Light‐Inducible Bispecific DNAzymes Reveal Mitochondrial Fe²⁺/Fe³⁺ Dynamics During Ferroptosis

ABSTRACT

Mitochondrial iron redox homeostasis plays critical roles in cellular function and disease progression; however, spatiotemporal measurement of Fe²⁺ and Fe³⁺ dynamics in mitochondria remains challenging, largely due to limited analytical tools. Here we report a light‐inducible bispecific DNAzyme sensor (LiBD) that enables simultaneous and spatiotemporally controlled imaging of mitochondrial labile Fe²⁺ and Fe³⁺ in living cells and in vivo. LiBD integrated two orthogonal iron‐specific DNAzymes into a single DNA scaffold, incorporating photocleavable (PC) linkers, rendering DNAzymes inactive until light‐triggered activation. Coupled with mitochondria‐targeted nanocarriers, LiBD demonstrated high spatial and temporal resolution for monitoring Fe²⁺ and Fe³⁺ level changes in mitochondria. We revealed remarkable mitochondrial Fe²⁺ accumulation in tumor cells during drug‐induced ferroptosis, whereas chemotherapy‐resistant tumor cells exhibited substantially decreased Fe²⁺ in mitochondria, both accompanied by slight increases in Fe³⁺. Moreover, an NIR‐activatable LiBD demonstrated in vivo visualization of mitochondrial Fe²⁺ and Fe³⁺ level changes during ferroptosis in tumor‐bearing mice. Collectively, this work presents LiBD as a powerful platform for investigating mitochondrial iron biology with spatiotemporal precision and provides insights into iron‐dependent mechanisms of ferroptosis and chemoresistance.

A light‐inducible bispecific DNAzyme sensor (LiBD) has been developed for spatiotemporally resolved, simultaneous detection of Fe²⁺ and Fe³⁺ in mitochondria of living cells and in vivo. Using an LiBD sensor, we revealed that mitochondrial Fe²⁺ was remarkably increased in ferroptotic tumor cells, whereas chemotherapy‐resistant tumor cells exhibited a substantial decrease in Fe²⁺, both accompanied by modest increases in mitochondrial Fe³⁺.

1. Introduction

Iron is a redox‐active metal ion that plays vital roles in a wide range of cellular processes, including metabolism, bioenergetics, biosynthesis, and gene regulation [1, 2, 3]. Notably, mitochondria serves as the primary organelle for iron utilization, storage, and interconversion between the oxidized ferric (Fe³⁺) and the reduced ferrous (Fe²⁺) states in living cells [4, 5, 6]. Mitochondrial iron contributes to various metabolic processes such as oxidative phosphorylation and regulating reactive oxygen species (ROS) production, functions as a critical cofactor in enzymatic reactions, and directly modulates redox homeostasis [7, 8, 9, 10]. It has been shown that dysregulation of Fe²⁺ and Fe³⁺ levels disrupts the redox equilibrium in mitochondria, leading to excessive ROS generation and subsequent oxidative damage to nucleic acids, proteins, and lipids [11, 12, 13]. Such mitochondrial iron imbalance has been closely implicated in the pathogenesis of diverse diseases, such as metabolic disorders, neurodegenerative diseases, and tumor progression [14, 15, 16]. Therefore, there is an urgent need to develop strategies that enable spatiotemporal and simultaneous analysis of Fe²⁺ and Fe³⁺ in mitochondria, which would provide more insights into the molecular mechanism of mitochondria‐related biological processes and facilitate the elucidation of disease progression and therapeutic intervention.

Various strategies have been developed for iron detection, including inductively coupled plasma mass spectrometry (ICP‐MS) [17], atomic absorption spectrometry (AAS) [18], electrochemical sensors [19, 20, 21], and fluorescent probes [22, 23, 24]. In recent years, several small molecule fluorogenic probes have been reported for selective and sensitive imaging of either Fe²⁺ or Fe³⁺ in living cells and in vivo [25, 26, 27]. Despite these important advances, the spatiotemporal measurement of both Fe²⁺ and Fe³⁺ in subcellular organelles remains highly challenging. RNA‐cleaving DNAzymes, a class of catalytic DNA molecules that cleave substrate strands in the presence of specific metal‐ion cofactors, offer a powerful and versatile platform for metal ion sensing [28, 29, 30, 31]. Benefiting from their modular design, high selectivity and excellent sensitivity, a broad range of DNAzyme‐based fluorogenic sensors have been developed for detection of metal ions in living cells and in vivo [32, 33, 34]. Very recently, Lu and co‐workers reported two DNAzymes exhibiting high specificity towards Fe²⁺ and Fe³⁺, respectively, thereby providing a simple and robust tool for imaging of labile iron in living cells [35]. Nevertheless, the dynamic changes of mitochondrial Fe²⁺ and Fe³⁺ during biological processes such as ferroptosis and related chemotherapy‐resistance remains incompletely understood, largely due to the lack of effective tools that enable spatiotemporal and simultaneous detection of both iron redox states in mitochondria.

