UVR exposure
A full description of UV exposure and calculations is given in Supplementary Information S1. Briefly, UV quantification at 1 nm intervals was performed using an SR9910-v7 UV-Vis dual monochromator spectrometer (Irradian Limited, Elphinstone, UK) equipped with a cosine level corrected sensor assembly. The culture plate was irradiated inside a dedicated tissue culture incubator (Hera Cell, Heraeus, Germany), maintained at 37 °C, 5% CO2 and 100% humidity, with an unaligned beam at 19 cm from the light guide opening of the UVR source , a 120 W mercury metal halide epifluorescent lamp (Excelitas Technologies Corp., NY, USA). Exposures were performed for 72 h at three irradiance levels, one at full intensity (full irradiance) and two with ø21.3 mm neutral density filters (ND 0.2 and ND 0.4 (ThorLabs Inc., NJ, USA)) placed in a filter housing of the culture wells (Fig. 1). For each exposure, a negative (dark) control—consisting of a solid black resin disc—was placed in one of the available filter positions. Comparison of the typical experimental UVR energy doses used in the present study and the estimated terrestrial UVR dose confirmed that the UVR doses used in the present study achieved parity with the range of UVR doses experienced by a given individual over their lifetime ( see “Supplementary data S1”).
Cell viability
Cell viability assay was performed using PrestoBlue cell viability reagent (Invitrogen, MA, USA) according to the manufacturer’s instructions. Briefly, culture medium was removed and cells were washed with phosphate-buffered saline (PBS, Merck, MA, USA) containing calcium and magnesium chloride (referred to here as PBS+/+). After washing, one well of the dark control cells was exposed to cell lysis buffer (RIPA Lysis and Extraction Buffer, Thermo Scientific, MA, USA) for 5 min at room temperature to act as a positive control. Then, 120 μL of a 1:10 dilution of PrestoBlue cell viability reagent (Invitrogen, MA, USA) and PBS+/+ was added to each exposed well. Cells were then incubated at 37 °C and 5% CO2 for 20 min to allow the assay to develop. After incubation, 100 μL of the developed reagent was transferred to a 96-well solid white plate and the fluorescence was read using a multimode plate reader (Ex 520 nm/Em 580 nm, GloMax Explorer, Promega, WI, USA).
Electrical Cell-Substrate Impedance Detection (ECIS)
ARPE-19 cells were seeded in 96-well ECIS culture medium with each well housing 20 300 µm cross-directional electrodes (96W20Eidf, Applied Biophysics, NJ, USA). Immediately before UV exposure and after tissue maturation, the electrode plate was installed in a 96-well ECIS station. Spectral measurements of electrical resistance (Z), resistance (Ω), and capacitance (C) between 500 Hz and 64 kHz were determined at 11-min intervals during UV exposure.
Cell staining and image analysis
Within 2 h of UV exposure, cells were processed for imaging as follows: an equal volume of carboxy-H2DCFDA (Invitrogen, MA, USA) diluted 1:500 in Hank’s balanced salt solution (HBSS, Sigma-Aldrich , MI, USA) was added to the in situ culture medium and mixed gently by trituration to give a final carboxy-H2DCFDA dilution of 1:1000. The cells were then incubated with the staining solution for 30 min at 37°C. During incubation, a second staining solution comprising Hoechst 33342 (Sigma-Aldrich, MI, USA) diluted 1:500, CellMask Orange (Invitrogen, MA, USA) diluted 1:500, MitoM Fracker Deep diluted at a ratio of 1:500 and propidium iodide (Sigma-Aldrich, MI, USA) diluted at a ratio of 1:1500 were prepared in HBSS. During the last 10 min of the 30 min incubation, an equal volume of the second staining solution was added to the first solution in the wells. This was then incubated at 37 °C for another 10 min. When all staining was complete, 50% triplicate washes were performed using HBSS. Live imaging was performed using an automated microscope developed for high-throughput imaging (Operetta, PerkinElmer, MA, USA), whose imaging chamber was maintained at 37 °C and 5% CO2 throughout image acquisition. Seventeen fields were recorded per well at 40X objective magnification (see Fig. 2 for representative images of fluorescent probes). Image analysis was performed using the Columbus Image Analysis Suite (PerkinElmer, MA, USA). Parameters determined included cell number, fluorescence intensity, texture resolution (saddle ridge (SER) textures), and STAR morphology (symmetry properties, solid threshold, axial properties, radial properties, and profiles) for each cell compartment. After quantification of the image bundle, all data were exported as a text file for analysis in Excel 2016 (Microsoft Systems, CA, USA) and R-studio26.
Data curation and statistical analysis
PrestoBlue (Invitrogen, MA, USA) data were scaled between positive (solvated) and negative (no UVR) “dark” controls using the formula: (\left( {\left( {\chi – MIN } \right)/ \left( {MAX – MIN} \right)*100} \right)) where ‘x’ refers to the single measurement, ‘MIN’ refers to the lowest value within the data set, and ‘MAX ‘ refers to the highest value. Similarly, ECIS data were scaled between positive (no cells) and negative (no UVR) controls before being normalized to time zero (t0). Where possible, impedance, resistance and capacitance data were used to model Rb [tight-junction integrity]a [electrode coverage & cell adhesion] and Cm [membrane capacitance] values (collectively referred to as RbA) as previously described27,28. In order to model UVR damage performance at each wavelength using PrestoBlue (Invitrogen, MA, USA) data, the viability of treated cells for each wavelength and irradiance (full irradiance, ND02 and ND04) was used to calculate the coefficients linear regression. the slope coefficient of which is indicative of the effectiveness of UVR damage. Since the calculated slopes were all negative, they were first squared so that they could be logarithmically plotted and normalized to the visible wavelength of 405 nm. Modeling of photodamage performance using ECIS was performed by first selecting a “common action” for all wavelengths as a 60% reduction in the electrostatic parameters (Z, Ω, C) from their initial values. By observing the time point at which the specified action is achieved, combined with knowledge of the irradiance of the light source, one can calculate the photon dose required to achieve the action. The inverse of the dose required to fulfill the energy—(\left( {{\text{i}}. {\text{e}}.;efficiency = \frac{1}{dose}} \right) \ )—at each wavelength provided an action spectrum for each of the electrostatic parameters. These action spectra were then normalized to 405 nm to provide context relative to the visible radiation results. High-content imaging data were averaged on a per-wavelength basis, then normalized to the UV-free “dark” control prior to action spectrum generation. As previously highlighted, a linear regression analysis was performed on the results from the three irradiance conditions (full irradiance, ND02, ND04) for each wavelength and the slope factor used to determine response efficiency. Each exposure was repeated three times for each irradiation condition with four technical replicates for each wavelength. The dataset was simplified using K-means clustering, supplemented by consensus modeling based on Monte-Carlo simulation, facilitated by the R package M3C11, to provide empirical justification for the optimal value ‘K’. All statistical analyzes were performed using R-Studio26 and Excel 2016 (Microsoft systems, CA, USA), with data derived from three biological replicates each including four technical replicates.
