Precise photopharmacological eradication of metastatic tumor cells

Chemotherapeutics – including alkylating agents, antimetabolites, topoisomerase inhibitors and microtubule targeting agents – are the mainstay in current cancer treatment plans because these agents have proven to be highly efficient against tumor cells (Gascoigne and Taylor, 2009; Grünewald et al., 2018). Nevertheless, as these drugs also act on healthy cells, they have dose-limiting toxicities and generate severe adverse effects, such as cardiotoxicity and neuropathy, or can even cause secondary malignancies and infertility (Albini et al., 2010; Leone et al., 2001; Poorvu et al., 2019; Seretny et al., 2014).

Thus, one challenge for improving cancer therapy is the development of new agents with high tumor cell selectivity and minimized adverse effects on healthy tissue. Many strategies have been developed towards this goal, including targeted drugs designed to exploit cancer-specific vulnerabilities, e.g. antibody–drug conjugates, which specifically bind to tumor cells and deliver toxins or trigger phagocytosis by macrophages, and chimeric antigen receptor (CAR) T cells, which bind and lyse tumor cells (Fu et al., 2022; Hong et al., 2020; Tsimberidou et al., 2020). Although successful agents have emerged from these concepts, limiting their effects to cancer cells is still an issue, and identifying epitopes to make antibodies or CAR T cells completely cancer-cell specific remains a major challenge and area of active research. Furthermore, the heterogeneity of tumor cells makes it difficult to identify a common ‘Achilles' heel’ for targeted intervention (Dagogo-Jack and Shaw, 2018).

Photopharmacology offers an alternative strategy to limit the effects of small compounds to tumors by adding a light-responsive molecular switch to highly efficient anticancer compounds in order to control their pharmacological activity in space and time with unrivaled resolution (Hull et al., 2018; Kobauri et al., 2023). Photopharmacology and the application of light-activatable compounds is an appealing concept for precise targeting of cancer cells. To render small molecule drugs light activatable, light-sensitive ‘photocage’ motifs are often added at crucial sites of the compound to block its activity until illumination [typically, ultraviolet (UV)] irreversibly removes the cage (Klan et al., 2013; Monteiro et al., 2021). Alternatively, photoswitchable compounds can be engineered to switch from inactive to active conformations when exposed to a specific wavelength without releasing any byproducts (Gao et al., 2021, 2022). Photoswitches are often reversible in the sense that the compounds can be switched back to the inactive structure by illumination with another wavelength, or through spontaneous relaxation (Kobauri et al., 2023). Because the activity of photoswitchable drugs can be controlled with spatial and temporal precision, they can be administered in their inactive state at much higher doses than constitutively active drugs. Spatially confined activation is then able to achieve high local doses of the active compound while keeping systemic levels low (Fuchter, 2020; Hull et al., 2018; Velema et al., 2014). Importantly, the concept of using light against tumor cells is already exploited by photodynamic therapy (PDT) in the clinics (Kwiatkowski et al., 2018), with PDT specifically relying on the generation of reactive oxygen species (ROS) upon illumination of photosensitizer compounds.

Adding to the portfolio, SBTubA4P, a photoswitchable compound derived from the microtubule inhibitor combretastatin A4 phosphate (CA4P), was recently developed (Gao et al., 2022). SBTubA4P can be switched from its biologically inactive trans-configuration (E) to its active cis-configuration (Z) by UV light, which binds to tubulin and blocks polymerization of microtubules, analogous to clinically applied compounds – such as colchicine and related agents – that have entered late-stage clinical trials (e.g. combretastatin, ZD6126, ombrabulin) (Tozer et al., 2008). We previously demonstrated that localized E-to-Z-photoswitching of SBTubA4P can be applied for spatially precise short-term inhibition of microtubule polymerization in vitro and in vivo in a variety of models, including zebrafish larvae (Gao et al., 2022). Here, we explore the potential of SBTubA4P for targeted elimination of metastatic tumor cells using zebrafish xenografts as a preclinical cancer model. We chose osteosarcoma and Ewing sarcoma as tumor entities for xenograft experiments owing to the high clinical need to develop novel therapeutic strategies for patients with metastases. Our study provides evidence in larval zebrafish that metastases of pediatric sarcomas can be eradicated with SBTubA4P with high spatial precision and without adverse systemic effects.