To investigate the iron redox dynamic at the organelle level, sensing strategies need to precisely define where and when the target irons are detected in living cells. In particular, this requires DNAzyme sensors endowed with both subcellular targeting and spatiotemporal activation capabilities that are not achievable with constitutively active DNAzymes. Recent advances in the development of chemically caged DNAzyme [36, 37, 38] and organelle‐targeted delivery strategies [39] provide key technologies toward this goal. Caged DNAzymes enable conditional control of their catalytic activities in response to specific stimuli, thereby avoiding unspecific signal leakage and improving imaging sensitivity. Meanwhile, targeted‐delivery systems typically relying on specific targeting ligands afford spatially resolved imaging within specific organelles. For example, we have recently developed a photoactivatable CRISPR‐DNAzyme sensor for selective imaging of Zn²⁺ in the nucleus [40]. Inspired by these considerations, we herein report a light‐inducible bispecific DNAzyme sensor (LiBD) that enables spatiotemporally resolved and simultaneous detection of Fe²⁺ and Fe³⁺ in mitochondria of living cells and in vivo (Scheme 1). LiBD is rationally designed by integrating two orthogonal DNAzymes, one specific for Fe²⁺ and the other for Fe³⁺, into a single DNA scaffold. Photocleavable (PC) linkers are incorporated into the connecting regions, rendering DNAzymes catalytically inactive until light‐triggered activation for iron sensing. Moreover, triphenylphosphonium (TPP)‐modified cationic polymer carriers (PEI‐TPP) are utilized for the targeted delivery of LiBD into mitochondria. Using this LiBD, we observed that in ferroptotic tumor cells, mitochondrial Fe²⁺ increased remarkably while Fe³⁺ remained steady at the early stage and then slightly increased, indicating a dominant role of Fe²⁺ accumulation in driving ferroptotic progression. We also revealed a significant elevation in mitochondrial Fe²⁺ levels in A549 tumor cells upon cisplatin treatment, whereas Fe²⁺ levels were remarkably reduced in cisplatin‐resistant A549 cells. Meanwhile, mitochondrial Fe³⁺ exhibited only modest increases in both cell lines, suggesting a differential iron redox regulation that correlates with ferroptosis sensitivity. In a mouse model bearing HeLa xenograft tumors, we further demonstrated the ability of LiBD for in vivo monitoring mitochondrial Fe²⁺ and Fe³⁺ levels during ferroptosis.

SCHEME 1.

Design and application of the LiBD sensor for mitochondrial Fe²⁺ and Fe³⁺ imaging. (a) Schematic illustration of the rational design of LiBD and the preparation of LiBD@TPP nanoparticles for mitochondrial delivery. (b) Schematic illustration of spatiotemporally controlled imaging of mitochondrial Fe²⁺ and Fe³⁺ in living cells using LiBD@TPP, and its application in revealing mitochondrial dynamics of Fe²⁺ and Fe³⁺ responses to chemotherapy in drug‐sensitive and drug‐resistant tumor cells.

2. Results and Discussion

2.1. Design of LiBD for Controllable Fe²⁺ and Fe³⁺ Detection

To engineer light‐inducible DNAzymes, we designed PC linker‐incorporated inhibitory strands to disrupt the catalytic conformation of DNAzymes. To allow orthogonal optical readouts, two spectrally distinct fluorophores, Alexa Fluor 488 and Cy5 (Figure S1), were conjugated to the Fe^{2 +}‐ and Fe^{3 +}‐specific DNAzymes, respectively, yielding fluorogenic DNAzyme sensors termed as cDz‐Fe^{2 +} and cDz‐Fe^{3 +}. In the inactive state, the fluorescence was effectively quenched by proximal black hole quenchers (BHQ1) via Förster resonance energy transfer. Upon irradiation with 365 nm light (20 mW/cm²), photolysis of the PC linker triggered a conformational switch that enabled the correct folding of DNAzymes to restore their catalytic activities. This photoactivation facilitated the target‐dependent cleavage of the substrate strands, thereby generating remarkable fluorescence increase in the presence of target iron (Figure 1a–c). These results were further confirmed by native polyacrylamide gel electrophoresis (Figures S2 and S3). Kinetic analysis of photoactivation revealed a time‐dependent increase in fluorescence intensity, reaching a plateau at ∼8 min post‐irradiation. Thus, an irradiation time of 8 min was used throughout subsequent experiments (Figures S4 and S5).

FIGURE 1.

Design and characterization of LiBD for Fe²⁺ and Fe³⁺ detection. (a) Schematic illustration of the light‐inducible DNAzymes cDz‐Fe²⁺ and cDz‐Fe³⁺. (b) Schematic illustration of the LiBD sensor integrating two light‐inducible DNAzymes for dual‐channel fluorescence readout. (c, d) Fluorescence spectra of (c) cDz‐Fe²⁺ (100 nM) and (d) cDz‐Fe³⁺ (100 nM), respectively, in the absence or presence of light irradiation upon incubation with Fe²⁺ (250 µM) and Fe³⁺ (25 µM). (e, f) Fluorescence intensity of light‐activated LiBD monitored at (e) 520 nm and (f) 667 nm in response to Fe²⁺ (250 µM) and Fe³⁺ (25 µM). (g, h) Fluorescence spectra of LiBD after light activation (8 min) in the presence of increasing concentrations of (g) Fe²⁺ and (h) Fe³⁺. Insets show the linear relationship between fluorescence intensity at 520 nm and Fe²⁺ concentrations (0–250 µM), and at 667 nm and Fe³⁺ concentrations (0–25 µM). (i, j) Selectivity of LiBD in the presence of different metal ions. The symbols “+” and “‐” represent the presence and absence of the indicated components, respectively. Error bars represent standard deviations from three independent biological replicates (n = 3).

