A tandem activity-based sensing and labeling strategy reveals antioxidant response element regulation of labile iron pools

To develop an ABS iron probe that can preserve intracellular information on bioavailable Fe(II) pools by proximal protein labeling, we coupled the well-validated Fe(II)-responsive TRX trigger for ABS of Fe(II) (33–36, 49) to a modified 3-O-methylfluorescein dye reporter containing a pendant fluoromethyl group ortho to the phenolic site of TRX conjugation. We reasoned that Fe(II)-mediated release of a highly electrophilic quinone methide species would allow for proximal covalent reaction of the dye with biological nucleophiles (Scheme 1). While previous reports for hydrogen peroxide or enzyme activity detection appended a reactive trigger directly onto the fluorophore phenol (50–53), direct conjugation of a TRX moiety onto this site would yield an unstable carbonate prone to hydrolysis within biological settings. To overcome this issue, we employed a (N,N’-dimethyl)ethylenediamine linker separating the TRX trigger and dye (15, 37), used previously to cage and release phenolic reporters (37, 54, 55). Upon reaction with Fe(II), the TRX endoperoxide is reduced to a cyclohexanone intermediate, which undergoes spontaneous β-elimination (33) to release an extended amine linker that then undergoes a self-immolative cyclization to uncage the free phenol of the fluorescein payload. Subsequent fluoride elimination generates a highly reactive ortho-quinone methide electrophile that is susceptible to attack by proximal biological nucleophiles, resulting in a covalently anchored fluorescent product near the site of activity-based Fe(II) detection (Scheme 1 and SI Appendix, Scheme S1). We hypothesized that such a tandem sensing and labeling strategy would enable turn-on Fe(II) detection as the dye is released upon reaction with Fe(II) and thus iron-dependent quenching is minimized. The dye product is also rendered membrane-impermeable, and thus the intracellular fluorescence signal is retained through covalent anchoring to the cell.

We developed a robust synthesis of IG1-FM in six sequential steps from the known (15, 37) enantiopure TRX intermediate 1, storing and handling the material as the HCl salt to prevent premature self-immolation of the extended linker moiety (Scheme 2). We found that activation of the terminal amine 1 for coupling was best achieved via inverse addition of triethylamine to a reaction mixture containing a large excess of phosgene (~33 equiv.), thereby favoring rapid carbamoylation of the neutralized amine to form 2 and minimizing the competing self-immolation reaction. These optimized conditions enabled shorter reaction times and afforded the carbamoyl halide 2 in high yields of up to 98%. We then coupled intermediate 2 with phenolic 3-formyl fluorescein S1, forming a stable dicarbamate linkage between the TRX sensor and the fluorescein dye. Careful reduction of the aldehyde using sodium borohydride at low temperatures ranging from −78 °C to 0 °C afforded alcohol 4, while preserving the endoperoxide bond of the TRX ring. Finally, conversion of the benzylic alcohol function to fluoromethyl using Deoxo-Fluor proceeded smoothly to afford IG1-FM. Synthesis of IG1, a control probe lacking the fluoromethyl group, was accomplished in four sequential steps. Hence, the phenolic function of fluorescein (5) was converted to the p-nitrophenylcarbonate intermediate 6, which was then reacted in situ with N-Boc N,N′-dimethylethylenediamine to afford 7 in 45% yield over two steps. Boc deprotection of 7, followed by coupling of amine 8 with the enantiopure (R, R) TRX intermediate S2 (56) afforded IG1. Lacking the latent ortho-quinone methide functionality, this probe served as an important control for evaluating the predicted trapping abilities of IG1-FM.