Z-SBTubA4P was shown to disrupt microtubule polymerization in multiple systems in vitro and in vivo, including in zebrafish larvae, with high spatiotemporal precision and without obvious toxicity in its inactive state (Gao et al., 2022). Because microtubule targeting agents are a major group of cancer chemotherapeutics, we examined whether SBTubA4P could also be applied for spatially confined induction of cell death in cancer cells with no adverse systemic effects on normal cells. This goal has proven elusive for photoswitchable reagents based on alternative chemical scaffolds, such as the popular azobenzene photoswitches (An et al., 2019; Borowiak et al., 2015), but we considered that the greater druglikeness and metabolic stability of the styrylbenzothiazole (SBT) scaffold at the core of SBTubA4P ought to improve its chances of long-term in vivo efficacy (Gao et al., 2021, 2022).

First, we reproduced the transient inhibition of microtubule polymerization by photoactivated Z-SBTubA4P in cancer cells using the osteosarcoma cell line OS143B. To be able to observe microtubule dynamics, we transduced OS143B cells to express mNeonGreen-tagged end-binding protein EB3 (OS143B-EB3-mNeon), which fluorescently labels microtubule plus ends. After incubating OS143B-EB3-mNeon cells with 10 µM E-SBTubA4P, we UV irradiated specific regions using a 405 nm laser on a confocal microscope. In accordance with our previous observations in zebrafish cells, EB3-mNeon comets disappeared immediately upon UV E-to-Z-activation of SBTubA4P, but recovered within a few minutes because, being a small molecule (0.4 kDa), the photoactivated inhibitory Z-reagent diffuses readily out of the cell (Fig. S1A, Movie 1). As a single-pulse UV irradiation only transiently affected microtubule dynamics, we wondered whether repeated illumination of OS143B cells would lead to cell death. Indeed, with SBTubA4P, 3 min of repeated UV scanning resulted in cell blebbing, cell contraction, swelling and loss of eGFP fluorescence (Fig. 1A; Movie 2). Propidium iodide (PI) live staining revealed that permeabilization of cell membranes of OS143B cells treated with SBTubA4P occurred as early as 5 min following UV exposure. This effect was restricted to the illuminated area, with the majority of cells inside this region dying within 30-60 min (Fig. 1B; Movie 3). UV-irradiated control cells without SBTubA4P did not show any signs of cell death or PI uptake, indicating that the observed effects were SBTubA4P dependent and not UV induced (Fig. 1B; Movie 4). We confirmed these results with a second cancer cell line, Ewing Sarcoma SK-N-MC (Fig. S1B), demonstrating that spatially restricted photoelimination of cancer cells is feasible with SBTubA4P in sarcoma cells within an hour.

We next addressed whether SBTubA4P could be applied for spatially confined induction of cell death in vivo. Previously, we found that zebrafish larvae could tolerate high concentrations of inactive E-SBTubA4P for up to 24 h with no notable signs of toxicity (Gao et al., 2022). Following exposure to UV light, SBTubA4P-treated larvae displayed morphological abnormalities at concentrations ≥1 µM. Short-term spot irradiation of individual cells in SBTubA4P-treated zebrafish embryos triggered inhibition of microtubule dynamics followed by recovery over the course of a few minutes (Gao et al., 2022), similar to what we observed in OS143B cells (Fig. S1A).

