Photocaged Chloroquine Derivatives for the Light-Dependent Inhibition of Autophagy in Cancer Stem Cells

Abstract

Chloroquine (CQ) and hydroxychloroquine (HCQ) inhibit autophagy and have shown promise as adjuvant anticancer agents, particularly for targeting therapy-resistant cancer stem cells (CSCs). However, their clinical utility is limited by systemic toxicity and poor tumor selectivity. Here, we report the design, synthesis, and photochemical evaluation of [7-(diethylamino)­coumarin-4-yl]­methyl (DEACM)-caged CQ and HCQ derivatives as visible-light-activated autophagy inhibitors. Selective caging of the aliphatic amine suppressed biological activity in the dark and enabled rapid release of the parent drugs upon illumination. The lead compound 1C displayed robust light-dependent cytotoxicity across multiple cancer cell lines and, upon photoactivation, recapitulated CQ’s effects on LC3-II accumulation. In CSC-enriched tumorspheres, 1C completely abolished sphere formation only if illuminated. Ex vivo and in vivo studies confirmed that visible light penetrates tumor tissue sufficiently to activate 1C and locally release CQ within the tumor. These findings establish the first proof of concept for light-controlled autophagy inhibition and provide a blueprint for spatiotemporally confined anticancer therapies based on photopharmacological modulation of CSCs.

Introduction

Chloroquine (CQ) and its derivative hydroxychloroquine (HCQ), initially developed and widely used as antimalarials, were later repurposed and approved for the treatment of autoimmune disorders such as rheumatoid arthritis and systemic lupus erythematosus. They have also been explored as antivirals, including for HIV-1, SARS, and MERS. − More recently, numerous in vitro and in vivo studies have investigated the potential of CQ and HCQ for oncology, either as single agents or in combination with cytotoxic regiments. − Several clinical trials are currently ongoing (https://clinicaltrials.gov/), mainly focused on evaluating their potential as adjuvant therapy combined with chemotherapeutic drugs and radiotherapy. −

Compelling evidence indicates that CQ and HCQ can target cancer stem cells (CSCs) – a minor, highly tumorigenic subpopulation that drives therapeutic resistance and relapse. , CSCs exhibit stem-like properties (self-renewal, differentiation), enhanced drug resistance, and long quiescent phases that render them less susceptible to antiproliferative agents. , Once the bulk cancer cells are eliminated through chemotherapy, CSCs can differentiate into bulk cancer cells and regenerate the tumor. CSC enrichment has been documented in residual disease and metastases across cancer types, including head and neck squamous cell carcinoma (HNSCC) and triple-negative breast cancer (TNBC). Therefore, developing effective therapies that specifically target treatment-resistant CSCs may prevent local recurrences and distant metastases, improving overall survival and patient outcomes.

Several in vitro and in vivo studies support the ability of CQ/HCQ to target CSCs, alone or in combination. , As examples, CQ sensitizes liver CSCs to the tumor microenvironment, targets CSCs in TNBC, , potentiates Temozolomide cytotoxicity in glioma cells, and, in combination with the tyrosine kinase inhibitor afatinib and with cisplatin eradicates CSCs of HNSCC. Clinical trials are also evaluating the ability of HCQ to overcome resistance in combination with chemotherapy, , including taxanes, vorinostat, Temozolomide, or bortezomib, and radiation therapy.

While the precise anticancer mechanisms of CQ/HCQ remain incompletely defined, inhibition of autophagy is widely recognized as a principal mode of action. CQ accumulates preferentially in lysosomes, elevating intralysosomal pH, size, and permeability owing to its diprotic weak-base character. Unprotonated CQ diffuses freely across membranes, but once inside acidic organelles such as the lysosome, it becomes protonated and trapped, leading to lysosomal alkalinization and dysfunction that compromise proteolysis, chemotaxis, phagocytosis, and antigen presentation. −

Even though CQ and HCQ show promise as adjunct therapies in oncology, and specifically in targeting the resistant CSC population, their clinical use in prolonged treatment regimens – such as those required for cancer – is limited by toxicity. Clinical studies have reported adverse effects including retinal toxicity, neuromyopathy, and lysosomal storage disorders that can progress to cardiomyopathy. − HCQ is generally better tolerated than CQ and can be dosed higher in humans – daily doses of CQ are safe up to 500 mg, while HCQ is dosed up to 1200 mg/day. For this reason, the majority of clinical trials employ HCQ for combination therapy. Nevertheless, high-dose HCQ can still cause adverse events (fatigue, anorexia, gastrointestinal effects), and in some examples, these doses produced only modest autophagy inhibition in vivo, and dose-limiting toxicity can preclude escalation to pharmacodynamically effective exposures.