To achieve precise stoichiometric control of two orthogonal DNAzymes for accurate and simultaneous monitoring of Fe²⁺ and Fe³⁺, which is particularly critical in live‐cell studies where delivery carriers stochastically encapsulate DNAzyme cargos, we further engineered the LiBD sensor by assembling the two DNAzymes onto a single DNA scaffold with a defined 1:1 architecture (Figures 1d and S6). We next examined the fluorescence response of LiBD. LiBD exhibited low background fluorescence, indicating effective caging of both DNAzymes. Upon light irradiation, LiBD showed a remarkable increase in Alexa Fluor 488 fluorescence selectively in the presence of Fe²⁺, whereas a strong Cy5 fluorescence enhancement was observed in response to Fe³⁺. Importantly, when both Fe²⁺ and Fe³⁺ were present, varying fluorescence responses were observed in the two channels (Figures 1e,f and S7). By contrast, a negative control lacking PC linkers (termed as nLiBD) exhibited negligible fluorescence responses either upon light irradiation or in the presence of target irons (Figure S8). These results demonstrated LiBD as a controllable and orthogonal platform for simultaneous detection of Fe²⁺ and Fe³⁺. Moreover, LiBD exhibited concentration‐dependent increases in Alexa Fluor 488 and Cy5 fluorescence signals with increasing Fe²⁺ and Fe³⁺ concentrations, respectively (Figure 1g,h). The limits of detection (LODs) were determined to be 15.04 µM for Fe²⁺ and 0.053 µM for Fe³⁺. Given that intracellular concentrations of the labile Fe²⁺ and Fe³⁺ are typically within the range of 50–100 µM [41], the sensitivity of LiBD is suitable for monitoring endogenous iron redox states in living cells. In addition, LiBD exhibited excellent selectivity towards Fe²⁺ and Fe³⁺, with minimal fluorescence signal observed for other metal ions, and also effectively discriminated between the two redox states (Figures 1i,j). Collectively, these results demonstrated the capability of LiBD for controllable, selective, and simultaneous detection of Fe²⁺ and Fe³⁺.

2.2. Spatiotemporal Imaging of Fe²⁺ and Fe³⁺ Dynamics in Mitochondria

To achieve efficient mitochondrial delivery of LiBD, TPP, a widely used mitochondria‐targeting ligand [42], was conjugated to polyethyleneimine (termed PEI‐TPP) (Figure S9). Encapsulation of LiBD with PEI‐TPP yielded spherical nanoparticles (termed LiBD@TPP) with positively charged surfaces (Figure S10). We then evaluated the mitochondrial targeting ability of PEI‐TPP using Cy5‐labeled Y‐shaped DNA structures (Y_Cy5). Confocal fluorescence imaging revealed that HeLa cells incubated with Y_Cy5@TPP for 4 h displayed strong Cy5 fluorescence that overlapped extensively with the mitochondrial marker MitoTracker Orange. Quantitative colocalization analysis yielded a Pearson's correlation coefficient of 0.95, indicating efficient mitochondrial targeting (Figure 2b, bottom panel). In contrast, Y_Cy5 delivered using PEI lacking TPP modification exhibited substantially reduced colocalization with MitoTracker Orange with a Pearson's correlation coefficient of 0.42. Notably, a similar mitochondrial colocalization pattern was also observed for LiBD delivered by PEI‐TPP (Figure S11). These results highlighted the essential role of TPP in mediating mitochondrial delivery. Together, these results demonstrated that PEI‐TPP enabled effective delivery of LiBD to mitochondria in living cells.

FIGURE 2.

Spatiotemporal imaging of mitochondrial Fe²⁺ and Fe³⁺ using LiBD@TPP in living cells. (a) Schematic illustration of LiBD@TPP for mitochondrial Fe²⁺ and Fe³⁺ imaging in living cells. (b) Confocal fluorescence imaging of HeLa cells transfected with Y_Cy5@PEI or LiBD@TPP. Line‐scan intensity profiles of Y_Cy5 (red) and Mito‐Tracker Orange (orange) along the white arrow are shown to evaluate mitochondrial colocalization. (c) Confocal fluorescence imaging of HeLa cells transfected with LiBD@TPP, followed UV light activation or no irradiation. Cells were untreated, treated with Fe³⁺ (50 or 100 µM) or holo‐Tf (50 µM) to elevate intracellular iron levels, or treated with DFO (100 µM) to reduce intracellular iron level. Scale bar: 10 µm. Quantification of (d) Alexa Fluor 488 and (e) Cy5 fluorescence intensities from flow cytometric analysis (Figure S16). (f) The intracellular Fe^2+/Fe³⁺ ratio calculated based on the mean fluorescence intensity of (d) Fe²⁺ and (e) Fe³⁺. Error bars represent standard deviations from three independent biological replicates (n = 3). Statistical significance was calculated using a two‐tailed unpaired Student's t‐test, with p values indicated as: ns, no significance, *p  ≤  0.05, **p  ≤  0.01, and ***p  ≤  0.001.

