Recent advances in reactive oxygen species (ROS)-responsive drug delivery systems for photodynamic therapy of cancer Authors: Danrong Hu, Yicong Li, Ran Li, Meng Wang, Kai Zhou, Chengqi He, Quan Wei, Zhiyong Qian Published: 2024 ABSTRACT AbstractReactive oxygen species (ROS)-responsive drug delivery systems (DDSs) have garnered significant attention in cancer research because of their potential for precise spatiotemporal drug release tailored to high ROS levels within tumors. Despite the challenges posed by ROS distribution heterogeneity and endogenous supply constraints, this review highlights the strategic alliance of ROS-responsive DDSs with photodynamic therapy (PDT), enabling selective drug delivery and leveraging PDT-induced ROS for enhanced therapeutic efficacy. This review delves into the biological importance of ROS in cancer progression and treatment. We elucidate in detail the operational mechanisms of ROS-responsive linkers, including thioether, thioketal, selenide, diselencide, telluride and aryl boronic acids/esters, as well as the latest developments in ROS-responsive nanomedicines that integrate with PDT strategies. These insights are intended to inspire the design of innovative ROS-responsive nanocarriers for enhanced cancer PDT.Key words: Photodynamic therapy, ROS-responsive, Drug delivery systems, Cancer therapy Introduction Cancer remains one of the leading threats to human health, with conventional treatments such as surgery, chemotherapy and radiotherapy having notable limitations1. While Surgery is effective for localized lesions, it often fails to eliminate all tumor lesions, leading to recurrence2. While effective against cancer cells, small-molecule chemotherapeutic agents indiscriminately exert cytotoxic effects on malignant and healthy cells, leading to undesirable side effects3. The effectiveness of ionizing radiation is compromised by the hypoxic conditions within solid tumor microenvironments, posing the additional risk of harming adjacent healthy tissues4. Despite advancements in endocrine therapy, targeted therapy, and immunotherapy, improvements in cancer survival rates have been modest5, 6, 7. In contrast, photodynamic therapy (PDT) is a promising alternative, that uses light-activated photosensitizers to selectively produce reactive oxygen species (ROS) within cancer cells, leading to oxidative stress and cellular damage8, 9, 10. The benefits of PDT include minimal invasiveness, high effectiveness and localized therapy with low systemic toxicity11,12. Nevertheless, PDT efficacy largely depends on several factors, such as photosensitizer solubility13,14, light penetration depth15, 16, 17, conversion efficiency18,19, targeting capability20,21, and the oxygen levels in the tumor microenvironment22, 23, 24, 25. Drug delivery systems (DDSs) are innovatively designed to achieve targeted and controlled release of therapeutics, enhancing treatment efficacy and minimizing side effects through their responsiveness to specific physiological triggers26, 27, 28. Recent advances in materials science and nanotechnology have synergistically fuelled the development of both localized and systemic drug delivery strategies29, 30, 31, 32, 33. Localized DDSs, such as hydrogels, microneedles and implantable devices, are valued for their precise targeting ability and ability to serve as local drug depots, ensuring the sustained release of high drug concentrations at tumor sites, which optimizes therapeutic outcomes while minimizing systemic toxicity34,35. Nonetheless, the application of localized DDSs is limited when tumor sites are inaccessible. Nanotechnology plays a pivotal role in systemic drug delivery, offering benefits such as biocompatibility, targeted delivery, and tailored physicochemical properties36, 37, 38, 39. Nanoparticles, encompassing a wide range of forms, including micelles, liposomes, dendrimers, and metal-based particles, serve as essential building blocks for developing efficient DDSs40, 41, 42, 43, 44. In particular, the development of stimuli-responsive DDSs with various predesigned functions, such as on-demand release of the payloads in a spatially and temporally controllable manner, is crucial for promoting the clinical translation of nanomedicine45, 46, 47. These typical stimuli can include endogenous stimuli, such as pH value48,49, enzymes50, 51, 52, ROS53, 54, 55, 56, glutathione57, 58, 59, 60, 61, and hypoxia62, 63, 64, or exogenous stimuli, such as light65, 66, 67, 68, ultrasound69, 70, 71, temperature72,73, electric field74,75, magnetic force76, and X-ray77, 78, 79. ROS are pivotal stimuli in ROS-based therapies such as PDT80,81, chemodynamic therapy (CDT)82,83, and sonodynamic therapy84,85, because of their dual ability to trigger responses and exert direct antitumor effects. Therefore, interest in developing ROS-responsive DDSs for diseases characterized by ROS dysregulation is increasing86, 87, 88. ROS, a double-edged sword within cellular processes, are integral to cell signaling under normal conditions but are also characteristically elevated in cancer cells, which exhibit ∼100 μmol/L levels—10 to 100 times higher than those in normal cells—are correlated with tumor progression89,90. This altered redox state within the tumor microenvironment has spurred the creation of ROS-responsive DDSs, leveraging the biochemical disparity between healthy and cancerous cells for targeted cancer therapy. By incorporating ROS-responsive moieties and linkers, these systems can activate and release drugs on demand at the tumor site91,92. Despite these advancements, current ROS-responsive DDSs face challenges due to their heterogeneous ROS distribution within tumors and inadequate endogenous ROS replenishment93,94. Consequently, the development of ROS-responsive PDT nanosystems that supply exogenous ROS is particularly appealing owing to their self-amplifying feedback mechanism and synergy in drug release95,96. These systems utilize optimized nanocarriers to deliver photosensitizers specifically to tumor sites, avoiding premature release into the systemic circulation or healthy tissues. Elevated ROS at the tumor site trigger drug release and photosensitizer activation by light to produce cytotoxic ROS. Concurrently, the exogenous ROS generated during PDT increase the release of photosensitizers, thereby increasing the overall effectiveness of PDT. Several reviews have documented the development and application of various stimulus-responsive DDSs in cancer therapy97, 98, 99. However, a systematic discussion on the integration of ROS-responsive DDSs with PDT is lacking. This review highlights advancements in ROS-responsive PDT nanosystems, emphasizing their potential for tumor-specific drug release and enhanced therapeutic efficacy (Fig. 1). By employing a self-amplifying feedback mechanism, these systems overcome the limitations of heterogeneous distribution and an insufficient supply of endogenous ROS. Specifically, this review begins with an overview of the intracellular redox system, including organelles and molecules involved in ROS production and ROS scavenging. The physiological and pathological roles of ROS, along with potential therapeutic opportunities in cancer treatment are discussed. Next, we examine the current status of ROS-responsive DDSs in PDT, covering the design and synthesis of ROS-responsive materials, controlled drug release, and targeting specificity, as well as the mechanisms of action of various ROS-responsive linkers, such as thioether, thioketal, selenide, diselenide, telluride, and aryl boronic acid/ester. This review aims to provide valuable insights and novel strategies that could significantly advance the clinical application of PDT and other ROS-based therapeutic regimens. Reactive oxygen species (ROS)-responsive drug delivery systems in photodynamic therapy (PDT). ROS: Biology and pathophysiological role in cancer Typically, ROS are a group of highly reactive oxygen-based chemical species that are physiologically formed in cells as byproducts of aerobic metabolism. They mainly include free radicals, such as superoxide anions (O2·−), hydroxyl radicals (·OH), alkoxyl radicals (·OR) and peroxyl radicals (·OOR), as well as nonradical molecules, such as hydrogen peroxide (H2O2) and hypochlorous acid (HOCl)100. One-electron reduction of an oxygen molecule produces O2·−, and O2·− can be further transformed to H2O2 through spontaneous dismutation or via a reaction catalyzed by superoxide dismutase (SOD)101. Compared with short-lived O2·− with a lifetime of milliseconds and limited diffusion ability, H2O2 is relatively stable in vivo102. Owing to its lipid-solubility, H2O2 can diffuse freely through membranes, acting as a second messenger for intercellular crosstalk and regulating multiple physiological signaling pathways by selectively oxidizing target proteins103. H2O2 is decomposed into water and O2 by the cytoplasmic antioxidant enzymes104. Alternatively, H2O2 with unstable oxygen bonds can oxidize Fe2+ to form Fe3+ and the most reactive ·OH via the Fenton reaction105. In addition, ·OH can be produced from electron exchange between O2·− and H2O2 via the Haber–Weiss reaction106. The ·OH is extremely destructive to almost all biological molecules. ROS were originally perceived as toxic byproducts of aerobic metabolism because they can damage cellular biomacromolecules, which eventually leads to cell destruction and is