations. Moreover, many of the statistical programs give only a measure of probability that is greater or less than a set probability level, e.g. 5%.

Our program may be described as follows: When there were N treatments each with n_N members, we employed a single random Gaussian distribution (Box & Muller, 1958) with a harmonic mean of the N treatments and a weighted average standard deviation of the N treatments (these are standard techniques used in most post-hoc procedures). For 1 million trials we selected n_N samples from this distribution for each treatment and found their mean values. The [N × (N− 1)]/2 differences between means were then calculated, ordered greatest-to least, and compared with the ordered differences of the corresponding experimental samples in the following manner: The largest difference of the random samples was first compared with the largest difference in the experimental data. If the difference of the random sample exceeded that of the experimental data, the frequency score for that experimental difference was incremented by one, and the second largest random sample difference was compared with the second largest experimental difference. However, if the largest random difference did not exceed that of the corresponding experimental difference, it was compared with the next-highest experimental differences until one was found which it exceeded. The frequency score for that experimental difference was then incremented by 1, and a new random difference was compared with the remaining experimental differences in rank order until one was found that it exceeded. This process was continued until all the random differences were exhausted. Then another set of random samples was drawn and the entire comparison procedure was repeated. After 1 million iterations the frequencies of which each experimental difference had been exceeded were computed and these frequencies were taken as the probabilities of finding these differences by chance. By repeating the Monte Carlo procedure on the same dataset a number of times we could estimate the variation in the random chance probability estimate of a difference which fell in the P = 0.05 and P = 0.001 regions. The standard deviation of the probability in the 0.05 region was 0.00019 (coefficient of variation = 0.5%) and in the 0.001 region 0.00004 (coefficient of variation = 2%). Hence we report our probability values with three-decimal point accuracy.

Our program was checked by comparing its results with that of the statistical program statview® (Anonymous, 1999). Our results usually fell in between those of the Fisher's PLSD test and those of the Student–Neuman–Keuls test, being closest to the Tukey–Kramer test. However, the significance results of our Monte Carlo test on the 21 comparisons of Table 5 (below) were identical with the results of the Student–Neuman–Keuls and Tukey–Kramer tests on the same data with regard to being greater or less than the significance level of P = 0.05.

Table 5.

Post hoc Monte Carlo probability values for comparisons of treatment pairs (row and columns) in Table 4. As in figures, the significance level was taken to be 5%

	CL-CLM	CL-CLMRA	CL-CLRA	CL-N	N-CL	N-N
CL-CL	NS	NS	NS	NS	< 0.001	0.017
CL-CLM		NS	NS	NS	< 0.001	0.004
CL-CLMRA			NS	0.004	< 0.001	< 0.001
CL-CLRA				0.012	< 0.001	< 0.001
CL-N					0.025	NS
N-CL						NS

‘NS’ without an additional probability value indicates P > 0.10.

Animal welfare

All animals were handled in strict accordance with good animal practice as defined by the N.I.H. and the Cornell Institutional Animal Care and Use Committee, and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Results

Experiment 1: effects of melatonin and RA on the cornea in vivo

The eye drop experiment afforded the opportunity to assess the in vivo effect after direct application of melatonin and RA on the shaping of the cornea during early post-hatching development in CL. There was no effect of melatonin or RA on vitreous chamber depth, lens thickness, or axial length (anova comparisons). However, there was a significant effect on anterior segment depth. In CL, chicks develop shallow anterior segments after 2 weeks, and are hyperopic (Li et al. 1995). RA and melatonin applied either individually or together, partially protected the anterior segment from these effects of CL (Fig. 2A; Monte Carlo, ‘M-C’, comparisons – Mel vs. sham P > 0.05, not significant, RA vs. sham P = 0.037, Mel + RA vs. sham P = 0.003). As expected, this correlated with significantly less hyperopia in the treated eyes (Fig. 2B; M-C comparisons – Mel vs. sham P = 0.042, RA vs. sham NS, Mel + RA vs. sham P = 0.002. In vivo, the percent water content of the chick cornea ranged from 80% in hatchlings to 78% in adults (data not shown), similar to data reported for normal human corneas (Waring et al. 1982). Among eye drop treatments, the corneas receiving RA were more hydrated than sham-treated eyes (P = 0.03), and those receiving melatonin did not differ from sham but were less hydrated the other groups (Mel vs. RA: P = 0.004, Mel vs. Mel/RA: P = 0.022; Fig. 3).