In such cases where toxicity limits the use of drugs, photopharmacology emerges as a promising solution. , It is based on the use of light to control the biological effect of drugs in a precise place and time. In oncology, it could allow us to restrict their effect to the tumor area only, avoiding undesired effects in other tissues. Rendering freely diffusible drugs responsive to light relies on two possible designs, giving photoswitches and photocages. Photoswitches contain a photoresponsive functional group, such as an azobenzene or a hemithioindigo, that isomerize under irradiation of a particular wavelength and intensity. The differences in polarity and geometry of the two isomers can result in a change in target affinity and efficacy, providing a reversible control of biological activity. The design is often based on modifying the structure of known drugs with the photoswitching moiety. Instead, for photocages the original drug is modified at a position that is essential for its biological activity with a photocleavable moiety, such as an o-nitrobenzyl or a coumarin.

Numerous CQ/HCQ analogues have been described, which enhance lysosomal accumulation and cytotoxicity through structural tuning. As representative examples, the 4-alkyl chain has been substituted with a cymantrene group and longer aliphatic chains, the primary alcohol of HCQ has been used to introduce larger substituents and conjugates, and the 7-Cl has also been modified. Yet, to our knowledge, photopharmacology has not been applied to spatiotemporally control the effect of autophagy inhibitors. Given that the therapeutic use of CQ/HCQ is limited by a lack of tumor selectivity, we hypothesized that introducing a light-responsive moiety could enhance its use. Here we describe the design, synthesis, and photochemical characterization of coumarin-caged CQ and HCQ derivatives and evaluate their light-dependent activity in breast and HNSCC models, including CSC-enriched tumorspheres.

Results

Design and Synthesis of Photocaged (H)­CQ Derivatives

Converting a known drug onto a light-responsive molecule often relies on introducing a photocaging moiety onto a functional group that is essential for its activity. Given the role of the basic amine(s) on the mechanism of CQ, we reasoned that protecting its basic center(s) with a photocage could abolish its biological activity, which would be recovered upon uncaging under irradiation. We prepared derivatives of both CQ and HCQ to compare photochemical behavior and biological responses, noting HCQ’s improved tolerability yet reports of greater CQ efficacy in some contexts. While we did not expect significant differences in their photochemical properties, we were interested in comparing their biological properties. Initial attempts with o-nitrobenzyl as the photocaging moiety led to low release yields and rates (data not shown); hence we present herein our results with [7-(diethylamino)­coumarin-4-yl]­methyl (DEACM) as the photocaging moiety.

Direct reaction of commercial chloroquine diphosphate with Br-DEACM resulted in low conversion to the protected product (Scheme ). Instead, upon prior neutralization with NaOH, CQ readily reacted with Br-DEACM to give a mixture of three major protected derivatives. CQ reacted at the tertiary aliphatic amine to give quaternary salt 1C, at the 4-aniline to give tertiary amine 2C, and at both sites to give the doubly protected product 3C. These were separated by reverse-phase column chromatography and isolated as their formate salts. Regiochemical assignments for 1C vs 2C derived from Heteronuclear Multiple Bond Correlation (HMBC) experiments, which showed correlations between the methylene bridge of coumarin and the quinoline skeleton for 2C, while for 1C the correlations of the benzylic position of coumarin were with the methylene groups adjacent to the aliphatic quaternary amine (Figures S1 and S2).

1. Synthesis of Caged Derivatives .

^a i. for 5C: NaOH, water, rt, 30 min, quant.; ii. for 5H: NH₄OH, CHCl₃, rt, 1 h, quant.; iii. MsCl, Et₃N, DCM, rt, 2 h, quant.; iv. LiBr, THF, rt, 18 h, quant.; v. MeCN, 60 °C, 18 h.

