For the comparison of RGB LEDs

LED lights with varying spectral characteristics were used as the sources for primary exposure treatments. As shown in Figure 22, single-wavelength blue LEDs (460 nm, 102.3 μW/cm²), green LEDs (530 nm, 102.8 μW/cm²), and red LEDs (620 nm, 102.7 μW/cm²) were custom-made for the exposure experiments (BlueDog Technology Corporation Ltd., Taipei, Taiwan). Each light source was pretested in an integrating sphere and programmed for 40 measurements on site.

Figure 22 LED light source
spectral power distribution (SPD) curves 3.2.3 Light exposure

3.2.3.1 For the comparison of CFL vs. LED

As shown in Figure 19. The animals were divided into 4 groups, and each rat was stored in an individual transparent cage with a dimension of 45 x 25 x 20 (cm). Each cage was placed in the center of a rack shelf with dimensions of 75 x 45 x 35 (cm). The light sources were set on the top of each shelf and were

started at 6:00 PM of Day 15 with the total exposure duration ranging from 3, 9, to 28 d under 12 hr dark/12 hr light cyclic routines. The animals were sacrificed for analysis after light exposure. However, a special treatment for 32 animals was performed, 8 from each group were returned to a dark environment for 14 d of recovery after 28 d of exposure. The objective of the recovery stage was to allow for possible removal of necrotic photoreceptor cell debris.

3.2.3.2 For the comparison of RGB LEDs

As shown in Table 2, the animals were randomly divided into 3 exposure groups, and each rat was housed in an individual transparent cage with dimensions of 45 x 25 x 20 (cm). As shown in Figure 23, each cage was placed in the center of a rack shelf with dimensions of 75 x 45 x 35 (cm). Each rack, equipped with 6 layers of shelving, was covered with a black curtain to keep the light intensity and quality separate. The light sources were set at the top of each shelf and were measured at 20 cm from each source to acquire the irradiance at the level of the cornea at 102.3 μW/cm², 102.8 μW/cm², and 102.7 μW/cm² for blue, green, and red, respectively. After 10 days of environmental adaptation, the light exposure was initiated at 6:00 PM on Day 11, with the total exposure duration ranging from 3 days to 9 days to 28 days under a 12 h-dark/12 h-light cyclic routine. The animals were sacrificed for analysis after light exposure.

Figure 23 Diagram of light exposure setting

3.3 Sample pretreatment

The animals were anesthetized, and both eyes were scanned using electroretinography (ERG) after completing the light treatment. They were sacrificed with pentobarbital sodium (> 60 mg/kg, intraperitoneal) immediately after the ERG scans. For hematoxylin and eosin (H&E) staining and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining, enucleated eyes were immersion-fixed in 4% paraformaldehyde in 0.1 M phosphate buffered saline (PBS) at pH 7.4 overnight before being embedded in paraffin. For the transmission electron microscopy (TEM) analysis, the eyeballs were immersion-fixed in 2.5% glutaraldehyde in PBS for 2 h before further

thickness were made in the glass slide and maintained at -80°C until analysis.

The mid-superior aspect of the retina was examined using H&E, TUNEL, TEM, and IHC. For the superoxide anion (O2-.) assay, the eyeballs were frozen immediately in liquid nitrogen after enucleation. The eyeballs were ground with saline (500 μL saline per eye) for extraction. Additionally, for the western blot (WB), the hydrogen peroxide (H2O2) assay, and the iron assays, retinal tissues were taken immediately for protein extraction after the eyes were enucleated (as Figure 24). As reported previously⁶⁵, the proteins were extracted from the retinal homogenates using radioimmunoprecipitation assay (RIPA) lysis buffer, which contained 0.5 M Tris-HCl (pH 7.4), 1.5 M NaCl, 2.5% deoxycholic acid, 10%

NP-40, 10 mM EDTA, and 10% protease inhibitors (Complete Mini; Roche Diagnostics Corp., Indianapolis, IN, USA).