Analogous to the rapid photoinduction of cell death in cancer cell lines in vitro after repeated UV illumination (Fig. 1; Fig. S1B), we went on to explore cell ablation in zebrafish larvae in vivo. To this end, we incubated wild-type AB zebrafish larvae at 3 days post fertilization (dpf) with 50 µM E-SBTubA4P in the dark for 4 h, followed by mounting for microscopy and PI addition to the fish medium. Repeated spatially confined UV illumination was performed via confocal laser scanning at 405 nm for 5 min (Fig. 2A). We observed rapid and locally confined tissue damage in SBTubA4P-treated larvae with PI uptake of dying cells within 30 min after illumination (Fig. 2B; Movie 5). Importantly, control UV-irradiated cells in the absence of SBTubA4P remained unaffected (Fig. 2B).

To determine the spatial restriction of cell death induction, we microinjected the transgenic zebrafish line Tg(UAS:Kaede)^rk8 with KalTA4 mRNA for ubiquitous expression of the photoconvertible fluorescent protein Kaede, which would provide us with a readout for actual UV irradiation, including the adjacent area affected by UV light scattering (Distel et al., 2009; Hatta et al., 2006). We observed rapid onset of cell death and subsequent extrusion of cells restricted to the photoconverted Kaede area in SBTubA4P-treated larvae, whereas untreated larvae retained photoconverted red cells (shown in magenta) and did not show signs of tissue damage by UV (Fig. 2C; Movie 6).

We thereby demonstrate SBTubA4P-photomediated killing of cells to be precise and locally confined in zebrafish larvae in vivo without apparent cytotoxicity outside the targeted region of interest (ROI) and the adjacent margin.

Having confirmed photoefficacy in tumor cells and healthy cells in vivo in zebrafish, we next set out to investigate whether SBTubA4P is suitable for precise photoeradication of disseminated tumor cells in a zebrafish xenograft model in vivo. Injection of OS143B-eGFP cells into mitfa^b692/b692; ednrba^b140/b140 zebrafish larvae at 2 dpf into the perivitelline space resulted in larvae with disseminated tumor cells forming metastases in the caudal hematopoietic tissue (CHT) (Fig. 3A). Injected larvae were selected for the presence of multiple individual tumor masses of similar size in the CHT. One day post injection (dpi), we incubated larvae with 50 µM SBTubA4P for 4 h before mounting on imaging dishes for targeted illumination using a confocal microscope. Small tumor masses in the CHT were specifically targeted with the 405 nm laser for 10 min. Spatially confined in situ illumination of SBTubA4P resulted in loss of eGFP fluorescence in targeted OS143B cells and contraction of the surrounding tissues immediately after UV irradiation (Fig. 3B; Movie 7). We observed none of these effects when disseminated cells were UV irradiated in the absence of SBTubA4P (mock treatment; Fig. 3B).

Because we found that SBTubA4P effectively photoeradicates xenografted tumor cells, we next attempted targeted ablation of single metastases in xenografts at 3 dpf by confined illumination. After UV illumination, larvae were recovered from imaging dishes, washed and kept until 5 dpf. Imaging was performed at 5 dpf to evaluate target-killing and whole-larvae toxicity of SBTubA4P compared to mock treatment (Fig. 3C-E). In SBTubA4P-treated and illuminated individuals, the size of the targeted metastasis was reduced in 55.6% (10/18), and the metastasis was completely eradicated in 44.4% (8/18), of xenografted larvae, whereas all metastases in mock-treated larvae maintained their size or proliferated further (Fig. 3C,D). We also reproduced these results in xenografts using the Ewing sarcoma cell line SK-N-MC (Fig. S2A). For this cell line, metastasis size reduction occurred in 47.4% (9/19), and complete eradication occurred in 52.6% (10/19), of SBTubA4P-treated xenografts (Fig. S2B). For the SK-N-MC control mock treatment group, we also observed regression in some cases; however, this occurred both inside and outside targeted ROIs, indicating that disseminated SK-N-MC cells are not as well maintained as OS143B cells. Treatment outcomes in both cell lines reached statistical significance comparing mock and SBTubA4P treatment (P<0.0001; Fig. 3D; Fig. S2B). Systemic toxicity was assessed by comparing lethality and occurrence of edema in xenografted larvae treated by SBTubA4P and illumination to that in mock-treated individuals. In the illuminated SBTubA4P-treated cohort, 14% of larvae died; 10% also died in the mock-treated cohort. Edemas occurred in 11.6% of the surviving SBTubA4P-treated individuals compared to 25% of the mock-treated individuals (Fig. 3E). Overall, we found no significant differences in toxicity between mock and SBTubA4P treatment.