Fig. 2.

(Experiment 1) (A) Anterior chamber depth and (B) refraction of 2-week-old chicks treated with eye drops containing melatonin (M), retinoic acid (RA), both melatonin and retinoic acid (M/RA) or no hormones (SHAM). Inspection of the graphs shows that the treated animals were, on average, less hyperopic and had larger anterior chamber depths. This was, in large part, reflected in the pattern of statistically significant differences. In this and subsequent figures, error bars represent one standard error of the mean, and the inserted matrix gives the post hoc significance probabilities of the differences in means of their respective rows and columns. If the difference is significant and < 0.05, only the probability is given. If the probability lies between 0.05 and 0.1, both the probability and the designation ‘NS’ for ‘not significant’ are given. If the probability is > 0.1, only the designation ‘NS’ is given. See ‘Statistics’ under Materials and methods for further information.

Fig. 3.

(Experiment 1) Final hydration status of corneas harvested from 2-week-old chicks raised in constant light with eye drop treatments as described in Materials and Methods. The Y-axis indicates the percent of water weight of the cornea.

Experiment 2: melatonin receptor blocking in vivo

Refractions of chick eyes in CL treated with either melatonin eye drops or subconjunctival melatonin injections (Fig. 4) were partially protected from the refractive effects of CL in this experiment (as shown also in Experiment 1). Eyes treated with luzindol eye drops or sub-conjunctival luzindole injections under 12/12 became hyperopic compared to control groups (see Fig. 4 for significance values). The pattern of changes in corneal curvature of these eyes reflected the changes in refraction, but were not significantly different among treatment groups (data not shown).

Fig. 4.

(Experiment 2) Refractive error after exposure to constant light or 12/12 light/darkness among birds treated with melatonin (in constant light) or luzindole (in 12/12), respectively. LD, luzindole eye drops; LDC, luzindole sham eyedrops; LI; subconjunctival luzindole injections; LIC, luzindole subconjunctival sham injections; MD, melatonin eyedrops; MDC, melatonin sham eyedrops; MI, melatonin subconjunctival injections; MIC, melatonin subconjunctival sham injections. For further explanation, see Materials and methods.

Anterior chamber depths of chick eyes treated with melatonin eye drops or subconjunctival injections under CL were significantly deeper than those of control eyes (Fig. 5). However, the anterior chamber depths of eyes treated with luzindole drops or subconjunctival injections under 12/12 did not differ significantly from their controls. Of course, animals raised in CL had significantly shallower anterior chambers.

Fig. 5.

(Experiment 2) Anterior chamber depths for various treatments. Key to abbreviations in legend of Fig. 4. The animal numbers are the same as values in Fig. 4.

Histology of cultured vs. in vivo corneas

The normal chick cornea at 32 days of age is shown in Fig. 6A. After 14 days in culture the corneal epithelium was healthy and resembled the in vivo condition (Fig. 6B). The fact that the anterior surface of the cornea was exposed to air allowed the epithelium to be maintained in a healthy condition, and it remained several cell layers thick throughout the experiment. The endothelium was continuous but fewer endothelial cell nuclei were observed than typical for in vivo corneas of the same age. Stromal cells appeared to be healthy but were rounded and disorganized compared to in vivo corneas. The loss of tension at the margins of the cornea caused it to contract, and this contraction reorganized the stroma. The fresh appearance of cultured corneas differed from in vivo corneas in that they were much thicker, and slightly cloudy in appearance. Overall, transparency remained fairly high. Text could still be read through the cultured corneas. There were no differences in the histological appearance or distribution of corneal epithelial, stromal or endothelial cells among any of the culture treatment groups, including RA and melatonin-treated cultures.