implicated in various diseases, including cardiovascular dysfunctions, neurodegenerative disorders, diabetes mellitus and cancers107, 108, 109, 110. Nevertheless, with increasing knowledge of cellular redox homeostasis, a certain level of ROS are indeed involved in numerous signaling pathways in both normal and cancer cells111,112. This section discusses the intracellular redox system, including ROS-generating and ROS-scavenging organelles and substances, the physiological and pathological roles of ROS, and ROS-related cancer therapeutic opportunities. The balance of intracellular redox homeostasis ROS-generating systems: Mitochondria ROS can be produced via multiple mechanisms, among which mitochondrial metabolism, the family of transmembrane nicotinamide dinucleotide phosphate (NADPH) oxidases (NOXs) and peroxisomes are the main sources of endogenous ROS that are implicated in cancer (Fig. 2)113. In most mammalian cell types, including cancer cells, mitochondria are the most significant contributors to intracellular oxidation114. Oxidative phosphorylation within mitochondria generates cellular energy through adenosine triphosphate (ATP). During this process, electrons flow across the mitochondrial electron transport chain (ETC), which is composed of a series of electron transporters, including four multisubunit complexes (complexes I‒IV) and two shuttles―coenzyme Q (CoQ) and cytochrome C115. According to studies on isolated mitochondria, complex I and complex III, where are located in the intermembrane space are believed to be the major sites mediating mitochondrial ROS or O2·− generation116. Cellular generation of reactive oxygen species (ROS). Key sources include the mitochondria, nicotinamide dinucleotide phosphate oxidases (NOXs), and other cellular organelles and enzymatic systems. Mitochondrial complexes I and Ⅲ generate superoxide (O2·−), which is metabolized to hydrogen peroxide (H2O2) by superoxide dismutase 2 (SOD2) in the matrix or SOD1 in the intermembrane space or outer mitochondrial membrane. Alternatively, H2O2 can interact with the Fe–S clusters, abundant in the electron transport chain (ETC), to form hydroxyl radicals (·OH) via the Fenton reaction. Membrane ROS are generated by the NOXs/dual oxidase (DUOX) system. Generally, NOX1, NOX2, NOX3 and NOX5 mainly produce O2·−, while NOX4, DUOX1, and DUOX2 mainly generate H2O2 directly. Another crucial source of H2O2 is the endoplasmic reticulum (ER). Peroxisomes can generate ROS, principally as H2O2, often cooperating with mitochondria. XDH, xanthine dehydrogenase; ERO1, ER oxidoreductase 1; PDI, protein disulfide isomerase; CoQ, coenzyme Q; Cyt c, cytochrome c; NO, nitric oxide; ONOO−, peroxynitrite; GPx, glutathione peroxisome; Trx, thioredoxin; CAT, catalase. Complex I, consisting of a hydrophobic arm and a matrix-protruding hydrophilic arm, is the entry point for electrons from nicotinamide adenine diphosphate hydride (NADH) into the ETC117. The flavin mononucleotide (FMN) cofactor accepts electrons from NADH and transfers them along the Fe–S clusters to the CoQ reduction site118. In mitochondria, the orderly flow of electrons down the mitochondrial ETC from complex I to complex IV results in their eventual deposition into molecular O2 to form water119. Nevertheless, due to the “leaky” nature of electron transport in the ETC, electrons often escape to reduce O2 to partially form O2·−. In complex I, the FMN domain in the NADH-binding site and the ubisemiquinone in the CoQ-binding site can generate O2·−120. Furthermore, O2·− production can also be induced by reverse electron transport (RET) within complex I when the CoQ pool is highly reduced121. During RET, electrons enter complex II via succinate oxidation to reduce CoQ before being sequentially driven back into complex I instead of being forwarded to complex III122. Complex III is another crucial site for the production of mitochondrial ROS in the ETC through the reaction between O2 and ubisemiquinone123. Usually, the electrons in the ETC flow from the CoQ pool to complex IV. Nevertheless, in the presence of antimycin A, an inhibitor of the Qi site of complex III, a large amount of O2·− is produced from the Qo site of complex III due to the interaction between O2 and a ubisemiquinone bound to the Qo site124. Compared with O2·− produced by complex I during the RET process, complex III produces very low levels of O2·− in the absence of antimycin A125. Moreover, O2·− generated from complex III is mainly released into the intermembrane space, whereas O2·− produced by complex I tends to be released into the mitochondrial matrix126. Typically, the O2·− generated in the mitochondrial intermembrane space can be degraded by copper/zinc superoxide dismutase (Cu-ZnSOD, also known as SOD1)127. The O2·− generated in the mitochondrial matrix can be scavenged by manganese-dependent superoxide dismutase (MnSOD, SOD2) into H2O2, which is then consumed by glutathione peroxisome (GPx) and thioredoxin (Trx) enzymes in the mitochondria or through free diffusion across the membrane bilayer into the cytosol to be catalase (CAT) to finally degrade into water128,129. Alternatively, H2O2 can interact with the Fe–S clusters abundant in the ETC to form ·OH via the Fenton reaction130. In addition to complex I and III, at least ten other mitochondrial enzymes also contribute to ROS generation, including complex II, which releases O2·− toward the mitochondrial matrix. ROS-generating systems: NOXs Another significant source of intracellular ROS is the family of NOXs, which are located primarily in phagosomes within phagocytes, as well as in the subcellular locations, such as the plasma membrane, endoplasmic reticulum (ER) and mitochondria131. The human NOX family is composed of seven known isoforms, including NOX1-5, dual oxidase 1 (DUOX1) and DUOX2. All homologs of this family share conserved structures containing 6–7 transmembrane domains, one flavin adenine dinucleotide (FAD) and the NADPH-binding cytosolic domains, which primarily produce ROS for host defense and intracellular signaling functions instead of generating ROS as the byproduct132. NOX produces O2·− via a series of reactions once the electrons from NADPH flow to reduce FAD and are then transferred to the heme moiety for O2 reduction133. Generally, NOX1, NOX2, NOX3 and NOX5 mainly produce O2·−, whereas NOX4, DUOX1, and DUOX2 mainly generate H2O2 directly134. NOX2, originally termed gp91phox, was the first discovered NOX member and is responsible for the phagocyte respiratory burst. When the NOX complex in the plasma membrane of phagocytes, such as neutrophils, recognizes the foreign particles, it rapidly secretes O2·− to kill invading pathogens135. Notably, most NOX isoforms do not produce O2·− on their own but use different regulatory cytosolic factors136. For example, the activation of NOX2 requires the membrane-bound domain p22phox, together with the recruitment of several regulatory subunits, such as p40phox (also known as neutrophil cytosolic factor 4 (NCF4), p47phox (NCF1), p67phox (NCF2) and the small guanosine triphosphatase (GTPase) protein Rac1113. Upon cell stimulation, the p47phox subunit undergoes initial phosphorylation and conformational changes and then triggers the translocation of the activator subunits p40phox and p67phox to the membrane site for the subsequent recruitment of Rac1 for ROS generation137. Generally, the NOX isoforms NOX1, NOX2, NOX4 and NOX5 are expressed predominantly in carcinoma cells. Previous studies indicated that NOX-derived ROS are involved in the initiation and maintenance phases of tumor development138. For example, NOX1-generated ROS reportedly modulate several signaling pathways responsible for the extent and direction of invasion139. The mRNA and/or protein expression of NOX family members or their regulatory factors has been shown to be increased in various cultured cancer cells or human tumor tissues140. ROS-generating systems: Other organelles and enzymes In addition to the mitochondrial ETC and NOXs, various subcellular organelles, such as peroxisomes and the ER, also contribute to ROS generation. Peroxisomes are small membrane-bound organelles, whose primary function is to metabolize long-chain and branched-chain fatty acids via β-oxidation141. Several peroxisomal ROS-producing enzymes, such as acyl CoA oxidase, urate oxidase, and d-amino acid oxidase, can generate ROS, principally H2O2, which often cooperate with mitochondria142. The O2·− and nitric oxide (NO) are also generated by xanthine oxidase and NO synthase, respectively. Similarly, ROS-scavenging enzymes within peroxisomes, including CAT, peroxiredoxin (Prx), epoxide hydrolase, and some peroxisome-specific membrane proteins, can delicately modulate redox homeostasis and physiological cell signaling143. Imbalances in redox homeostasis in peroxisomes have been linked to diseases such as cancer, neurodegeneration, and type 2 diabetes144,145. Another crucial source of H2O2 is the ER, a well-orchestrated organelle responsible for the biosynthesis, folding and trafficking of cellular proteins146. During the process of oxidative protein folding, which is catalyzed by ER oxidoreductase 1 (ERO1) and protein disulfide isomerase to facilitate protein disulfide formation, one mole of H2O2 is generated as a byproduct of every disulfide bond formed147,148. The cytochrome P450 family of enzymes functions in steroid and lipid