We additionally performed live staining of larvae with PI during SBTubA4P illumination to visualize the timeframe of cell-death induction following UV irradiation. This revealed first uptake of PI in cells around the targeted locations within an hour, with most cells being eliminated ∼6 h after initial UV treatment (Fig. S2C, Movie 8).

In summary, in situ illumination of SBTubA4P enables spatially precise and highly effective non-invasive elimination of metastases in vivo without overt toxicity in non-targeted tissues.

We initially anticipated that E-to-Z-photoactivated SBTubA4P would lead to cell death via mitotic arrest and apoptosis in phototargeted cells. However, the timescale required for this is much longer than the timescale of cell death we observed in our experiments with repeated UV illumination of SBTubA4P (Bates and Eastman, 2017; Borowiak et al., 2015; Lafanechere, 2022).

We went on to investigate whether treatment with CA4P, a microtubule inhibitor closely resembling Z-SBTubA4P, or treatment with Z-switched SBTubA4P would lead to similarly rapid cell death. We treated OS143B-EB3-mNeon cells with 100 nM CA4P or 10 µM Z-SBTubA4P. We observed that 100 nM CA4P treatment resulted in inhibition of microtubule polymerization, blebbing and round morphology of cells, but rupturing of cell membranes and subsequent swelling did not occur within the observed timeframe (1 h post treatment). Likewise, treatment of OS143B-EB3-mNeon cells with 10 µM pre-switched Z-SBTubA4P stopped microtubule dynamics, but did not result in cell death within 1 h, suggesting a role of illumination with UV light in induction of cell death (Fig. S3A). Indeed, additional UV irradiation of Z-SBTubA4P-treated, but not CA4P-treated, cells induced cell death inside the irradiated region within 30 min (Fig. S3B).

Thus, we investigated which modes of action besides microtubule inhibition might be present when UV irradiating SBTubA4P. Phenotypically, the observed blebbing, swelling, loss of membrane integrity and intact nuclear structure resembled the process of pyroptosis (Yu et al., 2021), for which ROS are major regulators (Li et al., 2023; Su et al., 2022; Wang et al., 2022; Zhang et al., 2023).

To test for potential ROS production upon SBTubA4P illumination (e.g. photochemical singlet oxygen generation), we performed live-staining experiments in OS143B cells with the ROS-sensitive dye CellROX Deep Red Reagent. Indeed, we detected a rapid increase in intracellular ROS following UV irradiation of 10 µM SBTubA4P-treated cells. UV irradiation in the control without SBTubA4P also activated the ROS sensor, but to a lesser degree than that with SBTubA4P (Fig. 4A,B; Fig. S3C). Initially, the CellROX signal appeared to localize to mitochondria. CellROX fluorescence subsequently diffused in SBTubA4P-treated cells as they underwent cell death (Fig. S3D, Movie 9). We next examined whether an antioxidant could rescue induction of cell death. Therefore, we co-treated OS143B cells with 10 µM SBTubA4P and 5 µM N-acetyl-L-cysteine (NAC). NAC-treated OS143B cells displayed lower baseline ROS around their mitochondria and a more homogenously distributed increase in CellROX signal upon illumination, which subsided 45 min afterwards (Fig. 4C; Movie 10). Furthermore, NAC protected SBTubA4P-treated cells from blebbing, swelling and cell death altogether (Fig. 4C). Unexpectedly, we found NAC to cause uptake of PI in cells without phenotypic indications of cell death (Fig. S3E,F).

