Comparative analysis of the phototoxicity induced by BRAF inhibitors and alleviation through antioxidants

Small molecules tackling mutated BRAF (BRAFi) are an important mainstay of targeted therapy in a variety of cancers including melanoma. Albeit commonly reported as side effect, the phototoxic potential of many BRAFi is poorly characterized. In this study, we evaluated the phototoxicity of 17 distinct agents and investigated whether BRAFi‐induced phototoxicity can be alleviated by antioxidants.


| INTRODUC TI ON
The identification of somatic mutations of the BRAF gene has paved the way for targeted therapy with small molecules in a variety of cancer entities in recent years. 1 Activating genetic alterations of codons encoding for the kinase domain of BRAF results in constitutive oncogenic signaling through the Ras-Raf-MEK-ERK mitogen-activated protein kinase (MAPK) pathway, leading to cell proliferation and, ultimately, tumor growth. 2 Mutations of BRAF have been found in more than 66% of human cancers and are suspected to be present in an even wider range of malignancies at a lower frequency. 3 Encouraged by these observations, major pharmaceutical efforts have been made at high pace to develop small molecules targeting mutant BRAF. As the target protein with and without the mutation of interest could be structurally elucidated in its active and inactive conformation via crystallography, a hitherto never utilized drug design strategy was employed, namely fragment-based lead discovery.
Here, smaller well-binding structures are joined covalently to form a superiorly target-binding drug candidate. 4 As a result, the BRAF inhibitors (BRAFi) vemurafenib and dabrafenib were developed and approved by the FDA for the treatment of metastatic or unresectable melanoma in 2011 and 2013, respectively, showing substantial survival benefits compared with chemotherapy. 5,6 In 2018, a third inhibitor, encorafenib, has been approved in combination with the MEK inhibitor binimetinib. 7 All substances selectively bind to and inhibit the active-state BRAF kinase, with most BRAFi sharing common structural motifs: the A ring binding in the nucleobase-binding pocket, the B ring as a sterically important stiff core, the BC linker (salt bridge linker) for ionic interactions, and the lipophilic C ring. 8 As the A ring resembles the aromatic, bicyclic adenine-moiety of the native substrate ATP, most inhibitors rely on a mono-or bicyclic, heavily substituted aromatic structure for strong binding characteristics. For this reason, most inhibitors exhibit strong UVA absorbance which is a prerequisite for UVA-induced phototoxicity. This cutaneous adverse event is well known for vemurafenib, while the phototoxic potential of dabrafenib and encorafenib is much lower in pivotal trials. [9][10][11][12][13] However, the phototoxic potential of other BRAFi is poorly characterized and has not been analyzed in a systematic approach yet. Therefore, in this study we comparatively evaluated the phototoxicity of 17 distinct BRAFi or multikinase inhibitors and tested whether phototoxicity can be reduced by antioxidants in vitro.

| Reactive oxygen species assay
The reactive oxygen species (ROS) assay was developed to test for the generation of a reactive species from chemicals following ab- and sulisobenzone (SIB) were used as positive and negative controls, respectively. 15 The test procedure for this assay included a 1.5 mL micro tube and a plastic clear flat bottomed 96-well microplate. The reaction mixtures were prepared by vortex mixing under UV-cut illumination. For each reaction mixture, triplicates of 200 µL per well were transferred into a 96-well plate and solubility and coloration were checked microscopically at 100-fold magnification. After shaking the plate for 5 seconds, baseline absorbance at 440 nm (A 440 ) for SO and 560 nm (A 560 ) for SA was measured prior to UV exposure.
The plate was irradiated with a UVA simulator for 1 hour, and A 440 and A 560 were measured again. Based on the assay protocol, SO was determined as a result of bleaching of p-nitrosodimethylaniline by oxidized imidazole. The measurement of SA was made upon the reduction of nitroblue tetrazolium 16 :