A similar strategy led to the three photocaged derivatives of HCQ 1H, 2H, 3H.

Photochemical Characterization of Photocaged Derivatives

With the six photocaged inhibitors in hand, we proceeded to evaluate their ability to release CQ/HCQ under irradiation (Figure ). Under our standard HPLC methods, CQ appeared as a broad band, likely due to its basic nature. To enable its quantification, a method for the simultaneous HPLC-MS/MS analysis of CQ, HCQ, and their cage derivatives was developed, evaluating different sample solvents, stationary and mobile phases. Optimal chromatographic performance was achieved using a C18 stationary phase, 50 mM ammonium formate at pH3 as the mobile phase, and samples were prepared in a water–acetonitrile mixture containing at least 90% water. The use of the buffered mobile phase and using predominantly water to dissolve samples improved peak symmetry and sharpness by stabilizing the ionization state of CQ and minimizing secondary interactions with the stationary phase.

1.

Photochemical characterization of compounds 1C–3C. (A–C) Absorbance spectra of compounds 1C – 3C, at 100 μM in PBS. (D–F) HPLC quantification of the photocaged derivative and CQ during illumination at 405 (2C and 3C, 19 mW/cm²) or 420 nm (1C, 13 mW/cm²) for the indicated times, with an initial solution at 50 μM in PBS. For the uncaging of 3C (F), 2C is generated as an intermediate product after uncaging of the first coumarin moiety.

First, we confirmed that the six compounds were stable in the dark under the conditions of the biological studies. After 3 days in complete cell culture medium at 37 °C, there were no significant changes in the concentration of the photocaged derivatives, confirming that the active (H)­CQ will not be released under lack of illumination (Figure S3).

Upon illumination, the coumarin at the aliphatic amine cleaved faster than at the aniline: 1C released CQ with t ₉₀ ≈ 3 min, whereas 2C required 20 min under matched conditions (Table , Figure ). The yield was also higher for 1C. Consistently, double protected analogue 3C released one coumarin rapidly to give 2C, which then uncaged more slowly, at an overall lower yield of 40%. The yield of CQ formation for 1C (68%) aligns with reported yields for coumarin PPGs. , HCQ derivatives behaved similarly: 1H released HCQ with t ₉₀ ≈ 2 min and a maximal concentration of 28 μM (57% yield) (Table , Figure S4).

1. Photophysical and Photochemical Properties for the Uncaging of 1C–3C and 1H–3H .

compound	λ max [nm]	ε 405 [×10 3  M –1  cm –1 ]	ε 420 [×10 3  M –1  cm –1 ]	t 90 [min]	yield (%)
1C	406	9.1	7.9	3	68
2C	360, 384	6.6	2.3	21	56
3C	386	12.9	7.4	20	40
1H	404	67	5.6	2	57
2H	358, 384	5.9	1.8	35	51
3H	388	17.0	10.2	10	24

^aWavelength at which absorption is maximum.

^bTime required to reach 90% of the final concentration of (H)­CQ.

^cChemical yield of released (H)­CQ measured by HPLC, after illumination at 405 nm (2C and 3C, 19 mW/cm²) or 420 nm (1C, 13 mW/cm²). In all cases, 100 μM in DMSO.

We also quantified the release of coumarin alcohol 6 during the photolysis of 1H and found that 6 only reached a concentration of 5 μM that remained stable after 2 min (Figure S4D), suggesting the formation of other photolytic products of the coumarin skeleton that might degrade, as we did not detect other discernible peaks in the HPLC spectra (Figure C).

2.