To confirm induction of the electrophilic stress response in vivo, which is triggered by ROS, through UV-illuminated SBTubA4P, we also made use of Tg(3EpRE:hsp70:mCherry) zebrafish, which visualize the activation of the Keap1–Nrf2 (also known as Nfe212a) pathway (Mourabit et al., 2019). In the evolutionarily conserved Keap1–Nrf2 pathway, Keap1 serves as a ROS sensor and inhibitor of Nrf2. In the presence of ROS, Keap1 no longer targets the transcription factor Nrf2 for proteasomal degradation. Consequently, Nrf2 accumulates in the nucleus and activates an antioxidant transcription program (Baird and Yamamoto, 2020). Upon SBTubA4P activation through UV irradiation, we observed fluorescent reporter expression in Tg(3EpRE:hsp70:mCherry) larvae confined to the target region in the tail. The UV-irradiated control group without SBTubA4P did not show any reporter expression (Fig. 4D).

Next, we examined whether antioxidants could also reduce tissue damage and cell death of cancer cells in xenotransplanted zebrafish. As we observed toxicity of high doses of NAC and NAC combined with SBTubA4P (Fig. S3G), we tested another widely used antioxidant that has been used in zebrafish before (Kang et al., 2019), mitoquinone mesylate (mitoQ), for cell death rescue experiments in vivo. We treated OS143B-GFP-xenotransplanted larvae with 50 µM SBTubA4P, with or without 500 nM mitoQ, prior to UV illumination. This resulted in rescue of SBTubA4P-mediated cell ablation in a third of illuminated larvae, whereas SBTubA4P treatment alone resulted in targeted cell ablation in all larvae (Fig. S3H).

Taken together, these findings suggest that, in our experimental setting, SBTubA4P induces cell death primarily through photogeneration of ROS and the resulting oxidative stress.

Current chemotherapeutics including microtubule-targeting agents are efficient antitumor drugs, but they also act on healthy cells, causing severe adverse effects – including neuropathy, cardiotoxicity and secondary malignancies – especially if applied early on in life, as is the case in pediatric cancer patients. Systemic toxicity of chemotherapeutic agents also exerts dose-limiting toxicity, which, in turn, hinders effective treatment. Here, we investigated whether the photoswitchable microtubule inhibitor SBTubA4P enables spatially targeted treatment of cancer.

We found that SBTubA4P illumination can be applied to rapidly eradicate cancer cells in vitro with high spatial precision by repeated UV irradiation over the course of <10 min. Similarly, repeated local illumination in SBTubA4P-treated zebrafish larvae resulted in rapid cell death induction in vivo, which itself might be of interest for cell ablation and regeneration studies. As a preclinical in vivo model, we established larval zebrafish xenografts for osteosarcoma and Ewing sarcoma, which are ideally suited for optical manipulation, with a direct optical readout of effects on tumor cells and healthy host cells (Sturtzel et al., 2023). Using a metastatic xenograft model in zebrafish, we demonstrate that targeted SBTubA4P illumination for <10 min at UV intensities that are not phototoxic in the absence of the drug is sufficient to eliminate tumor cells in a spatially confined manner without causing systemic toxicity.

Interestingly, we observed cell death in less than an hour after illumination of SBTubA4P. This timeframe contrasts with our expected mechanism of death through mitotic-arrest-triggered apoptosis (Bates and Eastman, 2017; Borowiak et al., 2015; Lafanechere, 2022). Furthermore, the observed morphological changes of cells undergoing cell death did not resemble apoptosis, but pointed towards pyroptosis (Cookson and Brennan, 2001; Yu et al., 2021). The role of microtubule-destabilizing agents in the induction of pyroptosis is currently inconclusive and likely to be context dependent. Studies have suggested that colchicine inhibits pyroptosis in the context of cardiovascular disease (Li et al., 2024; Yang et al., 2020); others have reported that vincristine drives neuropathy by activating pro-pyroptotic signaling (Starobova et al., 2021). CA4P, the closest structural relative to SBTubA4P, has not been explicitly explored for induction of pyroptosis, but is known to cause blebbing, loss of cell junctions, contraction and necrosis in endothelial cells (Tozer et al., 2002). Although we observed inhibition of microtubule polymerization by illuminated SBTubA4P in sarcoma cells, it is likely not the main mechanism of cell death in our study. This is further supported by our finding that pre-switched Z-SBTubA4P inhibits microtubule dynamics, yet only induces cell death after additional UV irradiation (Fig. S3A,B).