(A) (B) Figure 24 (A) Eye enucleation and (B) retina tissue removal

3.4 Analytical Methods

3.4.1 Electroretinography (ERG)

As shown in Figure 25 and Figure 26, ERG was performed as described

all rats before and after light exposure using ERG (Acrivet, Hennigsdorf, Germany). After 18 h of dark adaptation, rats were anesthetized using an intramuscular injection of 100 mg/kg ketamine and 5 mg/kg xylazine (WDT eG, Garbsen, Germany). One drop of tropicamide (0.5%) (Mydriaticum Stulln, Pharma Stulln, Germany) was applied for pupil dilation before ERG measurement. One drop of Alcaine (0.5%) (proxymetacaine hydrochloride;

Alcon Pharmaceuticals Ltd, Puurs, Belgium) was applied for local anesthesia before placing the active electrode onto the cornea (see Figure 25). Two subcutaneous needle electrodes (Ambu Neuroline Twisted Pair Subdermal, Bad Nauheim, Germany) served as the reference and ground electrodes. The reference needle was subcutaneously inserted between the eyes, and the ground needle was subcutaneously inserted between the rear legs to obtain the proper impedance levels, which were less than 10 kΩ at 25 Hz. LED flashes were stimulated without background illumination, and the flash interval was 1 s with a flash duration of 3 ms. The weighted average of 10 stimulations was computed by the program to produce the final detection values.

Figure 26 Diagram of retina response components to ERG stimulation ¹⁵

3.4.2 Hematoxylin and eosin (H&E staining)

Retinal histology was performed as described previously with modifications

sectioning was performed (4% paraformaldehyde in 0.1 M phosphate buffer [pH 7.4] for 1 h at 48°C), and the eyeballs were dehydrated in EtOH, infiltrated in xylene, and embedded in paraffin. Radial 5 μm sections were stored at 48°C.

The histologic analysis included quantification of the outer nuclear layer (ONL) and retina morphology alteration using a light microscope. The midsuperior aspect of the retina was examined for all histological analyses in this study.

3.4.3 Transmission electron microscopy (TEM) analysis

TEM was performed as described previously ⁶⁴ after 9 d of light exposure.

The processes were performed at the Electron Microscopy Facility at the Department of Pathology at National Taiwan University Hospital (Taipei, Taiwan). Retina slices of 1 mm (Figure 27A) were prefixed in 2.5%

glutaraldehyde in PBS, postfixed with 2% osmium tetroxide, and dehydrated for 10 min each in sequential baths of 30%, 50%, 70%, 90%, and 100% ethanol.

The specimens were placed into propylene oxide for 30 min, followed by a mixture of propylene oxide and epoxy resin for an additional 1 h; the samples were subsequently embedded into a gelatin capsule with epoxy resin at 60°C for one day. Subsequently, 80 to 90 nm ultrathin sections were obtained using an ultramicrotome. The sections were stained with 2% tannic acid in distilled water (DW) for 5 min, followed by 2% uranyl acetate in DW for 15 min and a lead-staining solution for 5 min. In the final step, the sections were coated with a thin copper grid-film and placed in a vacuum chamber for scanning (Figure 27B). As shown in Figure 28A, the specimens were examined using TEM with a high-resolution instrument at 80 kV (JEOL JEM-1400, Peabody, MA, USA), and

Figure 27 Specimen of a retina slice

(A) (B) Figure 28 TEM observation instrument (JEOL JEM-1400)

(A) (B)

3.4.4 Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)

The TUNEL assay was performed using a FragEL^TM DNA fragmentation detection kit (Calbiochem, Darmstadt, Germany) following the standard protocol with a minor modification to detect apoptotic cells after 9 d of light exposure.

Tissue sections were deparaffinized, rehydrated, and blocked using endogenous peroxidase with H2O2 for 30 min. Antigen retrieval was achieved by pressure-cooking in a 0.1 M citrate buffer at pH 6 for 10 min followed by cooling at room temperature before incubation with the enzyme. The TUNEL enzyme (1 h at 37°C) and peroxidase converter (30 min at 37°C) were applied to the 10 μm sections after incubation for 5 min in a permeabilizing solution of 0.1% Triton-X in 0.1% sodium citrate. The tissues were counter-stained with DAPI, and the DNA strand breaks were labeled with fluorescein FITC-Avidin D. The fluorescent signals were obtained by adding FITC-Avidin, which bound to the biotinylated-dU of the damaged DNA. After staining, image analysis was used to quantify the relative fluorescence intensity of the TUNEL-positive cells, with the number of TUNEL-stained nuclei quantified in 4 random slides per sample.

Sections were visualized on a fluorescent microscope over the entire retina excluding the RPE layer (Nikon Instruments Inc., NY, USA). The number of TUNEL-positive cells for each section was counted by Image-Pro Plus software (v.6.0).

3.4.5 Immunohistochemistry (IHC)

Immunohistochemistry was performed as described previously ^{67, 68} after 9 d of light exposure. In brief, cryosections of the retina samples were incubated

deoxyguanosine (8-OHdG) (1:50, JAICA, Tokyo, Japan)], lipids [Acrolein, (1:200, Advanced Targeting Systems, San Diego, CA USA)] and proteins [nitrotyrosine, (1:200, Abcam, Millipore Billerica, MA, USA)]. The same quantification method used for the TUNEL analysis was applied in the IHC analyses. The relative fluorescence intensity corresponding to the number of IHC-positive cells for each section was measured and quantified by Image-Pro Plus software (v.6.0).