| In vitro 3T3 NRU phototoxicity test
Identification by this test increases the likelihood of substances to be phototoxic in vivo after systemic or topical application. 17 Phototoxicity of the BRAFi was determined according to the OECD/ OCDE 432 guideline with minor modifications. 18 For the experiments, 3T3 cells were seeded into 96-well plates at a density of After incubation of two identical 96-well plates, one was exposed to either UVA (total dose: 5 J/cm 2 ) or UVB (total dose: 20 mJ/cm 2 ) light and the other one was covered in lightproof aluminum foil and incubated under the UV lamp as well. Subsequently, the cells were washed with PBS and incubated in DMEM supplemented with FBS at 37°C overnight. On the following day, cells were washed with PBS and incubated in DMEM without FBS containing 50 µg/mL neutral red (NR) dye at 37°C for 2 hours. Cells were washed with PBS and blotted to remove buffer remains. Precisely, 150 µL desorb solution (50% ethanol v/v, 1% acetic acid (v/v)) was added per well and the plate was incubated at room temperature for 10 minutes with gentle shaking. The absorbance of the resulting homogeneously pink solution was measured without a lid at 540 nm in a plate reader (Spectra MR, Dynex Technologies). The outer wells of each plate were used as reference.

| Phototoxicity alleviation through antioxidants
To test for an effect on the BRAFi-mediated toxicity, antioxidants were added to the 3T3 NRU phototoxicity test of selected inhibitors. The concentrations of the tested antioxidants were 100, 10, and 1 µmol/L for vitamin C, vitamin E phosphate, and trolox, and 10 mmol/L, 1 mmol/L, and 100 µmol/L for GSH. The concentrations of the BRAFi vemurafenib, dabrafenib, and encorafenib were 3.16, 1, and 10 µmol/L, respectively. where the "photo-irritation-factor" (PIF; ratio of IC 50noUV to IC 50UV ) was used to estimate the risk. 18 Based on the validation study, a PIF value less than 2 predicts no phototoxicity, a value between 2 and 5 probable phototoxicity, and more than 5 phototoxicity. 17 In some cases, not both IC 50 could be determined, so that the published ad hoc rules were applied (prediction model 1): First, if only one IC 50 can be measured, the other value is replaced by the highest concentration tested; the chemical is considered phototoxic if the ratio is greater than 1. Second, if no IC 50 can be measured, the chemical is considered non-phototoxic. 19 The correlation between the formation of SO or SA in the ROS assay and phototoxicity observed in the 3T3 NRU assay was calculated with Pearson's correlation.

| Data evaluation
Differences of phototoxicity after the addition of antioxidants were compared with the student's t test. A two-sided P-value was calculated in all cases and values of P < .05 considered as statistically significant.

| UV spectral analysis of BRAF kinase inhibitors
Initially, the UV absorbance spectrum of the 17 Raf inhibitors was determined. All inhibitors showed a certain amount of UVA (λ UVA = 315 to 410 nm) and UVB light (λ UVB = 280 to 315 nm) absorbance ( Figure 1A). The substance with the highest absorption of UVA light was GW5074. The area under the curve (AUC) of this compound was set to 1 for UVA ( Figure 1C). It was closely followed The highest absorbance of UVB was observed for NVP-BHG712 whose AUC and relative absorption was set to 1 ( Figure 1C).
Compared with its absorbance of UVA, sorafenib tosylate displayed a much higher absorption of UVB (relative absorption 0.5) in our spectral analysis. Interestingly, the highest amount of UVB absorption amongst the approved BRAFi was recorded again for vemurafenib (relative absorption 0.8) and the lowest for encorafenib (relative absorption 0.2) ( Figure 1B). The absorbance curve for vemurafenib showed a peak at the border between UVB and UVA and then decreased after 340 nm ( Figure 1A and B).

| In chemico generation of reactive oxygen species upon exposure to UVA and UVB light
In chemico SO and SA generation induced by BRAFi upon exposure to UVA and UVB was analyzed. Dabrafenib was the only chemical showing increased SA levels after UV irradiation while also leading to increased SO formation. Only 3 out of 17 compounds generated SO species following absorption of UVB light, whereas, after absorption of UVA light, 9 chemicals including encorafenib resulted in increased levels of SO. Interestingly, vemurafenib and sorafenib tosylate belonged to the 7 substances where no phototoxic response was detected (Table 1).