In our experimental setting, we found that SBTubA4P illumination increases intracellular ROS and triggers signaling responses to oxidative damage. Abrogation of cell death through NAC in vitro and also mitoQ in one-third of targeted xenografts in vivo implies a role of oxidative stress in the mechanism of SBTubA4P-mediated killing. We therefore hypothesize that ROS production is the primary mechanism by which SBTubA4P induces cell death in the context of this work. The application of light-responsive antitumor compounds that amplify ROS production to trigger pyroptosis in cancer cells has been successfully demonstrated previously, and is seen as a strategy to simultaneously eradicate cancer cells and promote antitumor immunity (Li et al., 2023; Liu et al., 2024; Wang et al., 2022; Yu et al., 2022; Zeng et al., 2023). Furthermore, combinations of photosensitizers and microtubule-targeting agents such as vincristine have been investigated, showing potential for synergistic effects (Ma et al., 1996); however, the underlying mechanisms remain elusive. SBTubA4P is an addition to this toolbox, showing dual action as a microtubule inhibitor and photosensitizer.

A major limitation for potential clinical applications of light-activated drugs is currently the limited ability to bring light into the body. Light-delivery strategies usually consist of approaches for direct delivery using e.g. near-infrared (NIR) light, which is able to penetrate deeper into the tissue than UV. In combination with upconversion nanoparticles (UCNPs), which need to be targeted to the tumor cells, NIR can also be used to activate UV-switchable compounds, as UCNPs convert NIR back to UV (Wang and Liu, 2009). Probably, most feasible is the use of optic waveguides/optic fibers to deliver the light to the target tissue (Shabahang et al., 2018; Wang and Dong, 2020). For translation into the clinics, the development of novel photopharmacological compounds will have to go hand in hand with the development of novel light-delivery strategies (Pearson et al., 2021).

In conclusion, we demonstrate that precise eradication of metastases is possible without systemic effects in zebrafish xenografts with SBTubA4P. Our data indicate that the mode of action of SBTubA4P is mainly through photogeneration of ROS. Zebrafish xenografts are an ideal preclinical in vivo model to test novel light-activatable anticancer compounds, with their optical properties avoiding the light-delivery issues associated with larger-animal models, while allowing for high-quality assays, utilizing intravital imaging, fluorescent dyes and transgenic reporter strains.

The following cell lines were used for this work: GFP-expressing SK-N-MC (SKshctrl; a kind gift from Beat Schäfer, University Children's Hospital, Zurich, Switzerland) and OS143B (kindly provided by Crystal Mackall, Stanford University, Stanford, CA, USA). SK-N-MC and OS143B cells (non-transduced, eGFP and EB3-mNeon) were cultured in RPMI 1640 medium with GlutaMAX^TM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin–streptomycin (10,000 U/ml; Gibco).

The plasmid pCDH-EB3-mNeon was constructed by amplification of the EB3-mNeonGreen sequence from the EB3-mNeonGreen plasmid (Addgene ID: 98881) (Chertkova et al., 2020 preprint), followed by restriction ligation into a pCDH backbone for lentivirus production.