Next, we investigated the capability of LiBD for spatiotemporal and simultaneous imaging of mitochondrial Fe²⁺ and Fe³⁺ in living cells. Prior to photoirradiation, HeLa cells transfected with LiBD@TPP exhibited minimal Alexa Fluor 488 and Cy5 fluorescence, indicating effective inhibition of DNAzyme activity in the caged state. Upon irradiation for 8 min, bright fluorescence signals were observed in both channels and colocalized well with mitochondria, confirming light‐triggered activation of LiBD for spatially and temporally controlled imaging of mitochondrial Fe²⁺ and Fe³⁺ (Figure 2c). Moreover, JC‐1 assays confirmed that the applied UV irradiation did not significantly affect mitochondrial integrity (Figure S12). In contrast, HeLa cells transfected with the nLiBD@TPP displayed negligible fluorescence even after irradiation, highlighting the essential role of PC linkers for sensor activation (Figure S13). Similar results were also observed in other two cell lines of HEK293T and MDA‐MB‐231 (Figure S14). Together, these data demonstrated the light‐triggered activation of LiBD for spatiotemporal imaging of Fe²⁺ and Fe³⁺ in living cells.

We then investigated whether LiBD could monitor the dynamic changes of mitochondrial Fe²⁺ and Fe³⁺ in living cells. Treatment of HeLa cells with 50 µM Fe³⁺, 100 µM Fe³⁺, or 50 µM Holo‐transferrin (holo‐Tf) resulted in remarkable increases in Alexa Fluor 488 fluorescence, indicating substantial accumulation of mitochondrial Fe²⁺. These observations were validated using Mito‐FerroGreen, a commercially available mitochondria‐specific Fe²⁺ probe (Figure S15). In contrast, only slight increases in Cy5 fluorescence were detected under the same conditions, suggesting relatively modest changes in mitochondrial Fe³⁺ levels. This phenomenon can be attributed to the endosomal reduction of Fe³⁺ to Fe²⁺ during transferrin receptor‐mediated endosomal uptake, followed by controlled partitioning of Fe²⁺ among intracellular iron‐utilizing pathways and subsequent mitoferrin‐mediated import into mitochondria [43, 44]. We also observed that HeLa cells treated with the iron chelator deferoxamine (DFO) resulted in significant decreases in both fluorescence signals (Figure 2c). These results were further validated in a large cell population using flow cytometric analysis (Figures 2d,e and S16). Compared to untreated cells, HeLa cells exposed to 50 µM Fe^{3 +}, 100 µM Fe^{3 +}, and 50 µM holo‐Tf exhibited 2.0‐, 2.9‐, and 2.8‐fold increases in Alexa Fluor 488 fluorescence, respectively, whereas the Cy5 fluorescence increased very slightly (up to ∼1.3‐fold). In addition, to quantify the relative iron redox changes in mitochondria, we calculated the Fe²⁺/ Fe³⁺ fluorescence ratio (Figure 2f). The result showed that the ratio increased markedly upon treatment with Fe³⁺ or holo‐Tf, indicating preferential accumulation of Fe²⁺ within the mitochondrial iron pool. Notably, DFO treatment also led to an elevated Fe²⁺/Fe³⁺ ratio, consistent with its reported preference for chelating Fe³⁺ over Fe²⁺ [45]. Collectively, these results demonstrated that the developed LiBD‐enabled spatiotemporal and simultaneous monitoring of mitochondrial Fe²⁺ and Fe³⁺ levels in living cells.