synthesis and oxidative detoxification of the xenobiotics149. The induction of the activity of ER-located cytochrome P450 2E1, which is the most active subunit for the production of ROS, may affect the progression and metastasis of advanced-stage breast cancer150. Disruption of the redox balance in the ER has been associated with the initiation of the unfolded protein response (UPR), which aims to restore ER homeostasis under conditions of ER stress150. Elevated levels of oxidizing or reducing agents within the ER can induce ER stress and trigger UPR activation151. This, in turn, can increase ROS levels by influencing mitochondrial activity. Numerous human diseases, including cancer, neurodegeneration, metabolic disorders, and chronic inflammation, have been linked to altered ER proteostasis and abnormal UPR signaling152,153. ROS-regulating antioxidant systems Under normal physiological conditions, the concentration of ROS is tightly regulated within a narrow range known as redox homeostasis. Maintaining redox homeostasis in cells depends on the delicate balance between ROS generation and elimination. To neutralize excessive ROS, cells are equipped with various antioxidant systems, which include (1) enzymatic scavengers: such as SOD, CAT, GPx, and Prx; and (2) non-enzymatic defenses: such as glutathione (GSH), vitamin C (ascorbic acid), vitamin E (mainly α-tocopherol), lipoic acid and bilirubin154. SODs constitute a group of the most important enzymatic metalloproteins for ROS detoxification, including cytosolic SOD1 (Cu-ZnSOD), mitochondrial SOD2 (MnSOD), and SOD3 (Cu-ZnSOD). SOD catalyzes the dismutation of O2·− into H2O2, and multiple enzymes further convert H2O2 to water and O2, including CAT, GPx, and Prx155. GPx is a group of enzymes that use GSH or Trx as electron donors to convert organic and inorganic hydroperoxides to alcohols, promoting H2O2 metabolism and safeguarding cell membrane structure and function against oxidative damage156. Mammals have eight identified GPx family members (GPx1‒GPx8), including five selenocysteine-containing proteins (GPx1‒GPx4 and GPx6) and three cysteine-containing proteins157. Moreover, GPx4 is an indispensable mammalian glutathione peroxidase that shields cells from harmful lipid peroxidation and governs a novel form of regulated necrotic cell death known as ferroptosis158. Hence, the interplay of diverse antioxidant cascades is crucial in determining the outcome of redox homeostasis. Among non-enzymatic compounds, reduced GSH and NADPH are the most widely recognized electron donors and relevant natural antioxidants159. GSH, a tripeptide-containing thiol group, is present in most living cells and functions as an electrophilic and oxidant scavenger through direct action or enzymatic catalysis160. When defending against ROS, GSH undergoes oxidation to form glutathione disulfide (GSSG), which is subsequently reduced back to GSH by the enzyme glutathione reductase (GSH-R) in the presence of NADPH, which is generated through metabolic coupling with the pentose phosphate pathway161. The GSH/GSSG ratio is widely recognized as an excellent indicator of redox levels162. Moreover, NADPH participates in various cellular processes in addition to the detoxification of ROS143. These include anabolic reactions that drive fatty acid and nucleotide synthesis. Additionally, NADPH plays a role in ROS generation in conjunction with NOXs, as discussed above. As a result, maintaining NADPH homeostasis pathways is crucial for balancing the production and utilization of NADPH to meet these diverse demands. Oxidative stress and the role of ROS in cancer The physiological and pathological role of ROS ROS are considered a double-edged sword capable of bestowing beneficial and toxic effects upon living systems90. ROS play a vital role in cell signaling pathways at moderate physiological levels through their ability to act as chemically modified biomolecules, such as proteins and lipids, which serve as critical messengers and modulators of various types of signal transduction, including cell proliferation, differentiation, and apoptosis143,163. Most cells exhibit a small oxidative burst upon stimulation by certain factors, such as cytokines and hormones164. This instantaneous redox imbalance can be quickly restored to cell homeostasis with the assistance of the endogenous antioxidant system165. Therefore, maintaining redox homeostasis is paramount for maintaining normal physiological functions and mitigating the incidence of diseases. However, oxidative stress occurs when excessive ROS production surpasses the cellular tolerance thresholds and the endogenous antioxidant defense system166. While oxidative stress is beneficial in some circumstances, such as inflammation for the immune response against invading pathogens, it is also detrimental to cells because of its irreversible changes to biomolecules. Specifically, oxidative stress can cause DNA damage, lipid peroxidation, and protein oxidation167. The oxidative-induced carbonylation, glycation, S-glutathionylation, and S-nitrosylation of proteins have been implicated in the pathogenesis of several diseases168, 169, 170. Chronic oxidative stress promotes double-stranded DNA breaks, especially in mitochondrial DNA, which is considered a significant contributor to diseases such as cardiovascular dysfunction, neurodegeneration, autoimmune diseases, diabetes, and cancer171, 172, 173. Despite decades of investigation, the precise roles of ROS in these diseases remain to be fully elucidated. Oxidative stress and cancer Cancer is a disease characterized by aberrant cellular proliferation with the potential to metastasize to distant sites within the body. The majority of cancer cells exhibit altered metabolism, commonly referred to as the Warburg effect or aerobic glycolysis174. This metabolic shift is characterized by a high rate of glucose uptake and fermentation to lactic acid, even in the presence of adequate O2 levels175. The pathophysiology of cancer is strongly associated with oxidative stress. Typically, the causal relationship between tumorigenesis and elevated ROS levels remains controversial176. The etiology of elevated levels of ROS in tumors can be categorized as follows: 1) Metabolic abnormalities: heightened metabolic activity and increased mitochondrial energetics result in an imbalance in intracellular ATP metabolism and weakened oxidative phosphorylation177, 178, 179. Electrons within ETC complexes tend to reduce O2, leading to elevated levels of ROS, as mitochondria are the primary source of ROS production180. 2) Tumor microenvironment factors: Oxidative stress and inflammatory reactions within the tumor microenvironment can stimulate ROS production in tumor cells109,181. For instance, hypoxia promotes the activation of key regulators, particularly hypoxia-inducible factor 1 (HIF-1), to facilitate ROS generation, which upregulates HIF-1, completing a positive feedback loop of ROS induction in cells182. Moreover, inflammatory cells, such as macrophages and T lymphocytes, surrounding tumor cells secrete related inflammatory factors and cytokines, such as tumor necrosis factor-α (TNF-α) and interferon-gamma (IFN-γ), which increase ROS levels183. 3) Activation of oncogenic signaling: During tumorigenesis, the activation of various proteins, such as nuclear factor-κB (NF-κB), nuclear factor E2-related factor-2 (NRF2) and phosphatidylinositol 3-kinase (PI3K), and their downstream signaling pathways can induce cellular ROS production184,185. In addition, DNA mutations, such as those in C-MYC, RAS, P53, and SIRT3, may also increase oxidative stress and inflammation in cancer cells186, 187, 188, 189. Importantly, the causes of elevated ROS levels within tumors are multifaceted and intricately linked to various factors, including the type of tumor, stage of tumor progression, and individual variability among patients190,191. Additionally, elevated levels of ROS also have multiple effects on tumorigenesis and cancer progression, which generally promote tumor cell proliferation, metastasis, and drug resistance, as well as the regulation of the tumor microenvironment89. The acquisition of DNA damage and genomic instability induced by oxidative stress can facilitate the activation of oncogenes and disable tumor suppressor genes192. For example, in normal cells, increased ROS can activate the wild-type tumor suppressor gene P53, which triggers stress responses and DNA repair to eliminate ROS-mediated damage193. Nevertheless, such DNA repair mechanisms are compromised by mutation of the P53 gene by ROS, thus accelerating tumorigenesis194. Moreover, oncogenic mutations in NRF2 and activation of NF-κB have been demonstrated to mitigate ROS damage by inducing the expression of various antioxidant enzymes, thereby promoting carcinogenesis195,196. The overexpression of DNA damage-inducible transcript 4 antisense RNA 1 (DDIT4-AS1) has been observed in numerous human malignancies, including prostate cancer, ovarian cancer, and breast cancer197. This recent study revealed that oxidative stress, ER stress, and hypoxia are associated with DDIT4 dysregulation, which promotes the proliferation, migration, and invasion of triple-negative breast cancer via the activation of autophagy. Other relationships between ROS and cancer have also been elucidated. For instance, ROS facilitate the invasion and metastasis of tumor cells by