Virus production to transduce OS143B cells with the EB3-mNeon construct was conducted using lenti-X 293T cells. Initially, lenti-X 293T cells were seeded in T25 flasks at a density of 1.5 million cells per flask in 6 ml Dulbecco's modified Eagle medium (DMEM; 31966-021, Gibco) supplemented with FBS and incubated for 18 h. The following day, a mixture of pCDH or lentiCRISPRv.2 plasmids, psPAX2 and pMD2g plasmids at a ratio of 3.3:2.7:1 was prepared in 250 µl DMEM. Prior to addition to the cells, the transfection reagent PureFection™ (System Biosciences, Palo Alto, CA, USA) was combined with the plasmid mixture at a ratio of 2:1 and incubated for 15 min at room temperature. Subsequently, the lenti-X 293T cells were washed, and their medium was replaced by DMEM, followed by addition of the transfection mixture. On the third day, the lenti-X 293T cells were washed, and their medium was changed to DMEM supplemented with 10% fetal calf serum and 1% penicillin–streptomycin. Harvesting of the lentivirus-containing medium occurred on the fourth and fifth days. The medium was centrifuged for 10 min at 500 g at 4°C, and the supernatant containing the virus was filtered through a 0.45 µm syringe filter (Nalgene™, Thermo Fisher Scientific, Waltham, MA, USA). The resulting virus-containing medium was then used to transduce OS143B cells. Finally, the transduced cells were sorted using FACSAriaFusion (BD Biosciences, Franklin Lakes, NJ, USA).

SBTubA4P was synthesized as described previously (Gao et al., 2022) and used at a final concentration of 10 µM for in vitro experiments on cultured cells or 50 µM for in vivo experiments on zebrafish larvae. Pre-switched SBTubA4P refers to SBTubA4P exposed to UV light for 5 min prior to treatment of cells. PI solution (Sigma-Aldrich, St Louis, MO, USA) was used at a final concentration of 5 µg/ml both in vitro and in vivo. CellROX Deep Red Reagent (Invitrogen, Waltham, MA, USA) was used at a final concentration of 5 µM in vitro. NAC (Sigma-Aldrich, USA) was used at a final concentration of 5 µM in vitro and 500 µM to 5 mM in vivo. mitoQ (MedChemExpress, Monmouth Junction, NJ, USA) was used at a final concentration of 500 nM in vivo experiments. CA4P was synthesized by the group of Oliver Thorn-Seshold (Ludwig Maximilian University, Munich, Germany; Gao et al., 2022) and used at a final concentration of 100 nM.

Zebrafish (Danio rerio) were maintained under standard conditions (Kimmel et al., 1995; Westerfield, 2000), according to the guidelines of the local authorities under licenses GZ:565304-2014-6 and GZ:534619-2014-4. For xenotransplantation experiments, transparent zebrafish mutants (mitfa^b692/b692; ednrba^b140/b140) were used, and experiments were performed under license GZ:333989-2020-4. ednrba^b140/b140 is also known as rose and contains a mutated endothelin receptor Ba, which leads to a decrease in melanocytes and lack of iridophores in adult zebrafish (Johnson et al., 1995; Parichy et al., 2000; Rawls et al., 2001). Tg(3EpRE:hsp70:mCherry), used for visualization of oxidative stress response in vivo, was generated by injection of the plasmid Tol2-3EpRE-hsp70-mCherry-polyA-Tol2 (Mourabit et al., 2019), generously provided by Dr Tetsuhiro Kudoh and Dr Aya Takesono (University of Exeter, Exeter, UK), along with transposase mRNA, according to standard procedures (Kawakami, 2007). Founders were identified, selected for low background and robust induction, and stable lines were established. Tg(UAS:Kaede) (Hatta et al., 2006) was used for photoconversion experiments.

Synthesis of KalTA4 mRNA was performed using a mMessage mMachine SP6 transcription kit (Thermo Fisher Scientific) after linearization of respective plasmids with NotI (New England Biolabs, Ipswich, MA, USA).