2.3. Evaluation of Mitochondrial Iron Redox Dynamics During Ferroptosis

Ferroptosis is a highly regulated form of cell death driven by iron‐dependent lipid peroxidation, and monitoring the iron redox levels in ferroptotic cells, especially in mitochondria where the irons are mainly stored and used, is critical for further understanding the molecular mechanism of ferroptosis [46, 47, 48]. However, the spatiotemporal measurement of mitochondrial Fe²⁺ and Fe³⁺ dynamics remain challenging, largely due to the lack of direct and selective analytical tools. We then applied the LiBD sensor to investigate mitochondrial iron redox states and their conversion during ferroptosis. HeLa cells treated with RAS‐selective lethal 3 (RSL3), a widely used ferroptosis inducer, was served as the model system. RSL3 does not directly alter total cellular iron levels but instead enhances the reactivity of existing iron pools by inhibiting glutathione peroxidase 4 (GPX4)‐mediated lipid peroxide detoxification [49]. Compared to HeLa cells without RSL3 treatment, cells treated with RSL3 (1 µM) induced remarkably elevated Fe²⁺ levels upon prolonging the incubation time, reaching up to 2.4‐fold increase in mitochondrial Fe²⁺ levels (Figures 3a,b and S17). By contrast, we observed much lower increase in Fe³⁺ levels under same conditions, with 1.2‐fold increase observed at 4 h (Figures 3a,c and S18). As a result, a continuous increase of the Fe²⁺/Fe³⁺ ratio was obtained during ferroptosis (Figure 3d), indicating a dominant role of Fe²⁺ accumulation during ferroptosis. In addition, co‐treating HeLa cells with RSL3 and iron chelators DFO for 4 h resulted in substantially decreased Fe²⁺ and Fe³⁺ levels compared to cells treated with RSL3 alone, confirming the RSL3‐induced iron changes in mitochondria. Another control experiment without light‐triggered activation of LiBD exhibited a negligible fluorescence increase in RSL3‐treated HeLa cells, confirming that the observed signals were caused by light‐triggered activation of DNAzyme for substrate cleavage (Figure S19). To verify the results obtained using LiBD, we then analyzed the mitochondrial Fe²⁺ and Fe³⁺ using a commercially available iron assay kit by an indirect manner [50]. The result showed that compared to untreated HeLa cells, mitochondrial Fe²⁺ was significantly increased by 2.4‐fold upon treating HeLa cells with RSL3 for 4 h, whereas Fe³⁺ exhibited a modest increase of 1.4‐fold. As a result, the total iron levels in mitochondria increased by 1.9‐fold (Figure 3e). These results together demonstrated the capacity of LiBD for spatiotemporally controlled monitoring of mitochondrial iron redox changes, providing deep insights into the effect of Fe²⁺ and Fe³⁺ during ferroptosis.

FIGURE 3.

Evaluation of mitochondrial Fe²⁺ and Fe³⁺ dynamics during RSL3‐induced ferroptosis. (a) Confocal fluorescence imaging of HeLa cells transfected with LiBD@TPP, followed by UV light activation or no irradiation. Cells were either untreated, treated with 1 µM RSL3 for the indicated times, or co‐treated with 1 µM RSL3 and DFO (100 µM) for 4 h. Scale bar: 10 µm. (b, c) Quantification of (b) Alexa Fluor 488 and (c) Cy5 fluorescence signals from the flow cytometric analysis (Figure S18). (d) Calculated mitochondrial Fe²⁺/Fe³⁺ ratio based on the mean fluorescence intensities shown in (b) and (c). (e) Quantification of mitochondrial Fe²⁺, Fe³⁺, and total iron levels in untreated cells and cells treated with 1 µM RSL3 for 4 h, determined using a ferrous/ferric iron assay kit following mitochondrial isolation. (f, g) WB analysis of (f) HO‐1 and (g) MFRN1 and FTMT levels in HeLa cells treated with 1 µM RSL3 for indicated time. The uncropped blot images are provided in Figure S32. (h–j) Quantification of HO‐1/β‐actin (h), MFRN1/β‐actin (i), and FTMT/β‐actin (j) corresponding to the WB results in (f) and (g). (k) Schematic illustration of possible pathways for mitochondrial Fe²⁺ and Fe³⁺ level changes upon RSL3 treatment. Error bars represent standard deviations from three independent biological replicates (n = 3). Statistical significance was calculated using a two‐tailed unpaired Student's t‐test, with p values indicated as: ns, no significance, *p  ≤  0.05, **p  ≤  0.01, and ***p  ≤  0.001.

Next, the possible pathways for the changes of mitochondrial iron redox levels were investigated. First, we found that the levels of intracellular ROS and heme oxygenase‐1 (HO‐1) proteins were increased in RSL3‐treated HeLa cells. Using a commercially available ROS detection probe DCFH‐DA, we detected remarkably elevated ROS levels in HeLa cells following the treatment of RSL3 (Figure S20), consistent with previous reports of GPX4 inhibition‐induced ROS generation [51]. As a result, excessive ROS can directly destabilize the Fe‐S clusters in mitochondrial, thereby leading to the in situ release of Fe²⁺ [52, 53]. Moreover, western blot (WB) analysis revealed that compared to untreated HeLa cells, the levels of HO‐1 protein were substantially increased by 2.9‐fold in cells treating with RSL3 for 4 h (Figure 3f,h). Lipid peroxidation‐induced oxidative stress is known to activate nuclear factor erythroid 2‐related factor 2 (NRF2) signaling, thereby resulting in increased expression of HO‐1 [54]. Notably, a fraction of HO‐1 has been reported to translocate to mitochondria under oxidative stress conditions, where it catalyzes heme degradation and releases Fe²⁺ [55]. Therefore, the destruction of these Fe²⁺‐containing species likely constitutes an important source for the elevated free Fe²⁺ in mitochondria during ferroptosis (Figure 3k).