activating multiple survival cascades, such as the PI3K/AKT/mTOR and P38/mitogen-activated protein kinase (MAPK), extracellular signal-regulated kinase (ERK) and c-Jun N-terminal kinase (JNK) pathways198,199. ROS can promote the activation and transformation of tumor-associated cells, such as the polarization of M1 macrophages and the activation of tumor-associated fibroblasts, while also affecting the antigen expression and immune escape mechanisms of tumor cells200,201. These effects can contribute to tumor growth and inhibit clearance by the immune system202. As tumors progress, they start to invade nearby tissues and disseminate to metastasize to organs or tissues, which is the cause of poor prognosis and death in patients with many types of cancer203, 204, 205. ROS have been demonstrated to participate in metastasis processes, such as invasion of surrounding tissue, intravasation into blood vessels or lymphatic vessels, survival in the circulation, extravasation from blood vessels or lymphatic vessels, and re-establishment and growth of secondary tumors206, 207, 208. Generally, ROS impact the behavior of both cancer cells and the stromal components of tumors to facilitate tumor metastasis89. For example, ROS-based signaling stimulates rearrangements of the actin cytoskeleton of cancer cells to form membrane protrusions, such as invadopodia or pseudopodia, which aid in the proteolysis and invasion of cancer cells209. Furthermore, studies have revealed that an increase in both membrane and mitochondrial ROS drives the degradation of the extracellular matrix by stimulating matrix metalloproteinases, thereby allowing cancer cells to filter and invade neighboring tissue210. In addition, epithelial–mesenchymal transition (EMT), a process by which epithelial cells lose their cell polarity and cell–cell adhesion ability and gain migratory and invasive properties, can also be induced by the increase in membrane and mitochondrial ROS211, 212, 213. N-acetyl-l-cysteine (NAC)-treated cancer cells can successfully revert to an epithelial phenotype via a process called mesenchymal–epithelial transition (MET)214. Although recent progress in ROS biology has shed light on the mechanisms underlying ROS-related physiology and pathology, a more comprehensive understanding of the interplay between ROS and cancer development is still warranted. Opportunity for the use of ROS in cancer therapy Oxidative stress within a certain range at the tumor site has a positive effect on the cancer progression. Many cancer cells possess robust antioxidant defenses, such as increasing GSH and NADPH generation to adapt to elevated levels of ROS, establishing a new state of steady redox equilibrium215. Accordingly, cancer cells rely more heavily on their internal antioxidant system to maintain the newly established redox balance216. Therapeutic approaches focusing on perturbing the redox status of cancer cells could constitute a viable strategy for cancer treatment176,217. Intuitively, it seems reasonable that the application of antioxidants, by decreasing ROS levels, could protect cells from oxidative damage and hence reduce cancer incidence and transformation. Nonetheless, epidemiological trials with antioxidant supplements either have shown no effect or, in some cases, have indicated that certain antioxidants might increase cancer risk218. This enhanced antioxidant capacity of cancer cells is often associated with resistance to chemotherapy, as several currently used anticancer drugs are known to drive cytotoxicity at least partially via oxidant generation219. If the adaptive endogenous redox balance state can be eliminated through external intervention, such as the use of nanomaterials that produce ROS, the resulting ROS levels beyond the threshold of cellular tolerance can overwhelm the antioxidant system, leading to the selective destruction of cancer cells220. Indeed, cell fate is largely tailored to a particular case, including considerations of the stage and type of tumor, oxidant levels in the tumor niche, and the endogenous antioxidant capacity of the tumor site143,221. Stepping from ROS physiology and pathology to ROS oncology, the development of ROS-relevant studies has promoted the emergence of several therapeutic modalities, such as chemotherapy, radiotherapy, and X (photo, chemo, sono, or thermo)-dynamic therapy222, 223, 224, 225, 226. Chemotherapy, which uses mainly small-molecule chemicals, is a conventional cancer treatment used in the clinic. Certain chemotherapeutic drugs generate free radicals that destroy cancer cells through oxidative reactions, inducing oxidative stress and damage to biological molecules such as DNA and proteins, ultimately leading to cancer cell death227. The oxidative chemotherapy agents primarily include polycyclic aromatic compounds (e.g., cyclophosphamide and methotrexate), platinum-based compounds (e.g., cisplatin and carboplatin), anthracene derivatives (e.g., anthracyclines) and doxorubicin (DOX), which inhibit DNA synthesis and cell division in cancer cells228, 229, 230. Radiotherapy damages cellular DNA by increasing ROS generation through various mechanisms, including extracellular water radiolysis, intracellular metabolic changes, or mitochondrial damage, resulting in oxidative bursts that can exceed the ability of cellular ROS scavengers ability and damage nearby biological substances231. Compared with other emerging redox-regulating X-dynamic therapeutic modalities, PDT dates back to 1900, when Oscar Raab first demonstrated the light-triggered cytotoxicity of Paramecium by acridine orange232. As a noninvasive therapeutic modality, PDT has been approved in the clinic for the treatment of various malignancies, including non-small cell lung cancer, head and neck cancer, pancreatic cancer, esophageal cancer, and bladder cancer99,233. It also treats some non-neoplastic diseases, such as macular degeneration, acne, and mucosal lesions234, 235, 236. PDT can be further categorized into two subtypes on the basis of the photochemical reaction mechanism: type I-PDT and type II-PDT. Under light irradiation, the photosensitizers transform from their ground singlet state to the excited triplet state and then participate in two pathways237. In type I-PDT, photosensitizers directly interact with the biological substrate to form free radicals, such as O2·− and ·OH238. In type II-PDT, the photosensitizers transfer the obtained energy to the surrounding ground state oxygen (3O2) directly for the generation of cytotoxic singlet oxygen (1O2)239. In recent years, our group and others have made significant progress with the assistance of nanotechnology to address the limitations of PDT in clinical applications, including improving the targeting ability, water solubility, photobleaching effect, and tissue penetration depth of photosensitizers, and alleviating tumor hypoxia to increase 1O2 generation15,58,240, 241, 242. The ROS generated during PDT can directly exert an anticancer effect by increasing intracellular oxidative stress. Additionally, along with the high level of ROS in the tumor microenvironment, ROS can initiate a second chemical cascade via specific ROS-responsive moieties, which can promote controlled drug release243. The tumor microenvironment has a unique redox microenvironment with an elevated H2O2 concentration at the micromolar level, which distinguishes it from the surrounding normal environment. However, the heterogeneity in intratumoral ROS concentrations due to differences in tumor types, stages, and patient conditions may hinder the ROS responsiveness of the designed DDSs92. Therefore, the development of ROS-responsive DDSs for cancer PDT is expected to establish a self-feedback effect and augment on-demand drug release, thereby minimizing the side effects caused by the off-target dispersal of drugs on healthy tissues. ROS-responsive DDSs for cancer PDT Polymer-based DDSs have been widely considered efficient vehicles for delivering numerous therapeutic agents244. The nanometer dimensions of these DDSs enable them to preferentially accumulate at the tumor site via the enhanced permeation and retention (EPR) effect after intravenous administration. In addition, hydrogel-based DDSs leverage their biocompatible and tunable matrix to achieve sustained drug release upon local application245. Additional engineering-specific targeting ligands on the surface of DDSs can achieve more efficient cancer cell internalization via receptor-mediated endocytosis pathway246. More importantly, the integration of DDSs with stimuli–responsive properties, such as sensitivity to elevated ROS levels, holds great potential for precise cancer treatment247. ROS-responsive DDSs can release their cargo via bond cleavage, hydrophilicity changes, and charge reversal induced by pathological oxidative stress within tumors248. Table 1 lists the chemical structures of the representative ROS-responsive groups and their oxidation products. This section discusses the recently designed ROS-responsive DDSs according to the types of ROS-responsive chemical moieties. Table 1Representative ROS-responsive groups and their oxidation products.Table 1ROS-responsive groupChemical structure and oxidation productThioetherImage 1ThioketalImage 2SelenideImage 3DiselenideImage 4TellurideImage 5Boronic esterImage 6Boronic acidImage 7OxalateImage 8Oleic acidImage 9 Representative ROS-responsive groups