Fig. 6.

(Experiments 3–6) Corneal sections from in vivo (A) and cultured (B) corneas at 4 weeks of age. Corneas were immersion-fixed in 4% paraformaldehyde at pH 7.4, embedded and sectioned at 6 μm, then stained with hematoxylin and eosin. In (b), the cornea was harvested from a 2-week-old bird and then cultured for 2 weeks. Because cultured cornea does not have a marginal attachment, it contracts and thickens. There was no apparent difference in morphology of cultured corneas undergoing different light and hormone treatments.

Experiment 3: CL vs. N light effects on water weight of hatchling corneas in culture

There was no significant difference in water weight among corneas from hatchling birds cultured in CL or N for 11, 14 or 21 days. The dry weights of these corneas also did not differ significantly (anova comparisons, data not shown).

Experiment 4: effect of melatonin on water weight of hatchling corneas in culture

Hatchling chick corneas from pairs of fellow eyes were divided into either N or N + melatonin conditions, or CL vs. CL + melatonin conditions. No significant effect of melatonin on hydration was observed on corneas cultured from hatchling birds in either CL or N conditions during 14 days of culture (results not shown).

Experiment 5: effect of N and CL on water weight before and after culture

Raising birds for 2 weeks prior to culture in either CL or N did not affect the final percent water weight after culture (Table 3). The percent of water in corneas from birds raised in CL conditions for 2 weeks, then divided into N and CL cultures, did not differ. Birds raised in N conditions prior to culture whose corneas were divided into N and CL cultures also did not differ after 2 weeks.

Table 3.

Corneal water weight in constant light (CL) vs. n in vivo pre-exposure followed by 2 weeks of CL orN culture (Experiment 5)

	Percent of total weight in water
Initial condition	CL culture	N Culture
CL in vivo	87.2% ± 0.456 SE	86.4% ± 0.487 SE
N in vivo	86.4 ± 0.552 SE	87.8 ± 0.385 SE

Experiment 6: effects of melatonin and RA on water weight in culture

Melatonin had a dehydrating effect in organ culture among corneas from birds raised previously for 2 weeks in CL conditions (Monte Carlo comparison, P < 0.001, Fig. 7). RA or RA + Mel-treated corneas were significantly more hydrated than Mel-treated corneas in culture (P < 0.001).

Fig. 7.

(Experiment 6) Hydration of corneas computed by subtracting dry weight from the initial wet weight and expressing the result as a percentage of the original wet weight. Among corneas cultured from 2-week-old chicks, there is an effect of melatonin (CLM) on the hydration status of corneas cultured from constant light (CL) birds. Melatonin dehydrates corneas, and retinoic acid probably hydrates them. The CL bar represents the mean of all fellow eyes.

Effects of CL, N, RA, and M on dry weights after 2 weeks in culture

Corneas cultured in CL from birds raised for 2 weeks in N had the greatest mean dry weights after 2 weeks in culture (Table 4). They were significantly heavier than all corneas that had CL pre-culture conditions. The greatest dry weights in each row of Tables 4 and 5 were associated with a change in light regime. CL corneas transferred to CL with any hormone supplement did not differ in dry weight from controls (Table 5).

Table 4.