Then, we assessed the contribution of cytosolic Fe²⁺ transport into mitochondria during ferroptosis. WB result revealed that the expression of mitoferrin‐1 protein (MFRN1), a mitochondrial membrane iron importer [56], was remarkably upregulated in HeLa cells upon treating RSL3 for 4 h in contrast to those of untreated cells (Figure 3g,i). Given that MFRN1 preferentially mediates the transport of Fe²⁺ into mitochondria [57], its elevation is expected to enhance the influx of cytosolic Fe²⁺ into the mitochondria, thereby serving as an additional source for accumulation of mitochondria Fe²⁺ during ferroptosis (Figure 3k).

On the other hand, the possible conversion between mitochondrial Fe²⁺ and Fe³⁺ during ferroptosis was investigated. Notably, the expression levels of mitochondrial ferritin (FTMT), a key regulator of mitochondrial Fe²⁺ oxidation and Fe³⁺ storage, was remarkably increased in HeLa cells upon RSL3 treatment (Figure 3g,j). Upregulation of FTMT is known to promote the oxidation of excess Fe²⁺ to Fe³⁺ in mitochondria, and its subsequent sequestration within ferritin [58]. Consistent with this role, a modest increase in mitochondrial Fe³⁺ was detected in cells following RSL3 treatment. While FTMT‐mediated oxidation and sequestration would be expected to partially counteract Fe²⁺ accumulation, our results suggested that the rapid release of Fe²⁺ from Fe²⁺‐containing species and the enhanced MFRN1‐dependent import of Fe²⁺ might play dominant roles in mitochondrial Fe²⁺ accumulation during ferroptosis.

We further investigated mitochondrial iron redox dynamics during ferroptosis in different cell types. Increasing evidence indicates that tumor cells can be induced to undergo ferroptosis by platinum‐based chemotherapeutics [59]. However, this ferroptotic effect is markedly attenuated in some platinum‐resistant cancer cells, while the molecular mechanism remains incompletely understood. To elucidate whether altered mitochondrial iron homeostasis contributes to differential ferroptosis sensitivity, we applied the LiBD sensor to directly evaluate mitochondrial Fe²⁺ and Fe³⁺ levels in platinum‐sensitive and platinum‐resistant tumor cells under drug treatment. Cis‐diamminedichloroplatinum(II) (DDP), a widely used chemotherapeutic drugs [60], was used as the ferroptosis‐inducing agent, A549 cells and cisplatin‐resistant A549/DDP cells were selected as model systems. Cell viability assay confirmed the distinct cisplatin sensitivities of the two cell lines, with a half‐maximal inhibitory concentration (IC₅₀) of DDP of 17.43 µM for A549 cells and 95 µM for A549/DDP cells, corresponding to a drug resistance index (RI) of 5.45 (Figure S21). These results validated A549/DDP cells as a suitable model for dissecting mitochondrial iron alterations associated with ferroptosis resistance.

Next, we applied LiBD sensor to evaluate mitochondrial Fe²⁺ and Fe³⁺ changes in A549 and A549/DDP cells. We observed that treating A549 cells with DDP for 4 h induced a 1.3‐fold increase in mitochondrial Fe²⁺ levels, whereas Fe³⁺ levels exhibited a marginal elevation (1.1‐fold) (Figures 4a,c and S22). Consequently, a significant increase in the Fe²⁺/Fe³⁺ ratio was obtained (Figure 4c), consistent with the Fe²⁺ and Fe³⁺ level changes observed in RSL3‐induced ferroptotic HeLa cells. By contrast, DDP treatment of A549/DDP cells resulted in a substantial reduction in Fe²⁺ levels (2.1‐fold), accompanied by a slight increase in the Fe³⁺ levels (1.2‐fold) (Figures 4b,d and S22). As a result, the Fe²⁺/Fe³⁺ ratio was significantly decreased in A549/DDP cells upon DDP exposure (Figure 4d), revealing an opposite mitochondrial iron redox response to DDP treatment compared to that of A549 cells.

FIGURE 4.

Mitochondrial iron redox responses to cisplatin in cisplatin‐sensitive and ‐resistant A549 cells. (a, b) Confocal fluorescence imaging of (a) A549 cells and (b) A549/DDP cells transfected with LiBD@TPP, with or without UV light activation. Cells were either untreated or treated with DDP (30 µM) for 4 h prior to imaging. Scale bar: 10 µm. (c, d) Quantification of Alexa Fluor 488 (Fe²⁺) and Cy5 (Fe³⁺) fluorescence signals from the flow cytometric analysis of (c) A549 cells and (d) A549/DDP cells (Figure S22). (e–g) Quantification of (e) HO‐1 levels, (f) MFRN1 and (g) FTMT levels in A549 cells and A549/DDP cells treated with 30 µM DDP for 4 h from WB analysis (Figure S24 and S25). (h) Schematic illustration of possible pathways for mitochondrial Fe²⁺ and Fe³⁺ level changes upon DDP treatment in cisplatin‐sensitive and ‐resistant A549 cells. The symbols “+” and “‐” represent the presence and absence of the indicated components, respectively. Error bars represent standard deviations from three independent biological replicates (n = 3). Statistical significance was calculated using a two‐tailed unpaired Student's t‐test, with p values indicated as ns, no significance, *p  ≤  0.05, **p  ≤  0.01, and ***p  ≤  0.001.