Uncaging profile of 1C. (A) Uncaging reaction. (B) Changes of the UV–vis absorption of 1C recorded in 20 s intervals during 420 nm (13 mW/cm²) irradiation at 100 μM in PBS; dotted lines show the UV–vis absorption of CQ and coumarin 6 at 100 μM in PBS. (C) Representative HPLC traces of CQ, 1C, and 1C after 420 nm (13 mW/cm²) illumination (1 and 2 min) at 100 μM in PBS. (D) Amount of CQ formed after illumination of 1C (50 μM in PBS) for 2 min at 420 nm of different intensities, quantified by HPLC. (E) Amount of CQ formed after illumination of 1C (50 μM in PBS) for 2 min at 365, 380, 405, 420, 455, 470, and 500 nm (11–13 mW/cm² except for 365 at 7 mW/cm²), quantified by HPLC. (F) ¹H NMR of 1C under dark and illumination (420 nm, 15 and 25 min), at 2 mg mL^–1 in DMSO-d ₆. Green regions indicate peaks from the CQ quinoline scaffold and orange from the coumarin scaffold.

We selected the fastest photocaged derivative 1C (Figure A) for a deeper photochemical analysis. Following the photolysis process by UV–vis spectroscopy confirmed that the uncaging reaction is completed after 3 min at 420 nm with loss of the coumarin band and retention of characteristic CQ band (Figure B); a similar profile was observed for 1H (Figure S5A). This is consistent with the photolysis studies by HPLC (Figure C), which had shown CQ as the major released product. We also followed the uncaging process by ¹H NMR (Figure F), which also showed disappearance of the coumarin peaks with minimal changes on CQ quinoline signals. In this experiment, the photolysis required a longer time to complete as the amount and concentration of photocaged inhibitor was larger (100 μM for HPLC studies, 4 mM for NMR studies).

Illumination of 1C for 2 min at 420 nm of different intensities (2–13 mW/cm²) gave an intensity-dependent release of CQ (Figure D), consistent with a photon flux-dependent reaction rate. Testing the photolysis at different wavelengths ranging from UV to green light (Figure E) indicated that, while 405 and 420 nm led to the fastest release consistent with its maximum absorbance at 406 nm (Table ), similar conversions are obtained with up to 470 nm. Although at 500 nm the photolysis reaction was significantly slower, the release was still considerable with only 2 min of illumination giving a 10% release.

Biological Evaluation in Vitro

Having confirmed that the prepared photocaged derivatives are able to release (H)­CQ under illumination, we tested their ability to induce a light-dependent decrease in viability of cancer cell lines. We first verified dark inactivity.

As breast cancer is one of the malignancies where autophagy inhibition by (H)­CQ has been validated, we started our studies with the common breast cancer cell line MCF7. Cages bearing coumarin at the aromatic amine (mono- (2C, 2H) or di- (3C, 3H) substituted) retained appreciable activity at the highest concentration, approaching the IC₅₀ of free (H)­CQ (Figures A and S6), and were therefore unsuitable as silent prodrugs. In contrast, the quaternized aliphatic-amine cages were minimally active in the dark: 1C was inactive up to 200 μM and 1H showed only weak effects at the top dose.

3.

Effects of photocaged derivatives in cancer cell lines on adherent culture. (A) Dose–response curves of CQ and of the CQ photocaged derivatives 1C–3C on the viability of MCF7 cells (MTS assay) under dark conditions for 72 h (IC₅₀ (CQ) = 34.1 μM; IC₅₀ (2C) = 45.6 μM; IC₅₀ (3C) = 72.6 μM). (B) Workflow for viability studies with adherent cells under dark vs illumination conditions. (C) Dose–response curves on the viability of MCF7 cells (MTS assay) of 1C under dark vs illumination (420 nm, 8 mW/cm², 40 s) conditions and of CQ for 72 h. (D) Dose–response curves on the viability of MCF7 cells (MTS assay) of 1H under dark vs illumination (420 nm, 8 mW/cm², 30 s) conditions and of HCQ for 72 h. (E) Viability of a range of cell lines upon treatment with 1C under dark vs illumination (420 nm, 8 mW/cm², 40 s) conditions and of CQ at 200 μM. (F) Immunoblotting analysis on the effect of 1C under dark vs illumination (420 nm, 8 mW/cm², 1 min) conditions and of CQ on the relative expression levels of LC3-II and LC3-I on MCF7 cells; cells treated for 2 h. Values in (A), (C), (D), (E), and (F) are represented as mean ± SD.