and their oxidation products. Sulfide-containing linkers Thioether Sulfur, an essential trace element critical for human health, is predominantly incorporated into proteins as part of their amino acid constituents249. The sulfur-containing polymers are vulnerable to ROS because of the relatively low bond dissociation energies of C–S at 272 kJ/mol and S–S at 240 kJ/mol250. When exposed to ROS, thioether moieties (generally R-S-R) are prone to be oxidized into polar sulfoxide or sulfone (Table 1), undergoing a transformative phase shift from a hydrophobic to a more hydrophilic state251. This precipitous transition from hydrophobicity to hydrophilicity can precipitate the rapid dissolution of polymers, facilitating the ROS-triggered release of therapeutic agents. Consequently, thioether-containing polymers are extensively used to develop ROS-responsive DDSs for cancer therapy252. There are two main strategies for the synthesis of thioether-containing polymers. The first approach involves strategically incorporating thioether groups into the polymer main chain through ring-opening polymerization, group transfer polymerization, and step-growth polymerization techniques, synthesizing polymers with C–S–C linkages along their backbone253. Poly(propylene sulfide) (PPS) is the most well-known example of step-growth polymerization. In 2004, Hubbell and coworkers developed an ABA block polymer consisting of hydrophilic poly(ethylene glycol) (PEG) and hydrophobic PPS to fabricate ROS-responsive polymeric vesicles. After treatment with H2O2, the most prevalent ROS in the body, the vesicles are destabilized and undergo a phase transition to release the payloads eventually254. By utilizing this concept to increase the application potential of black phosphorus quantum dots (BPQDs) in the biomedical field for PDT, Li et al.255 recently developed a near-infrared region (NIR)/ROS-sensitive BPQDs nanovesicles (BPNVs) self-assembly of amphiphilic BPQDs grafted with PEG (ended by pyrene) and ROS-sensitive PPS through π–π stacking (Fig. 3A). During the self-assembly process, immunoadjuvant CpG oligodeoxynucleotides (ODNs) are encapsulated in the cavities of BPNVs, forming BPNVs-CpG nanoparticles of approximately 100 nm in size for appreciable passive tumor-targeting ability (Fig. 3B). Upon reaching the tumor site, the BPNVs undergo dissociation to 15 nm under NIR laser irradiation, which occurs as a result of the degradation of the ROS-sensitive PPS triggered by the free radicals generated during the PDT process (Fig. 3C). Next, the loaded immunoadjuvant CpG was released into the tumor tissue, where it is captured along with the tumor antigens by the antigen-presenting cells. The combination of PDT with immunotherapy has garnered widespread attention because of its ability to evoke a robust immune response within the body256. The in vivo assessment shown in Fig. 3D revealed that the treatment of the mouse 4T1 breast cancer model with BPNVs-CpG with laser irradiation significantly increased the serum levels of TNF-α, interleukin-6 (IL-6), and IL-12, indicating a remarkable immune response. Moreover, BPNVs-CPG-mediated PDT-immunotherapy inhibited the proliferation of distant tumors (Fig. 3E and F) and blocked metastatic nodules in the lungs of the mice (Fig. 3F)255. Harnessing the “remote-controlled” disassembly capability of these nanovesicles via NIR laser irradiation represents an exemplary strategy for optimizing their therapeutic efficacy while minimizing the adverse effects of immunotherapeutic agents in cancer treatment257. Example of ROS-responsive black phosphorus quantum dots vesicles. (A) Schematic illustration of BPNVs–CpG synthesis and the synergistic cancer photodynamic-immunotherapy. (B) Morphology and characterisation of BPNVs. a) Transmission electron microscope (TEM) image of BPQD (scale bar = 100 nm). b) TEM image of BPNVs (scale bar = 200 nm). c) Scanning electron microscope image of BPNVs (scale bar = 1 μm). (C) Hydrodynamic diameter distribution of dissociated BPNVs under near-infrared region laser irradiation. (D) Serum levels of IL-6, IL-12, and TNF-α in different treatments. (E) Distant tumor growth curves. (F) The relative tumor volume changes and representative image of mice treated with 1) PBS, 2) PBS + 660 nm laser, 3) CpG alone, 4) BPNVs + 660 nm laser, and 5) BPNVs-CpG + 660 nm laser. Reprinted with the permission from Ref. 255. Copyright © 2020 Wiley. The incorporation of thioether groups into the side chain of polymer is the other prevalent method, which significantly expands the range of thioether-enriched polymers258. One advantage of this approach is that it leaves the polymer backbone intact and is unaffected by ROS. These thioether-containing polymers can be synthesized through the polymerization of specific thiol-pendant monomers and post-polymerization modification techniques such as radical thiolene/thiolyne reactions and nucleophilic substitutions of thiols. For instance, Xu and colleagues259 developed a red blood cell (RBC) membrane camouflaged polyphosphoester (PPE) nanoparticles (RBC@PPEMTO/PFA) for targeted delivery of the photosensitizer mitoxantrone (MTO) and the oxygen carrier perfluoroalkane (PFA) (Fig. 4A). Upon accumulation at the tumor site, the thioether side chains of PPE respond to the elevated intratumoral ROS levels induced by the photoactivation of MTO in the presence of oxygen, triggering a hydrophobic-to-hydrophilic transition within the PPE core, self-accelerating the release of MTO and enhancing the sequent therapeutic efficacy. In addition, various bioactive molecules, including peptides260, chemotherapeutic drugs261, photosensitizers262, and contrast agents, have been engineered into the thioether-containing side chains of polymers for ROS responsiveness253. Certain small molecule chemotherapeutic agents and photosensitizers have been shown to effectively induce pyroptosis, a highly inflammatory form of programmed cell death, thereby eliminating cancer cells and triggering strong antitumor immunity 263,264. Therefore, Ma et al.262 developed an intelligent laser and ROS/GSH triple-responsive polyprodrug nanoinducer, CCNP, to precisely regulate gasdermin E (GSDME)-mediated tumor pyroptosis (Fig. 4B). In this innovative nanoinducer, the chemotherapeutic agent camptothecin (CPT) is attached to the amphiphilic unimolecular polyprodrug via an ROS-sensitive 2′-thiobisacetic acid anhydride group. Moreover, the photosensitizer chlorin e6 (Ce6) is integrated through π-π stacking and hydrophobic interactions, amplifying the therapeutic efficacy and synergistically inducing pyroptosis in tumor cells (Fig. 4C). Pyroptosis enhances tumor immunogenicity by releasing cellular contents, eliciting antitumor immune responses, generating strong systemic immune effects against distant tumorigenesis and progression (Fig. 4D)262. Notably, ROS-responsive polymers containing thioether groups have not been widely used in the biomedical field because of the relative stability and inherent chemical properties of sulfides. Examples of thioether groups in the side chains of polymers. (A) Schematic illustration of the synthesis of RBC@PPEMTO/PFA and the sequential PTT/PDT/chemotherapy effect for synergised immunotherapy. Reprinted with the permission from Ref. 259. Copyright © 2023 Wiley. (B) Schematic illustration of the synthesis of CCNP and the mechanism of anti-tumor immune response by precisely inducing pyroptosis. (C) Representative brightfield photographs of pyroptosis and confocal laser scanning microscope images of high mobility group protein B1 (HMGB1) and calreticulin (CRT) in 4T1 cells under various treatments (scale bars = 50 μm). (D) a) Typical flow cytometry images of matured DCs (CD80+ CD86+) gating on CD11c+ MHC-Ⅱ+ cells in draining lymph node (DLN). b) Typical flow cytometry images of CD8+ and CD4+ T cells gating on CD45+ CD3+ cells in tumor. c) Typical flow cytometry images of Tregs (CD25+ and Foxp3+) gating on CD4+ cells. Reprinted with the permission from Ref. 262. Copyright © 2023 Elsevier. Thioketal Thioketal (TK) is a subclass of the thioether group, which is commonly recognized as a protective group for carbonyls or alcohols in organic synthesis because of its stability and ease of synthesis265. Upon oxidation in the presence of ROS, such as H2O2, O2·−, and ·OH, TK can be cleaved into two thiol fragments and acetone (Table 1). This process is considered biodegradable and nontoxic266,267. The TK group has recently played a significant role in the development of nanoprodrugs and nanocarriers, which can be designed to release drugs in specific diseased tissues, such as tumors54,268, spinal cord injury269, acute lung injury270, and myocardial infarction271. Recent studies have demonstrated that the sizes and shapes of nanoparticles significantly influence tumor penetration and retention, subcellular distribution, and intracellular drug release272,273. Small-sized nanoparticles (<50 nm) exhibit superior deep tumor penetration, whereas nanoparticles 100–200 nm in diameter are optimal for prolonged blood circulation274,275. Various size-transformable nanosystems have been constructed to surmount the dilemma of size requirements for different delivery processes. These nanosystems often maintain a relatively large diameter for efficient circulation while smartly changing into a smaller size within the tumor microenvironment in response to specific