We began by determining the photophysical properties of IG1 and IG1-FM, and their representative reaction products 3-O-Methylfluorescein and Fl-CH₂OH (SI Appendix, Table S1). As expected, both IG1 and IG1-FM have relatively low brightness versus their reaction products, suggesting that both sensors are capable of a robust fluorescence turn-on response in a buffered solution upon cleavage of the TRX group. To validate the reactivity of IG1-FM toward Fe(II), we evaluated its fluorescence turn-on responses in aqueous solution buffered to physiological pH with 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES). Upon incubation with 10 μM Fe(II), the probe underwent a time-dependent increase in fluorescence (Fig. 1A), a feature consistent with other probes that employ quinone-methide chemistry (50–53). Further validation of this product, using liquid chromatography analysis, revealed that it was hydrolyzed to free hydroxymethylfluorescein (Fl-CH₂OH, SI Appendix, Figs. S1 and S2). Treatment of IG1-FM with various concentrations of Fe(II) showed a dose-dependent turn-on response for added Fe(II) concentrations spanning 1 to 100 μM. Additionally, coincubation with Fe(II) and the iron-selective chelator deferoxamine (DFO) results in a substantial decrease in fluorescence versus treatment with Fe(II) alone (SI Appendix, Fig. S3). The control sensor IG1 similarly showed a dose-dependent response to Fe(II) that was quenchable with DFO (SI Appendix, Fig. S4). This turn-on response was also selective for Fe(II) over other biologically relevant metals, including ferric Fe(III) ions, the iron-bound protein myoglobin, or the biologically relevant reductant and oxidant glutathione (GSH) and H₂O₂, respectively (Fig. 1B). Moreover, this high selectivity was also retained at various incubation times (SI Appendix, Fig. S5). A small fluorescence turn-on response was observed with excess Cu(I) but not at lower, more physiologically relevant copper concentrations. To establish its iron-dependent protein labeling capabilities, we incubated IG1-FM with human serum albumin (HSA) as a model protein in the presence or absence of Fe(II); protein labeling was monitored via sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). For 1 to 10 μM of IG1-FM, the HSA-associated fluorescence intensity was elevated in the presence of Fe(II), consistent with Fe(II)-dependent protein labeling by IG1-FM (Fig. 1C and SI Appendix, Figs. S6 and S7). Analogous protein labeling with HEPG2 cell lysates showed elevations in total lane fluorescence only in the presence of Fe(II), demonstrating the high selectivity for IG1-FM for Fe(II) in whole cellular proteomes (SI Appendix, Figs. S8 and S9). We also measured cell lysate labeling efficiency in the presence of excess GSH or various nucleophilic amino acids (SI Appendix, Figs. S10 and S11), from which only cysteine had a significant effect on protein labeling. Taken together, these data establish that IG1-FM is a selective and sensitive Fe(II)-responsive fluorescent probe that can undergo protein labeling coupled to activity-based iron sensing.