Other important controls were confirming that the cells are not affected by illumination at 420 nm for 40 s to 2 min (Figure S7) and that the activity of CQ is not affected by illumination (Figure S8).

Short illumination immediately after dosing (Figure B) restored activity: 1C and 1H displayed dose-dependent cytotoxicity matching their parent drugs over 72 h (Figure C,D). This light-dependent effect was reproduced across a range of cell lines covering breast cancer (MCF7, MDA-MB-468, MDA-MB-231), HNSCC (JHU029, HTB43, CCL138), and colon cancer (HT29) (Figure E, Table ). In general, we observed a larger window between dark and light conditions for 1C than 1H, as 1H showed some activity at the highest concentration(s).

2. Effect of Selected Compounds on the Viability of a Range of Cell Lines under Dark or Illuminated Conditions, Measured by an MTS Assay .

			1C		1H
cell line	CQ	HCQ	dark	light	dark	light
MCF7	44.6	46.5	>200	42.6	≈200	53.5
MDA-MB-231	35.2	39.7	>200	51.5	≈200	41.5
MDA-MB-468	57.4	55.4	na	69.9	≈200	46.6
JHU029	≈200	91.2	na	≈200	≈200	85.8
HTB43	102.1	98.2	na	107.4	na	97.7

^aValues are shown as IC₅₀ values in μM.

^bCells kept in the dark only.

^cCells illuminated at 420 nm (8 mW/cm²) for 30–40 s before the 72-h incubation period.

^dEstimation, full dose–response not obtained.

^ena = not active, indicating no decrease in viability observed up to 200 μM.

To gain further evidence on the effect of 1C being triggered through the release of CQ, we assessed by Western blotting for LC3-II protein, a common marker of autophagy. We quantified the effect as the LC3-II/LC3-I ratio, which is widely used to monitor autophagosome formation. We observed that 1C under dark conditions had similar levels of LC3-II as the nontreated cells, while 1C under illumination triggered an increase on LC3-II accumulation similar to CQ (Figures F and S9).

Coumarin alcohol 6, which we had determined to be the major byproduct of the uncaging reaction and was stable to the illumination conditions employed (Figure S10), showed no activity by itself in cellular viability (Figure S11). Nonetheless, we did observe that at longer illumination times and higher intensities, 6 showed some activity by itself and the photocaged derivatives gave an increase in potency, suggesting that under these conditions they could be acting through an alternative mechanism alongside autophagy inhibition.

One of the promises of autophagy inhibitors such as CQ is their effect on CSCs, hence, we also assessed the light-dependent effect of photocaged inhibitor 1C on tumorspheres. Sphere cultivation is widely used to enrich CSCs from bulk cancer cells and is widely accepted as a functional assay of self-renewal property of CSCs. − It is based on plating a single cell suspension at a proper cell density on ultralow attachment surface with the serum-free culture medium in supplementation with several defined growth factors. We grew MCF7, JHU029, and HTB43 cells under nonadherent conditions to form 3D spheres and confirmed these to be enriched in stemness factors (Figure A,B).

4.

Effect of 1C on CSC spheres. (A) mRNA levels of stemness genes in JHU029 cells grown in either adherent or sphere (first generation) culture, determined by PCR. (B) Brightfield images of JHU029 cells grown in either adherent or sphere (first generation) culture. (C) Brightfield images of HTB43 cells grown in sphere culture, either nontreated or treated with 1C at 100 μM under dark vs illumination conditions (420 nm, 8 mW/cm², 40 s) and with CQ at 100 μM for 10 days. Average number of spheres counted on individual wells of 96-well plates for JHU029 (D) and HTB43 (E) cells grown in sphere culture, treated with 1C at 100 μM under dark vs illumination (420 nm, 8 mW/cm², 40 s) conditions and with CQ for 10 days. Values in A), D), and E) are represented as mean ± SD.

To evaluate compound effects on CSCs, cells were treated immediately after seeding, and both sphere formation and viability were assessed after 7–10 days, when untreated spheres reached an average diameter of ≥250 μm for HTB43 and ≥300 μm for JHU029. Spheres were markedly more sensitive to CQ than adherent cultures (e.g., JHU029: IC₅0 adherent ≈200 μM vs spheres ≈12 μM), consistent with autophagy’s role in CSC survival. As we did not observe major changes on the viability of spheres of first or third generation treated with HCQ (Figure S12), we continued the subsequent studies with first generation spheres.