stimuli, ensuring optimal delivery precision276,277. To address this issue, Zhang and coworkers278 have ingeniously crafted a unique, size-transforming nanoplatform, CPIM nanoparticles, which leverages ROS-sensitive TK groups, thereby enhancing the targeted delivery of therapeutics to the lesion site. Briefly, polymer–drug conjugates and polyamidoamine (PAMAM) dendrimers were used first to encapsulate the photosensitizer IR780 and the indoleamine 2,3-dioxygenase (IDO) inhibitor 1-methyl-d-tryptophan (1-MT). This process yielded small “bomblets” (<10 nm in diameter), which were further crosslinked with TK as the crosslinker to form larger “cluster-bomb”-like nanoplatforms with a diameter of approximately 100 nm (Fig. 5A)278. The negatively charged chondroitin sulfate (ChS) shell was designed to detach in the presence of locally overexpressed hyaluronidase (HAase) within the tumor microenvironment. Moreover, the particle size of CPIM decreased significantly after incubation with H2O2, which was further confirmed by transmission electron microscopy (TEM) images (Fig. 5B and C). These results indicated that the large “cluster-bomb” can release drug-loaded “bomblets” within the tumor microenvironment and promote their penetration within the tumor for photoimmunotherapy (Fig. 5D and E)278. In B16F10-tumor-bearing mice, the “cluster-bomb” exhibited a significant inhibitory effect on tumor growth due to cytotoxic phototherapy and IDO pathway inhibition (Fig. 5F). Example of ROS-sensitive, size-transforming nano-platform CPIM NPs. (A) Schematic illustrating the synthesis and disassembly of CPIM NPs and their performance in vivo. (B) Changes in the size distribution of CPIM with H2O2 treatment. (C) Changes in transmission electron microscope images of CPIM with H2O2 treatment. (D) Infrared thermal images of different solutions with irradiation. (E) Confocal laser scanning microscope images of B16F10 cells incubated with CPI or CPI (L+). (F) Primary tumor growth curves following different treatments (n = 5). Reprinted with the permission from Ref. 278. Copyright © 2021 Elsevier. Multimodal imaging-guided cancer therapy offers the advantages of precise lesion localization, real-time treatment monitoring, and therapeutic outcome assessment, thereby enhancing the accuracy and safety of the treatment while minimizing damage to surrounding healthy tissues279. Recently, to alleviate fibrosis in immune-excluded tumors and improve the recruitment of tumor-infiltrating T lymphocytes, Zhuo's group280 engineered a metal matrix protease-2 (MMP-2)/ROS-sensitive nanovesicle, which integrates the inhibitor of transforming growth factor β receptor 1 (TGFR1) LY2157299, the bromodomain-containing protein 4 (BRD4) inhibitor JQ-1, and the gadolinium (Gd3+)-chelated photosensitizer pyropheophorbide a (PPa) (Fig. 6A). The PPa within the nanovesicles induced notable ROS generation upon NIR laser irradiation, which in turn cleaved ROS-responsive thioketal linkages, enabling spatiotemporally tunable JQ1 release. This process abrogates programmed death ligand 1 (PD-L1) expression on tumor cells, overcoming adaptive immune resistance. Moreover, the Gd3+-chelated PPa enabled fluorescence, photoacoustic, and magnetic resonance triple-modal imaging-guided PDT. Examples of ROS-sensitive NPs with TK groups. (A) Schematic of the nanovesicles to avoid the immune resistance of immune-excluded tumors. Reprinted with the permission from Ref. 280. Copyright © 2023 Springer Nature. (B) Schematic diagram of P53 mRNA/ICG NPs self-assembly. (C) Histogram analysis (EGFP positive cells) of the in vitro transfection efficiency. Reprinted with the permission from Ref. 282. Copyright © 2023 American Chemical Society. Messenger RNA (mRNA)-based therapeutics constitute a new generation of medicine and have shown great potential for cancer treatment281. Zhou and coworkers282 reported self-assembled nanoparticles composed of a TK group-containing dihydrolipoic acid oligomer (o-DHLA) for the codelivery of P53 mRNA and the photosensitizer indocyanine green (ICG) (Fig. 6B). In contrast to the free mRNA group, o-DHLA nanoparticles markedly enhanced the transfection efficiency of P53 mRNA, resulting in a notably high percentage of EGFP-positive cells, which reached 89.5% (Fig. 6C)282. These pioneering studies pave the way for the codelivery of photosensitizers and mRNAs utilizing ROS-reactive polymers, facilitating the development of highly efficient synergistic therapies for PDT and mRNAs. Cuproptosis, a novel form of cell death, relies on copper (Cu) ionophores to transport Cu into cancer cells, inducing cell death283. However, existing Cu ionophores, which are small molecules with short half-lives in blood, fail to deliver sufficient amounts of Cu to cancer cells. Accordingly, Guo et al.284 presented the design of an ROS-sensitive polymer (PHPM) for the coencapsulation of Cu ionophores elesclomol (ES) and Cu into nanoparticles (NP@ESCu) (Fig. 7A). This design ensures that upon entry into cancer cells, ES and Cu are released due to high intracellular ROS levels, synergistically inducing pronounced mitochondrial dysfunction and cell death of in BIU-87 bladder cancer cells through cuprotosis (Fig. 7B‒D)284. Wu's group285 further advanced this approach with a ROS-responsive polymer that codelivers ES and cinnamaldehyde, demonstrating effective tumor suppression by combining cuproptosis and immunotherapy in a 4T1 tumor model (Fig. 7E). The ROS-responsive DDSs, as illustrated by these studies, effectively enhance the circulation stability and targeted delivery of Cu ionophores such as ES for cuproptosis. This strategy not only refines the specificity and efficacy of cuproptosis induction but also increases the capacity of the immune system to combat cancer cells. This dual-pronged strategy paves the way for the development of PDT supplemented by cuprotosis, offering a novel and promising horizon for more effective tumor suppression286,287. Example of ROS-sensitive delivery systems for Cu ionophores elesclomol. (A) Design of NP@ESCu to induce cuproptosis. (B) Representative confocal laser scanning microscope images of the 3D cell spheroids of BIU-87 cells with various treatments. (C) Bio-TEM images of BIU-87 cells with NP@ESCu. The white box and arrow marked the location of the mitochondria. (D) Schematic illustration of the possible mechanism of action of NP@ESCu in promoting cuproptosis. Reprinted with the permission from Ref. 284 Copyright © 2023 Wiley. (E) Schematic illustration of the preparation and mechanism of ECPCP. Reprinted with the permission from Ref. 285. Copyright © 2024 Wiley. Selenide and diselenide-containing linkers Selenium, an indispensable trace element integral to proteins such as selenocysteine, is recognized for its significant benefits in enhancing immunity and cancer prevention, driven by its robust antioxidant capabilities288,289. Notably, the comparatively diminished electronegativity of selenium relative to sulfur endows its diselenide (Se–Se) bonds with lower energy levels (172 kJ/mol), a characteristic that significantly influences its redox behavior and biological activity290. Hence, the lower bond energy of selenide and diselenide compounds, making them more sensitive to mild stimuli than sulfur analogs, presents a compelling rationale for their use in constructing ROS-responsive DDSs for PDT, capitalizing on the complex redox changes associated with cellular or organismal disorders. The diselenide bonds can undergo cleavage through both oxidation and reduction processes, resulting in the formation of selenic acid and selenol, respectively (Table 1)291. Inorganic mesoporous silica nanoparticles (MSNs), known for their porosity, large surface area, facile modification, and low toxicity, hold great promise for drug delivery across a spectrum of therapeutics, including proteins292,293, peptides294, and nucleic acids295,296. However, its utility is somewhat constrained by its limited drug loading capacity, slow metabolic clearance, and potential drug leakage issues. In recent years, several research groups have used diselenide bond-containing organosilica moieties to construct biodegradable MSNs to burst the release of biologically active ingredients in response to either the oxidative tumor microenvironment or the bioreductive intracellular environment (Fig. 8A)292,293,297,298. Examples of ROS-sensitive nanocarriers based on selenium. (A) Schematic of the preparation of Se-MSN-PEG for cancer photo-immunotherapy with cascade medication release. Reprinted with the permission from Ref. 298. Copyright © 2022 Elsevier. (B) Illustration of preparation NP-PDT@Reg. (C) Confocal laser scanning microscope images and the corresponding quantification of the expression of CD206 in RAW264.7 cells after incubation with different formulations. (D) Representative flow cytometry plots and quantification analysis of M1 (F4/80+ CD80+) macrophages in tumors of various groups. Reprinted with the permission from Ref. 301. Copyright © 2023 Wiley. Organic conjugated polymers, characterized by their delocalized π-conjugated backbones, are macromolecules that have gained significant attention in the fields of bioimaging, PDT, and photothermal therapy (PTT) due to their remarkable optical characteristics299,300. Nevertheless, the nondegradable nature of conjugated polymers in the in vivo environment and associated long-term toxicity concerns