3.4.6 Free radical assay (reactive oxidative species, ROS)

Measurement of reactive oxygen species in the retina was performed as described previously ⁶⁸ after 3 or 9 d of light exposure. In brief, 0.2 mL of homogenized extraction was loaded with 0.1 mL of 0.9% saline onto a 3-cm dish with a stir bar placed at the center. The dish was placed into the chemiluminescence analyzer chamber (Tohoku CLA-FS1, Miyagi, Japan). The ROS were quantified after adding the enhancer Lucigenin to the chemiluminescence analyzer. After 60 s of background detection, 1 mL of a Lucigenin (bis-N-methylacridinium nitrate) solvent (2.5 mg of Lucigenin dissolved in 50 mL 0.9% saline) was added for stimulation. The stimulated O2-.

and total oxidative products were captured every 10 s and computed for 7 min after 1 min of baseline detection.

3.4.7 Western blotting (WB)

Total protein was extracted from the retina by lysing the sample in radioimmunoprecipitation assay (RIPA) buffer [0.5 M Tris-HCl (pH 7.4), 1.5 M NaCl, 2.5% deoxycholic acid, 10% NP-40, 10 mM EDTA] and protease

boiled for 5 min. A 100-mg sample was separated on 10% SDS-polyacrylamide gels and then transferred to polyvinylidene difluoride membranes (Immobilon-P;

Millipore Corp., Billerica, MA, USA). The membranes were incubated with hemeoxygenase-1 [(HO-1); Abcam, Cambridge, MA, USA], ceruloplasmin [(CP); Santa Cruz Biotechnology, Dallas, Texas, USA], anti-cytosolic glutathione peroxidase [(GPx1); Abcam, Millipore Billerica, MA, USA], anti-poly (ADP-ribose) polymerase-1 [(PARP-1); Cell Signaling Technology Inc., Danvers, MA, USA], anti-superoxide dismutase [(SOD2);

Santa Cruz Biotechnology Inc., Dallas, Texas, USA], and anti-β-actin (Abcam, Millipore Billerica, MA, USA) antibodies. The membranes were incubated with horseradish peroxidase-conjugated secondary antibody and visualized by chemiluminescence (GE Healthcare). The density of the blots was determined using image analysis software after scanning the image (Photoshop, ver.7.0;

Adobe Systems, San Jose, CA, USA). The optical densities of each band were evaluated by comparison with the density of the β-actin bands.

3.4.8 Hydrogen peroxide (H2O2) assay

The assay was performed using a hydrogen peroxide colorimetric/fluorometric assay kit (BioVision, Milpitas, CA, USA) following the standard protocol to detect H2O2 concentrations after 3 days of light exposure. In brief, total protein was extracted from the retina and centrifuged for 15 min immediately after the extraction. Each well was loaded with 20 μl samples and brought to a volume of 50 μl with assay buffer. Reagents and H2O2

standards were mixed and then incubated for 10 min. A background detection

3.4.9 Total iron and ferric (Fe³⁺) assay

The assay was performed using a QuantiChrom Iron Assay (DIFE-250) Kit (BioAssay Systems, Hayward, CA, USA) following the standard protocol to detect total iron concentrations after 3 days of light exposure. In brief, total protein was extracted from the retina, and 50 μl of the extraction was loaded into a 96-well plate. Then, 200 μl of working reagent was added and incubated for 40 min at room temperature, followed by an optical density reading at 510–630 nm.

Sample readings were compared with the standard curve to calculate the concentrations.

3.5 Statistical analysis

Data are presented as the mean ± SD unless otherwise stated. Data were evaluated using an analysis of variance (ANOVA) with Tukey’s post hoc tests to show differences between the groups. A p value less than 0.05 was considered statistically significant.

4 RESULTS AND DISCUSSION

4.1 White LED at domestic lighting level to induce retinal injury 4.1.1 Electrophysiological response shows photoreceptor cell function loss

The representative ERG response curves of rats are shown in Figure 29A.