Cells treated with 1C and kept in the dark successfully formed spheres. In contrast, a brief 40-s illumination prior to the 10-day spheroid culture completely abolished their ability to form spheres in both JHU029 and HTB43 cells (Figures C–E and S13). Illuminated 1C exhibited lower IC₅₀ values compared to those obtained in viability assays in adherent cells, further supporting its impact on autophagy and CSCs (Table ). We confirmed that illumination alone did not affect the growth of nontreated spheres (Figure S14).

3. Effect of CQ and 1C on CSC Spheres of JHU029 and HTB43 Cells under Dark or Illuminated Conditions.

cell line	CQ (viability,  IC 50  [μM])	CQ (sphere count)	1C (viability, IC 50  [μM])		1C (sphere count)
			dark	light	dark	light
JHU029	12.2	5.5	110.6	22.5	134.0	13.2
HTB43	36.0	50.0	na	59.8	na	48.0

^aCells kept in the dark only.

^bMeasured by an MTS assay.

^cCells illuminated at 420 nm (8 mW/cm²) for 30–40 s before the 72-h incubation period.

^dna = not active, indicating no decrease in viability observed up to 200 μM.

Photorelease in Vivo

Having confirmed the fast release of photocaged inhibitor 1C under illumination and its light-dependent effect in cellular studies in vitro, we assessed its ability to release CQ in vivo; we chose MDA-MB-231 tumors as this cell line gave a big window between dark and illuminated conditions.

We first quantified the amount of light transmitted through ex vivo mouse breast tumor samples of varying widths. Three wavelengths were selected – 405, 430, and 470 nm (Figure S15) – corresponding to the maximum uncaging efficiency of 1C (Figure E). Illumination was performed using mounted LEDs from Thorlabs equipped with collimators to minimize scattering, and transmitted irradiance was measured with a photodiode power sensor (Figure S16). As expected, thicker tumors generally allowed less light to pass through (Figure S17A,B). However, this relationship was more heterogeneous compared to more homogeneous tissues such as chicken breast (Figure S17C,D), where attenuation correlated perfectly (r = 0.976) with thickness following a logarithmic relationship. Across all samples, transmitted irradiance increased with wavelength: 470 nm > 430 nm > 405 nm (Figure A), consistent with reduced absorption at longer wavelengths. At full LED power (130 mW/cm² at 470 nm), up to 20 mW/cm² crossed the tumors – approximately 30% of the incident light. These results support that light can reach these tumors at a sufficient photon quantity to activate photocaged inhibitor 1C, which we proceeded to confirm in a real animal setting.

5.

Photolysis studies ex vivo and in vivo. (A) Percentage irradiance detected after illumination of tumor samples of MDA-MB-231 cells ex vivo at the indicated wavelengths. (B) Workflow followed for the in vivo study, aimed at quantifying the amount of CQ released in an orthotopic breast cancer model with MDA-MB-231 cells. (C) Quantification of 1C detected by HPLC before and after illumination of the tumor at 470 nm for 10 min, using the setup shown in (B). (D) Quantification of CQ detected by HPLC before and after illumination of the tumor at 470 nm for 10 min, using the setup shown in (B). Values in (A), (C), and (D) are represented as mean ± SD.

As 470 nm resulted in the largest intratumor irradiance among the three tested wavelengths, and we confirmed that this light can be used to induce the same light-dependent effect of 1C (Figure S18) as the previously used 420 nm, we performed the in vivo study with this wavelength (Figure B). We chose an orthotopic breast cancer model of MDA-MB-231 cells; cells were inoculated at the intramammary fat pad, and the study was initiated when tumors reached 80–120 mm³. We injected 1C intratumorally at 10 mg/kg, and after illumination for 10 min, tumors were excised, homogenized, and analyzed by HPLC. We found that the 10 min illumination period was sufficient to consume 50% of the photocaged inhibitor (Figure C), with a consequent release of CQ (Figure D). Instead, nonilluminated tumors showed no significant changes in the amount of 1C nor CQ.