Weights in milligrams, followed by ±1 SEM and sample size. Corneas were harvested from chicks exposed for 2 weeks to either N or CL conditions (left hand column) and then cultured for 2 weeks as indicated in top row

	Culture condition
Pre-culture condition	N	CL	CLM	CLMRA	CLRA
N	2.486 ± 0.88 n = 18	2.625 ± 0.29 n = 18
CL	2.320 ± 0.23 n = 15	2.190 ± 0.24 n = 45	2.109 ± 0.13 n = 8	1.963 ± 0.15 n = 11	1.990 ± 0.16 n = 11

Discussion

These experiments show that melatonin and RA have opposite effects on corneal hydration. Interestingly, both molecules also ‘rescue’ CL-exposed eyes from the optical effects of CL (Experiment 1).

Melatonin did not have a significant effect on hydration in vivo in Experiment 1 (Fig. 3) but it did protect the eyes of hatchling chicks from CL-induced hyperopia (Fig. 2B). Melatonin had a dehydrating effect on 2-week-old CL corneas transferred to culture for 2 weeks, as we have seen in Experiment 6 (Fig. 7). How can we account for the protective effect of melatonin in the eye drop experiment on hatchling chicks, as no significant differences in hydration were observed? We speculate that protective effect of melatonin on refraction prior to 2 weeks of age is due to regulation of the composition of the stromal matrix, and we further hypothesize that such matrix regulation by melatonin produces an appropriate water gradient across the stroma. This effect may be too subtle to discern by comparing wet and dry weights at 2 weeks of age.

Further support for the protective effect of melatonin is seen in the results of Experiment 2. Here, luzindol blocked the melatonin receptors of chicks in 12/12 light regimens, producing hyperopia, as if the chicks were in CL. We cannot explain the lack of effect on anterior segment depths that usually accompanies the hyperopic refractions in CL.

From Experiment 3 we learned that direct effects of light on the hydration of cultured hatchling corneas were negligible, showing that corneal tissues are not directly sensitive to light. If there are cryptochromes present in the cornea they do not appear to be influencing those parameters of corneal growth that we measured in vitro. We found that the same was true when 2-week-old corneas were transferred to culture in CL or N conditions for 2 weeks (Experiment 5).

Corneas grown in culture increase in dry weight, as may be seen by comparing the data of Tables 4 and 5 with that in Wahl et al. (2009). During growth, corneal dry weight increases least in CL, both in culture and in vivo (Tables 4 and 5; culture data, and Fig. 3; in vivo data). RA reduces dry weight in culture more than melatonin does, but the RA-treated corneas are significantly more hydrated than CLM corneas. Thus, hydration state is under humoral influence. As the corneal histology is indistinguishable between treatment groups, it may be that the differences in hydration are due to differences in GAG ratios of the stromal matrix among groups.

Light regimen in corneas cultured from 2-week-old birds does not affect corneal hydration (Table 3, Experiment 4). Dry weights, however, differ between 2-week-old in vivo CL vs N corneas that are subsequently cultured in CL (Tables 4 and 5). This result suggests that the CL in vivo cornea is different from the N cornea when it is transferred to culture, although there is no measurable difference in hydration or weight at that time. It is possible that interruption of the dark phase by transfer of N corneas to CL culture may have stimulated proteoglycan synthesis.

There is no evidence that the adult chicken cornea responds to melatonin or RA by producing a change in shape. Why not? We speculate that these hormones may have an altered influence over adult keratocytes or endothelial cells, just as the effect of growth factors changes during ontogeny (Gale et al. 1996). Although melatonin is not typically considered a ‘growth factor’, its role may nevertheless vary with age.

In conclusion, this study provides evidence that a light-regulated humoral factor or factors, including melatonin, influence shaping of the chick cornea during growth through regulation of corneal hydration and dry weight. Melatonin, previously shown to cycle during normal periods of light and dark, protects CL-exposed eyes from hyperopia, and also reduces corneal hydration in culture. Although this study demonstrates that RA increases corneal hydration in culture and may also protect the eye from CL-induced hyperopia, its regulation by light regimens has not been demonstrated. Further studies that dissect the relative contributions of light cycles, melatonin, and RA to corneal growth and illuminate the underlying mechanisms are warranted.