limit their utility in biomedical applications. To increase the biodegradability and reduce the toxicity of conjugated polymers for cancer treatment, Wan and colleagues301 developed a pioneering ROS-sensitive diselenide-containing polymer (P1), which intricately coassembled with a Bodipy-enhanced pseudoconjugate polymer (PSPBodipy) and the multikinase inhibitor regorafenib (Reg), under the therapeutic platform NP-PDT@Reg (Fig. 8B). This system, upon 808 nm laser activation, efficiently releases Reg to counteract tumor hypoxia and normalize the vasculature, enhancing nanoparticle and oxygen tumor infiltration. Moreover, it increases ROS levels for potent tumor cell elimination, induces immunogenic cell death (ICD) to stimulate antitumor immunity, and redirects tumor-associated macrophages from the immunosuppressive M2 phenotype to the proinflammatory M1 phenotype, thereby promoting the tumor microenvironment and fortifying cancer PDT-immunotherapy (Fig. 8C and D). Leveraging the abundant resources, inherent biocompatibility, facile chemical modification, and intrinsic bioactivity of natural polysaccharides such as hyaluronic acid (HA), chitosan, and starch, a myriad of chemically modified polymeric conjugates have emerged, serving as efficacious drug carriers in advanced DDSs13,302. Our group has recently designed an NIR and redox dual-activatable platform based on selenium nanoparticles15. This advancement minimizes skin photosensitization in PDT by integrating the photosensitizer Ce6 with HA through a diselenide bond, creating an amphiphilic polymer (HSeC) capable of encapsulating the PTT agent IR780 into HSeC/IR nanoparticles. These nanoparticles, which are approximately 130 nm in size, mitigate skin phototoxicity via fluorescence resonance energy transfer from Ce6 to IR780, as evidenced by significantly reduced skin damage in vivo. In another representative study, Tan et al.303 integrated the anticancer drug doxorubicin with hydroxyethyl starch (HES) to fabricate HES-SeSe-DOX conjugate via a diselenide bond, which self-assembled with Ce6 into HES-SeSe-DOX/Ce6 nanoparticles. This demonstrated the remarkable in vivo safety of HES, with 3T3 cell viability exceeding 92% after incubation. Selenium/diselenide-bonded macromolecules, which span inorganic to organic and natural materials, offer a promising strategy to enhance delivery efficacy. Despite their potential, the development of such selenium-containing compounds is still in its nascent phase, presenting fertile ground for future innovation. The inherent ROS/glutathione reactivity of selenium raises concerns regarding its potential for nonspecific activation, which could inadvertently lead to premature drug release prior to reaching the intended tumor target. Telluride-containing linkers As a typical chalcogen element, tellurium (Te) has garnered significant interest due to its remarkable electronic properties and pronounced thermo-optic effects304, 305, 306. Organotellurium compounds can be oxidized by as low as 100 μmol/L H2O2 or 2 Gy of gamma irradiation307. In biomedical applications, telluride materials have proven useful in wound healing and tumor ablation308, 309, 310, 311. Notably, the Te–Te bond energy, standing at 138 kJ/mol, is considerably lower than that of its selenium counterpart, Se–Se. This unique characteristic endows telluride-containing compounds with heightened sensitivity to ROS, sparking considerable research enthusiasm, especially for the development of ROS-responsive drug carriers. Fan and coworkers312 fabricated a ROS-responsive tellurium-enriched platform via layer-by-layer assembly that integrated chemotherapy and PDT. The system capitalized on the ability of porphyrin to produce 1O2 upon light irradiation, while the tellurium component in the micelles was sequentially oxidized by ROS to form a more hydrophilic Te Created by potrace 1.16, written by Peter Selinger 2001-2019 O group (Table 1), enhancing drug release. In addition to porphyrin, they also harnessed ICG as a photosensitizer, employing the amphiphilic block copolymer PEG5000-Tellurium-Polyurethane-PEG5000 (PEG-PUTe-PEG) to encapsulate cisplatin (CDDP) and ICG concurrently (Fig. 9A)313. These nanoparticles effectively co-delivered both agents, with the Te–Pt coordination interaction facilitating a high CDDP loading of up to 30.62% (Fig. 9B and C). Notably, the presence of 808 nm laser irradiation triggered ICG-generated 1O2 to oxidize tellurium bonds, diminishing the coordination interaction and accelerating CDDP release from less than 20% to approximately 65% within 10 min (Fig. 9D and E). Postirradiation, the nanoparticles transition from a dispersed spherical structure to a distinct core‒shell morphology, with the shell retracting as the core expands (Fig. 9F). In vivo, the NIR laser promoted the rapid and targeted release of CDDP at the tumor site, synergistically enhancing PDT, PTT, and chemotherapy. These results underscore the dual role of Te–Pt coordination in bolstering drug delivery stability and precision313. Despite their promise, research into tellurium-containing nanomaterials for drug delivery is nascent, with challenges including structural diversity, degradation, metabolism elucidation, size control, surface modification, and biocompatibility optimization. Example of telluride-containing ROS-sensitive drug delivery systems. (A) Construction and disassembly of NPs-Pt-ICG. (B) The encapsulation yields of ICG. (C) The encapsulation yields of CDDP. (D) Drugs release profiles of NPs-Pt-ICG at room temperature. (E) CDDP release profiles of NPs-Pt-ICG with near-infrared laser exposure. (F) Transmission electron microscope image of NPs-Pt-ICG with the near-infrared laser exposure. Reprinted with the permission from Ref. 313. Copyright © 2017 Elsevier. Boronic acid/ester-containing moieties Boric acid, renowned for its biocompatibility, plays a significant role in biomedical research and has been endorsed by the Food and Drug Administration (FDA) for clinical use, as exemplified by the approval of ixazomib citrate in 2015, which incorporates a boronic ester group for multiple myeloma treatment314. The distinctive reactivity of boronic acids and their esters with H2O2 and ONOO−, resulting in the cleavage of the C–B bond and the generation of phenol (Table 1), positions these compounds to be particularly promising in the development of fluorescent probes, ROS-sensitive nanocarriers, and prodrugs, expanding their utility in the medical field315, 316, 317, 318. Autologous tumor cell-based vaccines (ATVs) are emerging as a transformable approach for personalized immunotherapy319. To enhance their therapeutic efficacy in patients, Fang's group320 designed PDT-motivated AVTs (P-ATV), which utilized a 9-fluorenyl methoxycarbonyl (Fmoc)–KCRGDK-phenylboronic acid (FK-PBA) hydrogel to stimulate the local immune system and activate neoepitope-specific CD8+ T cells (Fig. 10A). The FK-PBA component, featuring a sialic acid-targeting moiety, was injected directly into postsurgical tumor remnants. Concurrently, polyethyleneimine-conjugated Ce6 (PEI-Ce6) is applied to the surface of autologous tumor cells through electrostatic interaction, yielding PC-Cells. In the residual tumor environment, the FK-PBA hydrogel is activated by PC-Cells dispersed in a Na2CO3 solution on demand. In postoperative CT26 mouse models, treatment with PC-Cell@gel plus laser showed elevated levels of TNF-α and IFN-γ compared to treatments lacking either component (Fig. 10B and C). This synergistic therapy correlated with improved survival outcomes, significantly extending the lives of mice and preventing tumor relapse. In stark contrast, mice in the control groups without the combined PDT and ATV intervention did not survive beyond 25 days (Fig. 10D). These findings underscore the efficacy of the integrated ROS-responsive PDT and ATV approach in thwarting postoperative tumor recurrence. Examples of ROS-sensitive polymers based on boronic acid/ester. (A) Illustration of the formation and function of PC-Cell@gel. (B) Serum TNF-α and (C) IFN-γ levels on Days 3, 6, and 9 after treatments, respectively. (D) Survival percentage of CT26 recurrent tumor-bearing mice with different treatments. Reprinted with the permission from Ref. 320. Copyright © 2020 AAAS. (E) Schematic for multifunctional nanomedicine in targeted abdominal aortic aneurysm treatment. (F) Representative immunofluorescence staining photos of co-localization of Cy5-TPTN with α-SMA+ VSMCs, CD68+ macrophages, and CD177+ neutrophils. (G) The mean maximal diameter of the abdominal aortas. (H) The calcium contents in the abdominal aortas. Reprinted with the permission from Ref. 321. Copyright © 2022 Wiley. In addition to being useful for treating malignancies, carriers based on benzene boric acid esters have been extensively explored for their applications in treating a diverse array of conditions, including cardiovascular diseases321 and ischemic stroke322. Lin's group321 proposed an innovative multibioactive micelle strategy for targeted abdominal aortic aneurysm (AAA) treatment as a typical paradigm. They synthesized an amphiphilic conjugate, TPT, featuring a hydrophilic PEG segment, the SOD-mimetic agent tempol (TP), and the ROS-responsive 4-hydroxyphenylboronic acid pinacol ester (PBE) group (Fig. 10E). TPT self-assembled with rapamycin (RAP) into spherical nanomicelles (RTPTNs), averaging 104 nm in diameter