Discussion

CQ and HCQ have emerged as versatile agents with therapeutic potential beyond their original antimalarial applications. Their ability to modulate lysosomal function and inhibit autophagy has attracted significant interest in oncology, particularly for overcoming therapy resistance. − Accumulating evidence indicates that CQ and HCQ can selectively target CSCs – a subpopulation responsible for tumor relapse and metastasis – thereby enhancing the efficacy of conventional chemotherapeutic and radiotherapeutic regimens. , Through lysosomal alkalinization and autophagy inhibition, these agents can disrupt key survival pathways in CSCs and bulk tumor cells. ,, Together, these findings highlight the promise of CQ and HCQ as adjuvant anticancer agents, though their precise mechanisms and therapeutic window remain subjects of active investigation.

Despite the therapeutic promise of CQ and HCQ, their long-term use in oncology remains constrained by dose-dependent toxicity. Retinopathy, neuromyopathy, and cardiomyopathy have been documented with chronic administration, underscoring the narrow therapeutic window of these agents. − Although HCQ offers improved tolerability compared to CQ and is therefore favored in most clinical settings, achieving pharmacologically effective concentrations in tumors without systemic adverse effects remains challenging. These limitations highlight the need for strategies that enhance spatial and temporal drug selectivity. Photopharmacology provides an elegant solution by enabling light-controlled activation of bioactive molecules with high precision. − Through the incorporation of photoswitchable or photocleavable motifs, drug activity can be confined to the tumor site, minimizing systemic exposure. Building upon this concept and given the absence of photopharmacological approaches to modulate autophagy inhibition, we explored the design of coumarin-caged CQ and HCQ derivatives. This approach aimed to achieve localized, light-dependent activation of CQ/HCQ to overcome toxicity barriers while maintaining their capacity to target CSCs and disrupt autophagy in cancer models.

The design of photocaged CQ and HCQ analogues aimed to achieve complete suppression of activity in the dark and efficient recovery of the parent drug under visible light. Given that CQ’s pharmacological effect relies on its basic amine centers, caging these functionalities with a photolabile group was expected to block lysosomal accumulation and autophagy inhibition until photoactivation. Incorporation of the DEACM group provided a visible-light-sensitive photocaged inhibitor compatible with biological applications, while also improving absorption at wavelengths of lower phototoxicity. The synthetic approach enabled modification at both the aliphatic and aromatic amines, yielding mono- and bis-protected derivatives for both CQ and HCQ.

Photochemical studies confirmed that all DEACM-protected derivatives were stable under dark conditions yet efficiently released their parent drugs upon illumination with visible light. Among them, the aliphatic amine-protected species (1C and 1H) showed the fastest and most efficient uncaging, with t ₉₀ of ∼ 3 min and 68% (H)­CQ recovery under 420 nm light – consistent with yields reported for other coumarin-based photocaged derivatives. , The slower photolysis of the aromatic and bis-protected derivatives likely reflects their worse leaving group ability. Notably, 1C retained meaningful uncaging efficiency beyond its absorption maximum: substantial release occurred at 455–470 nm with short exposures, and even at 500 nm we detected ∼10% CQ release in 2 min. This breadth reflects both DEACM’s tailing absorption and the rapid photolysis of the quaternary adduct, enabling very short illuminations and the use of suboptimal (but more penetrating) absorbing wavelengths when necessary.

In adherent cancer cell lines, 2C, 3C, 2H, and 3H were not suitable as light-activated prodrugs as they showed activity in the dark similar to the parent molecules. Instead, caging the aliphatic amine in 1C and 1H effectively abolished CQ/HCQ cytotoxicity under dark conditions, confirming that lysosomal accumulation and autophagy inhibition were blocked prior to illumination. CQ-derived 1C showed a larger window than HCQ-derived 1H. Upon brief irradiation, 1C regained potent, dose-dependent cytotoxicity across multiple cancer cell lines covering HNSCC, breast, and colon cancer, recapitulating the effects of their parent drugs. This light-dependent restoration of activity, absent in control experiments with illumination alone, directly links the observed cytotoxicity to photorelease of the active compound. Furthermore, LC3-II accumulation following illumination of 1C mirrored that induced by CQ, supporting that autophagy inhibition remains the principal mechanism of action postuncaging.

