Temperature Is a Key Factor Governing the Toxic Impact of Ultra-Violet Radiation-Emitting Nail Dryers When Used on Human Skin Cells

Cell Culture

E6/E7 immortalized Human Foreskin Keratinocytes (E6/E7-HFK) were grown and maintained in keratinocyte serum-free media (KSFM 1X, Gibco) supplemented with 25 mg Bovine Pituitary Extract (BPE), 2.5 µg Human Recombinant EGF, and 1% Penicillin/Streptomycin (PenStrep, Gibco). Immortalized Normal Human Dermal Fibroblasts (NHDF) were grown and maintained in Dulbecco’s Modified Eagle Medium (DMEM, Gibco), supplemented with 10% Fetal Bovine Serum (FBS, Peak Serum) and 1% PenStrep. All cells were cultured at 37°C in a 5% CO₂ incubator.

UV Exposure Parameters

A 54W MelodySusie (MS) UV nail dryer (MelodySusie, Model: DR-6332B EOS9, Newark, CA) was used to determine the impact that commercial UV nail dryers have on the skin. This model of UV nail dryer houses 30 UV-LED lights that emit UVA light at 365 nm. As a comparative control, the LED UVR DNA damage induction system (LUDIS) was used with output calibrated using a microphotometer (International Light Technologies, Peabody, MA). The LUDIS is a high-throughput light system that allows for rapid and precise UVA (365 nm) exposure to cells in a 96-well format. The LUDIS platform was developed originally as a novel method to test compounds that may have utility as new sunscreen active ingredients. The use of UV-emitting high-powered LED lights is a major hardware innovation of the platform. Current UV-delivering devices (e.g., Philips TL-40 UVB; 40 J/m²s) are cumbersome with no ability to be used in high-through-put applications, having been in operation with essentially no major improvements for decades. Commonly used sources, such as mercury or neon lamps, are prone to wide-bandwidth UV output, making it particularly difficult to discern the relative contribution of a wavelength (UVA2 vs. UVA1) to inducing specific forms of DNA damage. Recent advances in LED technology have created microbulbs (like the lights that are used in the LUDIS) with very high light output with low heat induction emitting a narrow, very stable wavelength. UVA1 fluence must have physiological relevance to typical human exposure. While UVA irradiance fluctuates with weather, seasons, and time of day, roughly 5% of the 1000 J/m² terrestrial solar radiation is UVR, of which 70-75% is within the UVA1 waveband. Thus, “typical” solar UVA1 irradiance (or flux density, sometimes thought of as “intensity”) is roughly 37.5 W/m² (=1000 J/m² x 5% × 75%). Thus, our highest fluence “dose” of 14.2 J/m² represents a level of irradiation achievable in approximately one-hour (=142,000 J/m² ÷ 37.5 J/m²s ÷ 3600 s/h) midday sun exposure during the average day. For frame of reference, specific examples of measurements yielded a flux density of approximately 47 W/m² at summer solstice in Chilton, UK (51.6°N). The output of the LUDIS and MS was measured using a calibrated microphotometer (International Light Technologies, Peabody, MA). All experiments were conducted in a laboratory with no outside windows, stable temperature, and shielding from other artificial light sources.

A major component of this research was to test different experimental parameters. The 3 exposure parameters used in this study were termed Acute, ⁹ Chronic, ⁹ and Salon. Acute exposure was 2 × 20-min UVA exposures with a one-hour recovery between treatments. Chronic exposure was 3 × 20-min UVA exposure with 24-hr recovery between treatments. Salon exposure was designed to simulate real exposure to UV nail dryers found in salons and was 6 × 30 sec exposures, with 2-min recovery between treatments.