The possible pathway for this differential mitochondrial iron redox changes was investigated. Similar to observations in RSL3‐treated HeLa cells, DDP‐treated A549 cells exhibited substantially elevated ROS, HO‐1 and MFRN1 levels (Figures 4e,f and S23–S25), suggesting the enhanced release of Fe²⁺ from Fe²⁺‐containing species and increased mitochondrial influx of cytosolic Fe²⁺. Concurrently, elevated FTMT expression was also observed upon DDP treatment (Figures 4g and S25), likely promoting the oxidation of excess Fe²⁺ and subsequent sequestration of Fe³⁺, thereby leading to a modest increase in free Fe³⁺ levels. By contrast, in cisplatin‐resistant A549/DDP cells, DDP treatment only upregulated FTMT expressions, without inducing the increase of ROS, HO‐1 and MFRN1 levels (Figure 4e–g). Given that FTMT can promote Fe²⁺ oxidation and subsequent Fe³⁺ sequestration in mitochondria, this probably be attributed to the significantly decreased Fe²⁺ levels with slightly increased Fe³⁺ levels in mitochondria. Besides, pathways related to mitochondrial iron efflux (e.g., involving ABCB7, ABCB8 or FLVCR1b) may also contribute to the reduction of mitochondrial Fe²⁺ during this process [61, 62, 63]. This differential in mitochondrial Fe²⁺ and Fe³⁺ levels may represent an important determinant of ferroptosis sensitivity versus resistance in tumor cells. Collectively, these findings highlighted LiBD as a powerful tool for directly evaluating the chemotherapy response and resistance at the organelle level.

2.4. In Vivo Measurement of Fe²⁺ and Fe³⁺ Levels During Ferroptosis Using NIR‐LiBD

Encouraged by the spatiotemporal monitoring mitochondrial iron redox dynamics in living cells, we next extended the LiBD sensor to in vivo applications. Because UV light suffers from poor tissue penetration, and is therefore incompatible with in vivo activation, we then integrated LiBD with upconversion nanoparticles (UCNPs), which convert near‐infrared (NIR) excitation into UV emission [64], to construct a NIR‐activatable LiBD sensor (termed as NIR‐LiBD).

The mitochondria‐targeting NIR‐LiBD was constructed via layer‐by‐layer assembly of PEI, LiBD, and PEI ‐ TPP onto the surface of UCNPs (Figure 5a), yielding nanocomposites with an average hydrodynamic diameter of ∼200 nm and a positively charged surface (Figures S26 and S27). We first evaluated the capacity of NIR‐LiBD for spatiotemporally controlled imaging of mitochondrial Fe²⁺ and Fe³⁺ in living cells (Figure S28a). Upon NIR irradiation (980 nm, 10 mW/cm²) for 10 min, HeLa cells treated with NIR‐LiBD displayed remarkable increases in both Alexa Fluor 488 and Cy5 fluorescence signals compared with non‐pirradiated controls, with both signals colocalizing well with mitochondrial markers (Figure S28b, left panel). In contrast, cells treated with a control design lacking PC linkers (NIR‐nLiBD) showed negligible fluorescence responses even after NIR irradiation (Figure S28b, right panel). These results demonstrated NIR‐triggered activation of LiBD for mitochondrial Fe²⁺ and Fe³⁺ imaging.

FIGURE 5.

In vivo imaging of mitochondrial Fe²⁺ and Fe³⁺ dynamics during ferroptosis. (a) Schematic illustration of the synthesis of NIR‐LiBD from UCNPs and its application for NIR‐activated in vivo imaging of Fe²⁺ and Fe³⁺. (b, e) Representative whole‐body fluorescence imaging of HeLa tumor‐bearing mice in a bilateral tumor model following intratumoral injection of (b) NIR‐nLiBD or (e) NIR‐LiBD, with or without NIR irradiation. Tumor regions are highlighted by red circles. (c, d, f, g) Quantification of fluorescence intensities at the left (‐NIR) and right (+NIR) tumor sites in mice treated with (c, d) NIR‐nLiBD and (f, g) NIR‐LiBD at 0 and 4 h post‐injection. (h) Schematic overview of the experimental protocol for monitoring mitochondrial Fe²⁺ and Fe³⁺ dynamics during RSL3‐induced ferroptosis in HeLa tumor‐bearing mice. (i–k) Representative whole‐body fluorescence images of mice bearing HeLa tumors after the indicated treatments. (l, m) Quantification of (l) Cy7 and (m) Cy5 fluorescence signals from in vivo imaging at 0 and 4 h post‐treatment in (i–k). (n) Calculated mitochondrial Fe²⁺/Fe³⁺ ratio based on the mean fluorescence intensities shown in (l) and (m). Error bars represent standard deviations from three independent biological replicates (n = 3). Statistical significance was calculated using a two‐tailed unpaired Student's t‐test, with p values indicated as ns, no significance, *p  ≤  0.05, **p  ≤  0.01, and ***p  ≤  0.001.