Importantly, the light-controlled cytotoxicity of 1C extended to CSCs. We used tumorspheres as a model of CSCs and confirmed these to be enriched in stemness markers. Under dark conditions, 1C was inert and allowed normal sphere formation, whereas illumination fully prevented sphere growth and viability, matching the effects of free CQ. Moreover, the selected cell lines appeared to exhibit greater sensitivity to both CQ and 1C when cultured as spheres compared to adherent conditions, potentially underscoring an increased susceptibility of CSCs to autophagy inhibition. Collectively, these findings establish 1C as a potent photosensitive prodrug, enabling precise spatial and temporal control over autophagy inhibition and CSC fate in vitro.

Ex vivo and in vivo optical experiments demonstrated that visible light penetrates tumor tissue sufficiently to trigger uncaging of 1C. First, we confirmed that UV and blue light (405, 430, and 470 nm) can cross tumor samples of up to 9 mm in thickness ex vivo, with blue light (470 nm) exhibiting the highest irradiance. This light effectively induced CQ release in living mice bearing orthotopic breast tumors. A single 10 min illumination following intratumoral administration of 1C led to approximately 50% consumption of the photocaged inhibitor and measurable generation of CQ within the tumor, while nonilluminated controls remained unchanged. These results confirm that photochemical activation can occur efficiently in a physiological context under light conditions compatible with biological safety. Previous studies had modeled and studied the optical properties of tumors and malignant tissue. − Our work complements these by confirming that, even with scattering and absorption caused by the tumor tissue, clinically relevant light doses can achieve localized activation, supporting the development of photoresponsive therapeutics.

Collectively, these data demonstrate proof of concept for spatially confined activation of an autophagy inhibitor in vitro and in vivo. The ability to trigger CQ release within tumor tissue using minimally invasive light exposure highlights the translational potential of this approach. While external illumination will find utility for easily accessible cancers such as breast cancer, HNSCC, and skin cancer, ongoing work with implantable and wireless LEDs and fibers − will expand the use to other tissues. Future studies will focus on using other more selective autophagy inhibitors, using red-shifted photocaging groups, and establishing a proof of concept in efficacy studies in vivo.

Conclusions

This study demonstrates a successful strategy to impart light-responsiveness to CQ/HCQ-based autophagy inhibitors through photocaging of their basic amine centers. The use of the DEACM group allowed efficient and reversible control of CQ/HCQ activity under UV and blue light, achieving full suppression of cytotoxicity in the dark and rapid restoration upon illumination. Among the prepared derivatives, the aliphatic amine–caged analogue 1C emerged as the most effective photoresponsive compound, displaying fast uncaging kinetics and robust light-dependent biological activity.

Photochemical and biological analyses confirmed that 1C retains the mechanistic hallmark of CQ – autophagy inhibition – while enabling precise spatial and temporal activation. Its ability to eradicate CSC-enriched tumorspheres under illumination underscores the therapeutic potential of light-controlled autophagy modulation to overcome resistance and recurrence. Moreover, in vivo activation of 1C within tumor tissue validates the feasibility of visible-light-triggered drug release under physiologically relevant conditions.

Overall, these results establish the conceptual and practical foundation for photopharmacological control of autophagy inhibitors. The design principles demonstrated here could be extended to other weak-base chemotypes or photoswitchable scaffolds, paving the way toward safer and more selective autophagy-targeted therapies in oncology. Further development toward clinical use could represent a more selective and safer therapy for cancer treatment.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.5c00962.

• Chemistry methods, photochemistry methods, biology methods, supplementary figures, HPLC chromatograms of final compounds, ¹H and ¹³C NMR spectra for final compounds (PDF)

The authors declare no competing financial interest.

The version of this paper that was published ASAP March 24, 2026, contained an error in the surname of author Matilde Esther Lleonart. This error was corrected and the paper reposted September 25, 2026.