| In vitro 3T3 NRU phototoxicity assay
We compared the IC 50 values acquired in the light and dark experiments by calculating the PIF. Since most BRAFi showed no toxicity

| Phototoxicity alleviation through antioxidants
Comparing the results of the ROS and the 3T3 NRU phototoxicity assay, we detected a significant correlation between the phototoxicity observed in the 3T3 NRU assay and the formation of both SO (Pearson r = .5365; P = .026) and SA (Pearson r = .9139; P < .001) in the ROS assay. To test for an inhibitory effect on the BRAFi-mediated toxicity, antioxidants were added to the 3T3 NRU phototoxicity test of selected inhibitors. The phototoxicity of the approved BRAFi vemurafenib, dabrafenib, and encorafenib was analyzed in the presence or absence of UVA light for different concentrations of glutathione (GSH), trolox, vitamin C, and vitamin E. Of all tested antioxidants, high-dose GSH was able to fully rescue the phototoxicity observed with vemurafenib after exposure to UVA ( Figure 3A). In contrast, the phototoxicity induced by dabrafenib and encorafenib was left unaltered after antioxidants were added at different concentrations ( Figure 3B and 3C). The generation of reactive oxygen species following UV light irradiation can lead to oxidative damage to the cell. UVA light plays a more significant role in causing phototoxicity than other UV ranges. 22,23 This is consistent with the results of the ROS assay that was performed in this study with UVA and UVB light. Only 3 out of Inhibitor ROS assay +UVB +UVA  with UVB light compared with 9 agents which generated ROS after UVA light absorption. Surprisingly, vemurafenib elicited no ROS release in our assay, although its UVA-dependent phototoxicity is well established in daily care and in the literature. 9,24 These results imply that the phototoxicity observed with vemurafenib clinically is not mediated by ROS. In contrast, we observed that dabrafenib, which in the clinical practice is considered much less phototoxic than vemurafenib, was the only substance leading to the formation of both SO and SA after UVA and UVB exposure. Thus, our results fit well with a recent study, in which vemurafenib but not dabrafenib impaired the repair of UV-induced DNA damage in keratinocytes. 25 These results suggest that the phototoxicity experienced by patients under BRAFi in vivo may not primarily be mediated by ROS formation and imply that also other mechanisms are likely to be involved. Above that, we conclude that the ROS assay may not accurately predict the phototoxicity which is clinically relevant and observed in vivo.

Singlet oxygen Superoxide anion Singlet oxygen Superoxide anion
Another test which has proved to be predictive of acute phototoxicity effects in animals and humans in vivo is the 3T3 NRU phototoxicity test. 18  To alleviate the BRAFi-induced phototoxicity, we tested a panel of antioxidants as an attempt to find feasible alternatives for the management of phototoxicity. 28,29 Accordingly, the physiological antioxidants vitamin C, vitamin E, trolox, and GSH were added to the 3T3 NRU phototoxicity assay with the BRAFi vemurafenib, dabrafenib, and encorafenib. Interestingly, we observed that high-dose GSH was able to fully rescue the UVA-induced phototoxicity of vemurafenib. GSH is a tripeptide best known for its role as antioxidant by neutralizing ROS. However, as no induction of ROS was observed with vemurafenib, it is likely that other functions of GSH are involved. It has general cytoprotective properties and can stabilize cellular components after DNA damage. 30 Thus, our data support a model where UV-induced damage repair is impaired by vemurafenib and that this process can be alleviated by GSH, independently from the generation of ROS. Although our understanding of the protective effects of GSH on the vemurafenib-induced phototoxicity is certainly limited, substances that increase GSH may represent an interesting option to protect patients from phototoxicity also in vivo.
Further studies are warranted to fully explore the potential of GSH in both the treatment and prevention of phototoxic reactions due to vemurafenib.

ACK N OWLED G EM ENTS
None.