Cell Viability and Cytotoxicity Assays

Cells were seeded in 96-well plates at 10⁴ cells/well and were left overnight in a 37°C incubator prior to UVA exposure. Cells were then exposed to Acute, Chronic, or Salon UVA exposure as described. Viability measurements were conducted 48 hr after final exposure; fibroblast viability was measured using an XTT Cell Viability Assay Kit (Biotium, Fremont, CA, USA). Plates were incubated at 37°C for 4 hr, and absorbance was read at 450 nm using the BioTek Cytation7 cell imaging multimode reader. Keratinocyte viability was measured using DAPI stain after cells had been fixed with 4% paraformaldehyde and permeated with 0.2% Triton. Plates were incubated for 5 min at 4°C, and cell count was taken using the BioTek Cytation7 cell imaging multimode reader.

Assessment of Superoxide Formation with Dihydroethidium

To measure ROS accumulation in the cells exposed to UVA, Dihydroethidium (DHE) was used. ¹⁰ Cells were seeded at roughly 70000 cells per well in a black-sided, optical-bottom 96-well cell culture plate and were left overnight. Working concentrations of N-acetylcysteine (NAC) (2 mM), Antimycin A (50 µM), and DHE (10 µM) were prepared in phenol- and serum-free media. DHE alone was used to measure endogenous ROS, Antimycin A was used as a positive control, and NAC was used as a negative control. DHE at 10 µM was added to both the positive and negative control solutions. All media was removed, and cells were washed with 100 µL of pre-warmed serum- and phenol-free media before all working reagents were applied and placed in a 37°C incubator for 20 min. After incubation, a preliminary ROS reading was taken prior to cells being exposed to 0-14.2 J/cm² UVA. A final ROS reading was taken immediately after UVA exposure. Endogenous ROS was subtracted from UVA-induced ROS in order to accurately visualize the amount of ROS induced by UVA.

Assessment of DNA Damage Using the Alkaline Comet Assay

To measure the amount of DNA damage accumulated in the cells after UVA exposure, the cells were run in an alkaline comet assay. ¹¹ Cells were seeded in 96-well plates at 10⁴ cells/well and were left overnight in a 37°C incubator prior to UVA exposure. Cells were then exposed to Acute, Chronic, or Salon UVA exposure as described. Immediately after the final treatment of UVA, cells were gravity-loaded into the comet chips (Biotechne, Minneapolis, MN) for 15 min. After loading, the comet chip was washed with PBS and sealed with 0.8% low-melting-point agarose (LMPA) before being placed in pH 10 lysis buffer solution for 40 min. The comet chip was then transferred to alkaline buffer for 40 min, with fresh alkaline buffer being replaced after 20 min. The comet chip was subjected to electrophoresis for 50 min with a constant setting of 22 V. Immediately after electrophoresis, the comet chip was neutralized in pH 7.5 400 mM Tris-HCl prior to staining with SYBR green diluted in 40 mM Tris-HCl. Comet chips were imaged using the BioTek Cytation7 cell imaging multimode reader, and tail DNA was measured using the Trevigen Comet Assay Analysis Software.

Measuring the Impact of Temperature on Cell Viability

To determine the role that temperature plays on cell viability, the temperature inside the MelodySusie (MS) unit was measured during the acute exposure protocol. Initially, the temperature was tested using a generic thermometer at two locations within the MS unit. From this, it was determined that indeed there were temperature changes. To measure temperature fluctuation more precisely, we used a pico-temperature measurement system. This apparatus measured temperature in the LUDIS and MS units over a 20-min irradiation in 8 distinct well locations. Temperatures were within the 40-65°C range. To determine the impact the high temperature had on cells, both keratinocyte and fibroblast cells were seeded in 96-well plates at 10⁴ cells/well and were left overnight in a 37°C incubator. To replicate and examine the effect of the temperature changes in the MS device in the absence of UVA, cells were then placed in an incubator set to 55°C for 20 min (twice, with a 60-min break between treatments to replicate the acute exposure protocol), or 31°C for 15 min (to align with the 6 × 30 sec salon exposure schedule), before being returned to a 37°C incubator. Measurements were conducted 48 hr after final exposure; NHDF viability was measured using an XTT Cell Viability Assay Kit (Biotium, Fremont, CA, USA). Plates were incubated at 37°C for 4 hr, and absorbance was read at 450 nm using the BioTek Cytation7 cell imaging multimode reader. E6/E7 viability was measured using DAPI stain after cells had been fixed with 4% paraformaldehyde and permeated with 0.2% Triton. Plates were incubated for 5 min at 4°C, and cell count was taken using the BioTek Cytation7 cell imaging multimode reader. Finally, to determine the impact of temperature changes within the MS unit (localized regions of elevated temperature), a plate of cells was covered in aluminum foil and placed in the MS unit for an acute 20-min protocol. Cells were allowed to recover for 48 hr before cell viability was measured.