with a negatively charged zeta potential. Cryosections revealed that Cy5-TPTN colocalized with key AAA-associated cell types, including CD177+ neutrophils, CD68+ macrophages, and α-SMA+ vascular smooth muscle cells (VSMCs), indicating the multifaceted targeting ability of the designed nanomicelles (Fig. 10F). In the AAA rat model, RTPTN outperformed free RAP in preventing aortic aneurysm progression and calcification (Fig. 10G and H). This “one stone for multiple birds” approach pioneers a novel nanotherapeutic paradigm for multitarget disease treatment. Others In addition to the above-listed ROS-responsive DDSs, other innovative ROS-responsive DDSs that use chemical groups, such as oxalate and oleic acid, continue to emerge (Table 1). For instance, Shen and colleagues323 synthesized a novel nanotherapeutic system that uses ROS-responsive HA-oxalate-CPT assemblies to overcome the chemotherapeutic resistance of cancer stem-like cells. Furthermore, the chemical energy harnessed from the reaction between bis[2,4,5-trichloro-6(pentyloxycarbonyl)phenyl] oxalate (CPPO) and H2O2 can directly activate the photosensitizer, circumventing the need for light excitation to produce 1O2 for PDT324. While offering considerable promise, ROS-responsive DDSs often encounter significant ROS depletion during drug release, compromising the potency of PDT. To counteract this, Wang and colleagues325 integrated a PDT and chemodynamic therapy approach to enhance tumor treatment outcomes. They synthesized an ROS-responsive polymersome, DOX-RPS, encapsulating doxorubicin hydrochloride within the hydrophobic core of an amphiphilic block copolymer composed of poly(ethylene glycol)-poly(linoleic acid) and poly(ethylene glycol)-(2-(1-hexyloxyethyl)-2-devinylpyropheophorbide-α)-iron chelate (PEG-HPPH-Fe) (Fig. 11A). The superior anticancer efficacy of DOX-RPS over its nonresponsive counterpart, DOX-NRPS, was attributed to the ROS-triggered accelerated drug release and the ROS regeneration mechanism, revealing a promising approach to overcome ROS scarcity and boost the potency of cancer therapies (Fig. 11B and C). Innovative examples of ROS-sensitive hydrogel and polymers. (A) Schematic illustration of the construction of DOX-RPS. (B) Comparison of TMB oxidation by laser pre-irradiated RPS. (C) Relative viability of different samples treated with U87MG cell. Reprinted with the permission from Ref. 325. Copyright © 2021 Wiley. (D) and (E) Schematic illustration of aCD47/Ce6@PPG hydrogel construction and its Raman-traceable, ROS-responsive degradation. (F) Raman images of Ce6@PPG. (G) Day 8 post-surgery representative flow cytometry examination of tumor-infiltrating CD4+ and CD8+ T cells gated on CD3+ T cells. Day 8 post-surgery relative quantifications of CD8+ T cells (H) and CD4+ T cells (I) in tumors. (J) Day 8 post-surgery relative quantification of mature DCs. Reprinted with the permission from Ref. 326. Copyright © 2022 Springer Nature. In addition to nanometer-scale carriers, ROS-responsive chemical groups are frequently employed to construct stimuli-responsive hydrogels, expanding the scope of applications for ROS-sensitive materials in targeted DDSs. A case in point is a ROS-responsive hydrogel-based DDS, aCD47/Ce6@PPG, designed for the sustained codelivery of the photosensitizer Ce6 and the anti-CD47 antibody (aCD47) (Fig. 11D)326. This hydrogel is synthesized by crosslinking poly(deca-4,6-diynedioic acid) (PDDA) with pullulan. Among these natural polysaccharides, PDDA is engineered to degrade in response to ROS, generating succinic acid and potentially neutralizing excess ROS to increase the safety and efficacy of PDT (Fig. 11E). Interestingly, the degradation behavior of the aCD47/Ce6@PPG hydrogel in 4T1 tumor-bearing mice was monitored via Raman imaging and correlated with a significant increase in CD8+ T cells and immune-related cytokines, indicative of robust immune response activation (Fig. 11F‒J). These results highlight a synergistic effect between PDT and immune checkpoint blockade therapy, suggesting a promising approach to invigorate immunologically “cold” tumors. Additionally, ROS-responsive hydrogels have been explored for localized pancreatic ductal adenocarcinoma (PDAC) treatment, combining gemcitabine and STING agonists to expand the horizons of ROS-mediated cancer therapeutics327. Summary and prospects ROS are pivotal in the intricate dynamics of cellular metabolism and tumorigenesis, serving as both physiological regulators and pathological catalysts in cancer. This review has shed light on the multifaceted role of ROS, from their homeostatic functions to their disruptive role in oncogenesis and their potential as targets for therapeutic intervention, particularly in PDT. The emergence of ROS-responsive DDSs represents a giant leap toward precision cancer medicine. These systems are designed to harness the unique redox environment of tumors to selectively activate therapeutic agents, offering a paradigm of intelligent therapeutics tailored to the oxidative stress signatures of cancer cells. For example, the strategic use of thioethers and thioketals as ROS-sensitive linkers has resulted in enhanced drug release specificity within the oxidative tumor microenvironment. Similarly, incorporating selenides and tellurides has provided a tunable response to the oxidative stress found across various cancer types. However, translating ROS-responsive DDSs from the laboratory to clinical practice is a journey fraught with challenges that require innovative and multidisciplinary solutions. A pivotal challenge lies in calibrating the sensitivity threshold of these systems to ROS levels. As elaborated in Section 3, the diverse ROS-reactive groups or bonds exhibit varying degrees of responsiveness and reaction kinetics. The strategic design and customization of ROS-responsive DDSs, specifically tailored to the distinct ROS levels found in physiological environments and different tumor microenvironments, is of paramount importance. This precision in tailoring ensures that the DDSs are adept at discriminating between normal and pathological states, thereby optimizing therapeutic efficacy and minimizing adverse effects. The activation of photosensitizers within ROS-responsive DDSs is another complex challenge. Overcoming the limitations of light penetration requires the design of photosensitizers with optimal absorption profiles for deeper tissue penetration. Alternative activation methods, such as NIR light, represent promising research avenues due to their capacity to penetrate deeper into tissues. The precision of ROS production is equally important. Advanced imaging techniques, such as fluorescence lifetime imaging, can provide real-time monitoring of ROS levels, guiding the precise activation of DDSs. Moreover, the synergistic combination of ROS-responsive DDSs with other targeted therapies, like immune checkpoint inhibitors, may offer a multifaceted attack on cancer. Recent studies highlight the immunomodulatory effects of ROS in the tumor microenvironment, suggesting that ROS-responsive DDSs could be integrated into a broader immunotherapy strategy. The path forward in the development of ROS-responsive DDSs for PDT is clear yet complex. To refine these systems, concerted efforts in materials science, bioengineering, and clinical research have been made. The future of cancer therapy may well lie in our ability to harness the power of ROS, turning a cellular threat into a therapeutic. With a deeper understanding of the underlying molecular mechanisms and a commitment to innovative design, ROS-responsive DDSs have the potential to unlock new horizons in PDT, offering a new generation of cancer treatments that are more effective, targeted, and safe. As we advance, addressing the challenges of sensitivity, photosensitizer activation, and ROS production control is crucial. By doing so, we can enhance the clinical applicability of ROS-responsive DDSs, ensuring that they provide a therapeutic window that is both expansive for tumor targeting and protective for healthy tissues. The integration of ROS-responsive DDSs with PDT holds great promise for the future of cancer therapy, offering hope to patients and a beacon for researchers in the field. Author contributions Danrong Hu: Writing – review & editing, Writing – original draft, Visualization, Methodology, Conceptualization. Yicong Li: Writing – review & editing, Writing – original draft, Visualization, Methodology, Conceptualization. Ran Li: Writing – review & editing. Meng Wang: Writing – review & editing. Kai Zhou: Writing – review & editing. Chengqi He: Writing – review & editing. Quan Wei: Writing – review & editing. Zhiyong Qian: Supervision, Project administration, Funding acquisition. Zhiyong Qian: Supervision, Project administration, Funding acquisition. Conflicts of interest The authors have no conflicts of interest to declare. Acknowledgments This work was financially supported by the National Natural Science Funds (NSFC 32001003, U21A20417 and 31930067, China) and 1·3·5 project for disciplines of excellence, West China Hospital, Sichuan University (ZYGD18002, China). We also thank the support of the Natural Science Foundation of Sichuan Province (2022NSFSC1282, China). Footnotes Peer review under the responsibility of Chinese Pharmaceutical Association and Institute of Materia Medica, Chinese Academy of Medical Sciences. References