We next evaluated the in vivo performance of NIR‐LiBD using BALB/c nude mice bearing HeLa xenograft tumors. To enable orthogonal fluorescence readouts in vivo, Alexa Fluor 488 in the Fe²⁺‐specific DNAzyme was further replaced with a NIR fluorophore Cy7 (Figure S29). Following intratumoral (i.t.) administration of NIR‐nLiBD (10 ng mg⁻¹), minimal Cy5 and Cy7 fluorescence signals were detected in tumors regardless of NIR irradiation (Figure 5b–d). By contrast, i.t. injection of NIR‐LiBD at the same dose, followed by NIR irradiation, resulted in remarkable increases in both Cy7 (1.8‐fold) and Cy5 (2.5‐fold) fluorescence within 4 h, whereas tumors without irradiation exhibited negligible signal changes (Figure 5e–g). Ex vivo analysis of dissected tumors further validated these results (Figure S30). These results demonstrated the ability of NIR‐LiBD for spatiotemporally controlled imaging of endogenous cellular Fe²⁺ and Fe³⁺ in vivo.

Finally, we applied NIR‐LiBD to monitor mitochondrial iron redox dynamics during ferroptosis in vivo (Figure 5h). Compared with mice co‐receiving NIR‐LiBD and NIR irradiation (Figure 5i), mice pretreated with RSL3 via intraperitoneal injection (10 ng mg⁻¹, 24 h) followed by co‐treatment of NIR‐LiBD and NIR irradiation exhibited an obvious increase in Cy7 fluorescence (∼2.2‐fold), accompanied by a modest elevation in Cy5 fluorescence (1.2‐fold) (Figure 5j,l,m). As a result, a 1.9‐fold increase in Fe²⁺/Fe³⁺ ratio was observed in RSL3‐treated mice compared to that of the untreated mice (Figure 5n). These results suggested the concurrent increases in mitochondrial Fe²⁺ and Fe³⁺ levels during ferroptosis, with Fe²⁺ displaying more pronounced changes. Notably, sequential treatment with RSL3 followed by DFO remarkably suppressed both Cy5 and Cy7 fluorescence signals (Figure 5k–m), confirming the iron‐dependent generation of the fluorescence responses. Ex vivo analysis of dissected tumors further validated these results (Figure S31), highlighting NIR‐LiBD as a powerful tool for spatiotemporal and simultaneous imaging of mitochondrial Fe²⁺ and Fe³⁺ dynamics during ferroptosis in living organisms.

3. Conclusion

In this study, we developed a novel light‐inducible bispecific DNAzyme sensor that enables spatiotemporally resolved and simultaneous detection of Fe²⁺ and Fe³⁺ in mitochondria of living cells and in vivo. By integrating two iron‐specific DNAzymes into a single DNA scaffold via PC linkers‐contained binding arms and further coupling with mitochondria‐targeted nanocarriers, LiBD overcomes challenges of conventional DNAzymes associated with uncontrolled catalytic activity and subcellular‐level specificity. LiBD revealed distinct mitochondrial iron redox dynamics in biological processes such as ferroptosis and chemoresistance. Our results exhibited a substantial increase in mitochondrial Fe²⁺ in ferroptotic tumor cells, probably due to the rapid release of Fe²⁺ from Fe²⁺‐contained biomolecules and the influx of cytosolic Fe²⁺ into mitochondria. Conversely, cisplatin‐resistant tumor cells exhibited upregulated FTMT‐mediated Fe²⁺ oxidation and Fe³⁺ sequestration, thereby resulting in significantly decreased Fe²⁺ levels in mitochondria. Additionally, the free Fe³⁺ levels in mitochondria were slightly increased in these processes, possibly due to the upregulated FTMT levels. These findings indicate the critical role of mitochondrial iron redox for ferroptosis susceptibility and chemotherapeutic response.

The LiBD sensor offers distinct advantages in its simple, modular design and easy incorporation of light‐activatable elements for spatiotemporally controlled Fe²⁺ and Fe³⁺ imaging. Moreover, by leveraging a mitochondrial‐targeting delivery system, LiBD enables the monitoring of iron dynamics at the subcellular level. However, due to its DNAzyme‐mediated cleavage mechanism, LiBD primarily detects cumulative rather than real‐time, reversible changes of labile iron species. Collectively, LiBD expands the toolbox for iron detection by enabling spatiotemporally controlled and organelle‐resolved imaging of iron redox dynamics. We anticipate that this platform will facilitate a deeper understanding of iron‐dependent biological processes and inspire the development of next‐generation molecular tools for precision bioimaging and therapeutic evaluation.

Author Contributions

Yang Shi: methodology, investigation, supervision, data curation, validation. Rong Wang: validation, investigation. Jiaqi Wang: validation, formal analysis. Hang Xing: resources. Jingjing Zhang: funding acquisition, resources, project administration, conceptualization. Jian‐Hui Jiang: resources, writing – review and editing.Zhenkun Wu: funding acquisition, writing – original draft, writing – review and editing, project administration, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information [65, 66, 67, 68, 69].Supporting File 1: anie72814‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.