Statistical Analysis and Assay Controls

Statistical evaluation of data was conducted initially using Excel (Microsoft) to annotate and organize raw data that was generated from the cell viability and comet analysis assays. The data was then exported to a dedicated statistical analysis and visualization software (Prism version 10.1.2, Graphpad, Los Angeles, CA). Statistical approaches used in this research include one-way ANOVA with post-hoc analysis unless otherwise specified. In Figure 1(F), the modeling was changed to a mixed effect analysis (REML) without the assumption of sphericity and Dunnett’s multiple comparison testing and resulted in the same adjusted P value of <0.0001. Figures 4(D) and 4(G) show the two sets of data on each of the graphs that passed normality testing (Shapiro–Wilk, Pearson, Anderson–Darling, and Kolomogorov), with each irradiation time point compared between acute and chronic exposures using one way ANOVA with multiple comparisons corrected for using Tukey post-hoc testing. Statistical significance was measured in Figure 4(D), A9.8 vs. C9.8 adjusted P value of <.0001, A14.2 vs. C14.2 adjusted P value of .013 (*). In Figure 4(G), A9.8 vs. C9.8 adjusted P value of <.0001, A14.2 vs. C14.2 adjusted P value of .0047 (**). The Figure 5(F) data were analyzed using Kruskal–Wallis multiple comparison testing with Dunn’s post-hoc test. Statistical significance was gauged as * = P < .05, ** = P < .01, *** = P < .001, and **** = P < .0001.

The DNA damage measurements were conducted using company-established Standard Operating Procedures (SOPs), including the inclusion of positive and negative controls as previously described in Ref. 12 and also in Ref. 11. One of the many innovations presented herein is the use of stable DNA damage controls. These positive controls have been described in the publication ¹² above and are cells that have been exposed to etoposide and then fixed. They have a stable level of DNA damage that is used to normalize multiple comet assays. There is no negative control for the comet assay as all cells have some level of endogenous DNA damage, nor is a blank no-cell control appropriate for this methodology.

Measurement of Cell Viability in Human Cells after Exposure to UVA from a Commercial Nail Dryer Unit