For microinjection of mRNA, borosilicate glass capillaries with filament (GB100F-10, Science Products GmbH, Hofheim am Taunus, Germany) were pulled with a needle puller (P-97, Sutter Instruments, Novato, CA, USA). Needles were loaded with 3 µl DNA/mRNA injection mix (nuclease-free water, Phenol Red solution, mRNA), mounted onto a micromanipulator (M3301R, World Precision Instruments, Sarasota, FL, USA) and connected to a microinjector (FemtoJet 4i, Eppendorf, Hamburg, Germany). Droplets of ∼1 nl of the injection mix containing 25 pg mRNA were injected into zebrafish embryos at one-cell stage.

mitfa^b692/b692; ednrba^b140/b140 embryos were raised until 2 dpf at 28°C, dechorionated and anesthetized with 1× tricaine (0.16 g/l; Sigma-Aldrich, Hamburg, Germany), adjusted to pH 7 with 1 M Tris-HCl pH 9.5, in E3 medium. For transplantation, anesthetized larvae were aligned on a Petri dish lid coated with a solidified 2% agarose solution as described previously (Grissenberger et al., 2023; Sturtzel et al., 2023). For injection of tumor cells, borosilicate glass capillaries without filament (GB100T-8 P, Science Products GmbH), pulled with a needle puller (P-97, Sutter Instruments), were used. Needles were loaded with ∼5 μl tumor cell suspension, mounted onto a micromanipulator (M3301R, World Precision Instruments, Friedberg, Germany) and connected to a microinjector (FemtoJet 4i, Eppendorf). Cells were injected into the perivitelline space of zebrafish larvae. Xenotransplanted larvae were sorted for tumor cells at the CHT at 2 h post infection (hpi) and subsequently kept at 34°C.

One day after xenotransplantation (3 dpf), zebrafish larvae were anesthetized, incubated with SBTubA4P for 2-4 h, then washed and mounted in 1.2% low-melting agarose (Sigma-Aldrich, Germany) on #1.5 Glass Bottom Dishes (Cellvis, Mountain View, CA, USA). Cells were seeded on imaging µ-slides (ibidi, Fitchburg, WI, USA) and left to settle overnight before illumination experiments. Targeted UV irradiation on zebrafish larvae and cultured cells was performed using the 405 nm UV diode laser of a confocal microscope (SP8 WLL, Leica, Wetzlar, Germany) set to ∼4.5 µW power as measured by a Nova II power meter (Ophir, North Logan, UT, USA). A temperature control system (The Cube 2, Live Imaging Services, Basel, Switzerland) was used to perform zebrafish experiments at 28°C, cell-line experiments at 37°C and xenograft experiments at 34°C. After illumination experiments, zebrafish larvae were removed from agarose, washed and kept for follow-up imaging at indicated timepoints. All imaging was performed using the aforementioned Leica SP8 WLL confocal microscope. Generation of maximum projections, and image and video exports were done using Leica LAS X Software (Version 3.7.1). Video overlays were generated in Fiji (Schindelin et al., 2012).

CellROX fluorescence and GFP fluorescence area were measured in Fiji (Schindelin et al., 2012) for each frame of the confocal microscopy timelapses during UV illumination. Increase in CellROX area was calculated by dividing the area of CellROX by the area of GFP, normalized to the baseline (first frame) of each treatment condition.

Statistical analysis was performed in GraphPad Prism 9. Chi-square test was employed to compare toxicity and treatment efficacy between mock treatment and SBTubA4P illumination in xenograft experiments.

We are grateful to all members of the Distel laboratory for advice and helpful suggestions. We thank Oliver Thorn-Seshold, Kilian Roßmann and Johannes Broichhagen for valuable feedback on experiments and the manuscript; Crystal Mackall for the OS143B cell line; Beat Schäfer for the SK-N-MC cell line; Benjamin Natha for excellent zebrafish care; and Dr Tetsuhiro Kudoh and Dr Aya Takesono for providing the Tol2-3EpRE-hsp70-mCherry-polyA-Tol2 plasmid.