The normal retina showed a high b-wave peak, but the injured retina curved a low b-wave peak as a result of cell function loss. As shown in Figure 29B. Two LED groups and the white CFL group all demonstrated a significant decrease of b-wave amplitude at day 9 and day 28 after light exposure (ANOVA followed by Tukey post hoc test p < 0.001). The b-wave amplitude of the yellow CFL group did not decrease significantly at day 9; however, it had 21% of decrease at day 28 after light exposure. The data from each of the four exposure groups was not statistically different at 28+14 d as compared to 28 d of exposure, and this trend was also applied to the H&E staining results (data not shown). No significant development was found after 3 d of light exposure, and therefore data were not shown as well.^

Figure 29 ERG responses after light exposure

Both LED groups demonstrated a significant decrease of b-wave amplitude at day 9 and day 28 after light exposure. The fluorescent lamp groups developed severe loss of b-wave amplitude until 28 d of light exposure. n = 3 for controls, n = 3 for 3 days of exposure groups, and n = 8 for each exposure group at each time of exposure (Curve scale: amplitude = 250 µV and stimulation = 50 msec). (**,

*** p < 0.01, 0.001, respectively, compared to the “normal” group by ANOVA with the Tukey post hoc test).

4.1.2 Retinal histology–H&E staining showing layer damages

As shown in Figure 30a-b. White LED light exposure can lead to morphologic alterations in the rat retina. The group that was exposed to 750 lux white LED light for 28 d exhibited the adverse effect of light exposure including the pyknotic photoreceptor nuclei (arrow), swelling of the inner segment (arrow head), and a disorganized outer segment (asterisk). As shown in Figure 30c-f.

The ONL thickness of white and blue LED groups decreased significantly at day 9 and day 28 (data not shown) after light exposure (ANOVA followed by Tukey post hoc test p < 0.01), whereas, the ONL thickness of the white and yellow CFL groups did not decrease significantly at day 9 after light exposure.

Figure 30 Retinal light injury after 9 d or 28 d of exposure analyzed by H&E staining

GCL: ganglion cell layer. INL: inner nuclear layer. ONL: outer nuclear layer. IS: inner segment. OS:

outer segment. *RPE: the retinal pigment epithelium (usually next to the OS layer) is detached and cannot be found within this scope.

(A) (a) Normal retina layers, and (b) light exposure-induced retinal injury, including the absence of photoreceptors and INL degeneration. (B) The ONL thickness of the LED groups decreased significantly at day 9 and day 28 after light exposure, whereas the ONL thickness of white and yellow CFL groups did not decrease significantly at day 9 after light exposure. Both blue (c) and white LED (d) light exposure caused the disappearance of photoreceptors; the white CFL group (e) exhibited distortion of the OS and ONL; and the yellow CFL group (f) exhibited less movement in each layer. n = 3 for controls and n = 8 for each group after 9 or 28 days of exposure (** indicates

4.1.3 Apoptosis Detection - TUNEL staining detects nuclear apoptosis

The retinal TUNEL stains are shown in Figure 31. Light exposure induced significant retinal cell apoptosis in all groups. However, more apoptotic cells were shown in the retina of the LED groups than in the retina of the CFL lamp groups after 9 d of exposure (ANOVA followed by Tukey post hoc test p <

0.001 for LED groups; p < 0.01 for CFL groups).

Figure 31 Light-induced retinal cell apoptosis tested by TUNEL labeling

GCL: ganglion cell layer. INL: inner nuclear layer. ONL: outer nuclear layer. RPE: the retinal pigment epithelium.

The damaged retina cells correspond to the positive labeling. (A) The result shows that more apoptotic cells (arrows) appear in the retina of the LED groups than that of the CFL groups after 9 days of light exposure. (B) The LED groups exhibit higher fluorescence intensity. n = 3 for controls and n = 8 for each exposure group (**, *** p < 0.01, 0.001, respectively, compared to the “normal”

group by ANOVA with the Tukey post hoc test; scale bar = 50 µm).

4.1.4 TEM demonstrations on the cellular injury

As shown in Figure 32 (samples were taken after 9 d of white LED light exposure). Nucleolus damage of photoreceptors occurred after exposure including early stage of nucleolus condensation (32b), karyolysis (32c), pyknosis (32d-e), and karyorrhexis (32f). Another crucial observation of photoreceptor injury included disruption of the inner and outer segments, which is shown in Figure 32g-l.