We investigated the irradiation characteristics of the MelodySusie UV-curing unit, exposing non-isogenic human keratinocytes (E6/E7) and normal human dermal fibroblasts (NHDF) to three different treatment protocols (Figure 1). The acute treatment protocol was 2 × 20-min irradiation with 60 min between the two treatments, while chronic exposure protocol was 3 × 20-min exposures with 24-hr recovery between each of the three treatments. ⁹ The third treatment protocol was based on a typical salon setting, with each (or all) nail(s) cured immediately after the coating is applied (5 × 30 sec) and 2 min between treatments (during which time the next nail is coated). The last cure (30 sec, 6 × 30 sec total) hardens the protective sealant coating on all five nails; this salon protocol is depicted in Figure 1. The MelodySusie (MS) unit has a total light output of 54 W and 3 pre-programmed durations of continuous exposure (30, 60, and 90 sec). Depicted in Figure 2(A), the LED configuration of the machine is arch-shaped to sit above the locations of each nail of a palm-down flat hand. The curing LEDs emit UVA at 365 nm but are not all at a uniform height or density, resulting in the unit emitting approximately three times more UVA (5 to 15 mW/cm²/sec) at the rear than at the front (Figure 2(B)), as measured using a handheld photometer with a microscale UVA-calibrated detector. We irradiated human-derived keratinocytes, and cell viability was measured after 48 hr of recovery (Figure 2(C)-(F)). After acute exposure, high levels of cell death correlated with LED location in the MS unit (Figure 2(C)). That is, no loss of viability was measured at the front of the unit (wells H5-7). In contrast, viability dropped to below 10% of control values in wells directly below the main arch-shaped array of LEDs (R) in the unit. A similar pattern of cell viability was achieved when using the 3-day chronic exposure schedule (Figure 2(D)). While these exposure protocols are informative regarding the general cytotoxicity of nail dryer LEDs, they do not model a typical level of LED exposure likely encountered when using these units in a nail salon setting. To reproduce the conditions more likely to be encountered in a nail salon, we exposed keratinocytes to a third experimental paradigm (6 × 30 sec exposure with 120 sec between exposures). This lower level of irradiation did not cause a reduction in cell viability (Figure 2(E)-(F)). Further, the three exposure protocols resulted in significantly different viability results (Figure 2(F)). The chronic exposure protocol reduced average viability to the greatest extent (25 ± 9%; P < 0.0001) compared to acute (39 ± 16%) and salon (142 ± 11%) irradiation protocols. In brief, when the unit was used according to the manufacturer’s specifications, there was no reduction in cell viability.OPEN IN VIEWER

UVA can penetrate through the epidermis into the lower dermal layer containing predominantly fibroblasts. Using the previously described three treatment schedules (Figure 1), we investigated the impact of the UVA exposure on normal human dermal fibroblast (NHDF) cells. Acute exposure resulted in an approximately 50% decrease in fibroblast viability (Figure 3(A)) that, unlike keratinocytes, was independent of location of cell wells in the dryer. In contrast, chronic treatment of the NHDF cells resulted in a pattern of reduced viability in wells directly below the UVA LEDs (Figure 3(B)), as previously seen in the keratinocytes (Figure 2(D)). Under salon conditions, the NHDF cells showed a reduction in viability correlative to positional LED density (Figure 3(C)), although the pattern was less defined than that observed using the chronic treatment protocol (Figure 3(B); summarized in Table 1).OPEN IN VIEWER

OPEN IN VIEWER

Table 1.

Cell Viability After the Three Different Exposure Protocols.

	Acute (%)	Chronic (%)	Salon (%)
Keratinocyte	39 ± 16	25 ± 9	142 ± 11
Fibroblast	36 ± 3	37 ± 16	63 ± 18

Determining the Amount of UVA Exposure at the Irradiation Surface of the MS Unit

Previous methods to accurately measure the output from nail dryers have yielded varied results. ¹³ This variability can be partly attributed to the non-uniform light distribution and density in commercial nail dryers (Figure 2(A)). We therefore evaluated different methods to measure the fluency inside the MS unit. Initially, we trialed moving a UV-detecting sensor in a designated grid pattern within the MS irradiation zone. However, the size of the sensor affected the light scatter patterns within the machine, and the sensor was not at the same irradiation height as the cells. We then sought to correlate the output to a device with a previously determined UVA output. The LED UV-DNA damage induction system (LUDIS) is an in-house patented prototype that was built by our consortium as part of a body of research investigating the impact of UVA on human skin. The LUDIS LEDs are in a 96-well configuration (Figure 4(A)). Similar to the MS unit, the LUDIS contains UVA (365 nm)-emitting LED bulbs. Each LUDIS LED has an individual light-directing channel (light cone) to prevent light from scattering into adjacent wells. We tested how the keratinocytes and fibroblasts responded to UVA irradiation from the LUDIS using the Acute exposure schedule. Cells were exposed to 0 (Rows A-B), 4.5 (C-D), 9.6 (E-F), or 14.2 (G-H) J/cm² (Figure 4(A)). Keratinocyte viability after LUDIS UVA exposure was similar to the MS results (Figure 4(B)-(C)) with chronic treatment inducing greater loss of cell viability than acute treatment (Figure 4(D)) at 9.6 and 14.2 J/cm². The E6/E7 used in this study is particularly amenable to high-throughput screening platforms but has aberrant p53 and pRB pathways. ¹⁴ The DNA damage induction profile of E6/E7 cells after UVA1 irradiation (acute) was compared to normal human adult keratinocytes (NHEKa) that were not immortalized (ATCC, PCS-200-011) (Figure S1). The amount of UVA1-induced DNA damage was higher in the NHEKa cells but with a comparable pattern of response. The LUDIS exposure experiments were also carried out using fibroblast cells under identical experimental conditions. As reported for the acute MS fibroblast results, cells did not respond to the acute UVA exposure in a linear manner with increasing fluency correlating with decreasing viability. Rather, the fibroblast gave a binary response, with the acute lowest exposure (4.5 J/cm²) and high exposure (14.2 J/cm²) inducing a comparable reduction in overall cell viability (Figure 4(E)). In contrast, the chronic LUDIS exposure in the fibroblasts did induce a reduction in cell viability correlating with UVA fluency (Figures 4(F) and 4(G)). The LUDIS linear response data from the keratinocytes was then used to create standard curves (Figure S2A and S2B). The linear regression equations derived from these curves were then used to calculate the absolute UVA exposure level (J/cm²) for each well in the MS unit (Figure S3A and S3B). However, depicted as in Figure S3C, the two exposure protocols significantly differed (overall 2.6 ± 4.0 J/cm²), particularly along the left edge of the plate (columns 1 and 2, difference 4.4 ± 1.5 J/cm²) and in the UV-LED arch area (5.4 ± 3.3 J/cm²). From these data, we concluded that a second endpoint would be needed to enable cross-comparison, potentially improving the precision of the predictive approach.OPEN IN VIEWER

The Role of Oxidative DNA Damage in the Loss of Viability after UVA Exposure

Oxidative DNA damage was evaluated as a second endpoint to correlate UVA exposure between the LUDIS and MS units. UVA can cause DNA damage through indirect mechanisms, including the induction of reactive oxygen species (ROS). Intracellular superoxide, a potent ROS molecule, was measured after exposure to UVA (Figure 5(A)). As per previous reports, ⁹ UVA exposure elevated levels of superoxide. This impact was observed in both fibroblasts and keratinocytes, with levels of induced superoxide plateauing at higher fluency (>5 J/cm²). We investigated whether levels of DNA damage correlated with superoxide induction using single cell gel electrophoresis (SCGE, aka comet assay). SCGE allows for the visualization of DNA damage, including lesions caused by oxidative damage, such as AP-sites and single-stranded breaks (SSB). Acute exposure of keratinocytes using LUDIS resulted in significantly lower levels (P = .043) of DNA damage than that of 3-day chronic treatment (Figure 5(B)). We then replicated the experiment in the MS unit. The pattern of DNA damage was correlative with the position of the LEDs; however, there were regions of high DNA damage at the periphery of the plate after any of the 3 (acute, chronic, and salon) irradiation protocols (Figure 5(C)–5(E)). Overall, the salon exposure schedule caused significantly less DNA damage than that of the acute (P > .001) and 3-day chronic (P > .0001) treatments. Of note, the chronic exposure caused more DNA damage than measured after the acute treatment (Figure 5(F)). In comparison to the keratinocytes, the LUDIS induced significantly (P > .0001) higher levels of DNA damage in the fibroblasts after acute and chronic exposure (Figure 5(G)). Using the MS for fibroblast, UVA irradiation caused many wells to reach DNA damage levels, which were deemed to be beyond assay linearity (>60%) (Figure 5(H)–(J), (K)). Because the acute and chronic measurements went beyond assay linearity, further statistical analysis of the fibroblast data was not appropriate. Overall, the fibroblasts were more susceptible to UVA-induced DNA damage compared to the keratinocytes (P > .0001), yet the pattern of DNA damage induction did not clearly correlate with the arch-shape of the LEDs. We used the acute LUDIS keratinocyte data to determine whether the measured linear DNA damage (r² = 0.98) could correlate and predict the UVA output of the MS unit. To this end, a standard curve was created from the LUDIS DNA damage results (not shown), and this curve was used to predict the UVA exposure in the MS unit (Figure S4A). The DNA damage levels did not correlate with the position of the LEDs; rather high levels of DNA damage were measured at the plate peripherals. Hence, the fluency calculations also estimate high UVA levels in these areas of the plate. This irradiation prediction was compared to the acute keratinocyte cell viability result (Figure S4B) with the difference between the two results presented as Figure S4C. As with the cell viability correlation, but using DNA damage as an endpoint, we were not able to predict the level of UVA exposure within an acceptable level of accuracy. Specifically, the calculations overestimated the amount of DNA damage around the periphery of the plate, in particular in row B of the acute treatment; this row sits closest to the back of the MS unit. As with the cell viability predictions, the DNA damage prediction data suggested unknown additional factors beyond UVA exposure were influencing both DNA damage and cell viability in the MS unit.OPEN IN VIEWER