Figure 32 Retinal cellular injury studied by TEM

The photoreceptor nucleolus damage after LED light exposure result in (A) ONL nuclear deformations (arrows) shown as (a) normal ONL nucleus; (b) nucleolus condensation; (c) karyolysis;

(d and e) pyknosis; (f) karyorrhexis. (B) Photoreceptor deformations and (g) normal photoreceptor, IS and OS; (h and i) showing minor disruption; (j, k, and l) and IS disappearance followed by OS shrinkage and the formation of several small round shapes (scale bar = 2 µm for g, h, and k; scale bar = 1 µm for the rest of others). n = 3 for controls and n = 5 for white LED group after 9 days of

4.1.5 Immunohistochemistry (IHC) staining results indicating retinal light injury

Oxidative injury results in adducts on macromolecules that can be detected by immunostaining. The antibodies that specifically recognize these adducts provide evidence of the oxidative injury. Three antibodies were used to detect cell conditions in these experiments after 9 d of light exposure, including acrolein for lipid recognition (Figure 33A), 8-OHdG for DNA detection (Figure 33B), and nitrotyrosine for protein identification (Figure 33C). The results show that LED groups exhibit higher fluorescence intensity with 8-OHdG, acrolein and nitrotyrosine in ONL (ANOVA followed by Tukey post hoc test p < 0.001 for LED groups) and that the fluorescent lamps induced lower fluorescence intensity of 8-OHdG, acrolein and nitrotyrosine in ONL.

Figure 33 Retinal light injury labeling after 9 d of exposure by IHC

(A) Acrolein was used to detect the lipid adducts on macromolecules; (B) 8-OHdG was used to detect the DNA adducts; and (C) Nitrotyrosine was used for protein adduct recognition. The result shows LED groups exhibit higher fluorescence intensity on ONL, and the fluorescent lamp groups

B

4.1.6 Oxidative Stress -- superoxide anion O2-. shows the injury

As shown in Figure 34A. Lucigenin-stimulated superoxide anion (O2-)and total oxidative products were computed for all groups. After 3 d of blue LED light exposure, the retina O2- exceeded 60000 in 8 min, the white LED group exhibited a high total count close to 40000, and the fluorescent groups accumulated smaller total counts from 20000 to 30000. However, the plot exhibited an opposite trend when the exposure duration was increased to 9 d (Figure 34B). This result suggests that retinal oxidative stress may be induced by light exposure in the early stage.

Figure 34 A reactive oxygen species assay after 3 d and 9 d of light exposure

CL: chemiluminescence

(A) After 3 d of blue LED light exposure, the lucigenin-stimulated superoxide anion (O2-) exceeded 60000 in total count; the white LED group had a high total count close to 40000; and the fluorescent groups accumulated less total counts from 20000 to 30000, whereas normal rats exhibited only a count of approximately 1000. n = 3 for controls and n = 3 for each exposure group. (B) After 9 d of exposure, the O2- total count for the blue LED light group decreased to 8000; the white LED light group decreased to 18000; and both fluorescent light groups remained at the same level at 20000 to 30000. n = 3 for controls and n = 8 for each exposure group (**, *** p < 0.01, 0.001, respectively, compared to the “normal” group by ANOVA with the Tukey post hoc test).

ROS - Lucigenin

Reactive oxygen species -- superoxide anion O₂^-.

CL Intensity (in thousands)

Reactive oxygen species -- superoxide anion O₂^-.

CL Intensity (in thousands)

4.2 Mechanism of LED induced retinal injury and its wavelength dependency 4.2.1 Functional and morphological alterations

The representative ERG response curves of the testing animals are shown in Figure 35. Whereas the control group showed normal ERG a- and b-waves, the blue LED significantly weakened the ERG responses after 3 days of exposure. Moreover, the b-wave amplitudes for the blue, green, and red light exposure groups all showed a significant decrease compared with the control group after 9 days of exposure. The H&E images in Figure 36A show an uneven morphological alteration in the rat retinas after 28 days of light exposure. Figure 36B quantifies the thickness of the ONL and shows that the blue LED exposure group has the least thickness.

The TEM histopathology analysis (Figure 37) highlighted that the injury in the RPE and photoreceptors area from blue light could be lethal after 9 days of exposure; more details were described in section 4.1.4. Referring to the retinal remodeling phases ^{62, 63}, Figure 37A shows the normal ONL nucleolus and its pyknosis in phases 1 and 2. Figure 37B shows the normal photoreceptor outer segment (POS) and its disorganized disks in phase 1 and the round POS in phase 2. Figure 37C displays the normal oval-shaped RPE nucleus and shrinkage in phase 1 and the RPE condensation and deformation in phase 2.

As shown in Figure 38, the apoptotic analysis by TUNEL staining in Figure 38A and 38B shows significant fluorescence response increases after 9 days of light exposure. Both the histological and apoptotic results showed that light exposure may cause RPI for blue, green and red LEDs. However, the blue

death. ⁶⁹ The caspase-independent apoptotic marker PARP-1 also showed higher

death. ⁶⁹ The caspase-independent apoptotic marker PARP-1 also showed higher