Considering the lack of onboard cooling in the MS, we investigated whether the zone of irradiation inside the unit had fluctuations in temperature. To measure changes in temperature, we used a thermometer inside the MS unit. Towards the opening of the machine (F), temperatures increased over the 20-min exposure duration to a maximum recorded heat of 45°C (Figure 6(A)). This correlated to the area with the highest viability after exposure (Figures 2 and 3). When the probe was placed at the back of the unit, the temperature reached a maximum of 55°C (Figure 6(A)). The magnitude of temperature change was unanticipated and required more precise analysis. To this end, a digital microprobe temperature detection system was used. The internal well temperature rose above 60°C during the acute treatment protocol in the MS (Figure 6(B)). To determine the impact this temperature change may have had on the cell viability, a plate of fibroblast cells was put in an incubator pre-warmed to 55°C for 20 min. The fibroblasts showed a well, row, and column-independent reduction in viability with an average reduction in viability of approximately 50% (Figure S5A). The incubator cell viability experiment was repeated using the salon protocol. This also reduced the cell viability by an average of 25% (Figure S5B). In contrast to the position-independent decrease in cell viability observed in the fibroblasts, the keratinocytes showed an exposure pattern that showed the greatest loss of viability (near 100% loss) at the peripherals of the plate while the center of the plate had a 50% loss of viability (Figure S5C); in these cells, there was a clear gradient from the outside to the center of the plate, presumably as a result of more rapid warming of peripheral wells in the incubator. To complete the set of temperature experiments, we covered a plate of fibroblast cells in a single layer of aluminum foil and exposed the plate in the MS under the acute protocol conditions. Aluminum foil was selected because it transfers heat readily and completely blocks the UV light. After exposure, the cell recovered for 48 hr, and cell viability was measured (Figure 6(D)). A loss of cell viability was measured at the peripherals of the plate similar to Figure S5C, with the greatest loss of cell viability measured at the left peripheral side of the plate. The experiment confirmed that using the MS induces an internal temperature increase that impacts cell viability independently of UVA exposure. This temperature increase is not ubiquitous throughout the unit and impacts both UVA-induced loss of cell viability and also, to a greater extent, levels of induced DNA damage. In the research conducted by Zhivagui et al., ⁹ the authors used a larger format 6-well plate rather than the 96-well format used herein. To determine whether the temperature fluctuations were an artefact of using the smaller well format, we also measured the temperature using a generic 6-well plate (Figure S6A-B). The wells irradiated at the front of the unit (A1, A2, and A3) all had a significant increase in temperature after 20 min in the nail dryer. This was highest for well A3 that had a peak temperature of over 50°C. In contrast, the three rear wells did not have a significant increase, suggesting that differences in heat dissipation patterns and the pattern of LED lights in the unit may play a major role in the elevated temperatures measured in both plate formats. The ramifications of elevated temperature on 2D mammalian cell culture and DNA damage are well established and will be discussed further herein.OPEN IN VIEWER