Upon establishing the selectivity and reactivity of IG1-FM for Fe(II) detection, we next sought to evaluate its ability to monitor changes in labile Fe(II) pools in living cells. IG1-FM showed an increased fluorescence response in HEPG2 and U2OS cells preincubated with 100 µM ferrous ammonium sulfate (FAS) as an iron supplement, whereas with the nonfluorinated probes IG1 or the hydroxymethyl compound analog 4, we saw no fluorescence changes upon treatment with FAS (SI Appendix, Figs. S12 and S13). Moreover, we observed that IG1-FM has a more uniform staining pattern compared to IG1, and IG1-FM had a higher basal intracellular fluorescence compared to 4 in HEPG2 cells. Thus, it is likely that IG1-FM provides enhanced cellular retention that is necessary for mapping intracellular labile Fe(II) pools. To further validate this notion, we incubated HEPG2 cells with IG1-FM or the IG1 reaction product 3-O-Methylfluorescein, washed three times with Hank’s Balanced Salt Solution (HBSS), and monitored the cellular retention of the probes over the course of 4 h (SI Appendix, Figs. S14 and S15). For IG1-FM, we also treated cells with and without FAS to determine whether Fe(II)-dependent differences in signal would persist over time. Indeed, the loss in signal for IG1-FM is minimal over time, only decreasing by approximately 1.3-fold, whereas 3-O-Methylfluorescein has a high initial intracellular fluorescence but dramatically decreases in signal by approximately threefold over 4 h. The poor cellular retention of 3-O-Methylfluorescein may explain the minimal intracellular staining observed with IG1 after reaction with iron. Additionally, we found that the difference in signal for IG1-FM between cells treated with and without exogenous FAS is persistent over the 4-h time course, further demonstrating the utility of the probe in monitoring intracellular labile Fe(II) pools over long time periods. We next monitored changes in labile Fe(II) in HEPG2 cells via external exposure to exogenous iron supplement and/or iron chelator and staining with IG1-FM. To monitor changes in intracellular labile iron induced by external Fe(II) supplementation, we incubated cells with FAS alone or sequentially with the iron chelator DFO followed by FAS. Changes to basal intracellular labile Fe(II) pools were evaluated by incubation of HEPG2 cells with DFO alone or FAS with the hormone hepcidin, which binds and inhibits the iron exporter ferroportin to increase basal intracellular iron levels. As expected, treatment with DFO, or DFO and FAS, produced a significant decrease in IG1-FM fluorescence signal versus basal conditions, while treatment with FAS, or FAS and hepcidin, produced significant increases in IG1-FM fluorescence with the sequential treatment of hepcidin and FAS producing a higher fluorescence signal than treatment with FAS alone (Fig. 2 A and B). Treatment with hepcidin alone did not result in significant changes in IG1-FM fluorescence (SI Appendix, Fig. S16). Analogous chelation treatments were performed with an additional iron chelator bathophenanthroline disulfonate, which produced similar results to those obtained with DFO (SI Appendix, Fig. S17). Additionally, we treated HEPG2 cells with various concentrations of FAS and the Fe(III) supplement ferric ammonium citrate (FAC) to test whether higher levels of external iron supplementation would lead to higher levels of intracellular labile iron. While IG1-FM should be unreactive to FAC, a source of Fe(III), the ferric ions may still be imported and reduced to Fe(II) by canonical iron acquisition machinery (57). Thus, treatment with FAC serves as an additional control to rule out the possibility of extracellular activation of IG1-FM. FAS produced fluorescence increases, but total signal reached a maximum at lower metal supplement concentrations (SI Appendix, Fig. S18), while treatment with FAC resulted in IG1-FM fluorescence increases that were dose-dependent on the metal supplement concentration (SI Appendix, Fig. S19). IG1-FM exhibits minimal cellular toxicity in the presence of iron supplementation/chelation treatments, as confirmed with Hoechst 33342 nuclear stain (SI Appendix, Fig. S20). The collective data establish IG1-FM as an effective ABS and labeling probe that is capable of monitoring changes in intracellular labile Fe(II) pools upon metal supplementation and/or chelation.

The ability of IG1-FM to monitor both increases and decreases in labile Fe(II) pools upon external addition of metal supplement and/or chelator, along with requisite sensitivity to detect basal levels of intracellular labile Fe(II), led us to interrogate how these metal pools may be affected by other cellular pathways for dynamic iron storage and sequestration. In this context, ferritin is the principal protein responsible for limiting the bioavailability of iron inside cells. A single ferritin can accommodate up to 4,500 iron atoms through the conversion of ferrous ions to ferric oxide via ferroxidase activity and storage in its nano-cage-like protein structure (58). Recent evidence suggests that nuclear receptor coactivator 4 (NCOA4) is responsible for lysosomal uptake and subsequent degradation of ferritin, a process coined as ferritinophagy (59, 60). Additionally, the autophagy receptor Tax1 binding protein 1 (TAX1BP1) has been reported as a NCOA4-binding protein that is required for NCOA4-ferritin trafficking to the lysosome (Fig. 3A) (61). Previous studies have shown that a deficiency in NCOA4 increases basal ferritin levels which in turn limits the bioavailability of intracellular iron (59); however, reports of iron-selective fluorescent imaging applied to directly measure how ferritinophagy regulates labile Fe(II) remain limited (62). Furthermore, whether TAX1BP1 can also regulate intracellular labile Fe(II) levels, and if so, whether it can do so to a comparable degree to NCOA4, remains understudied. Thus, we applied IG1-FM to measure labile Fe(II) pools in HeLa cells with CRISPR-Cas9 generated (63) genetic knockout (KO) of either NCOA4 or TAX1BP1. IG1-FM imaging reveals that both NCOA4 KO and TAX1BP1 KO show lower levels of fluorescence compared to wild-type (WT) cells (Fig. 3 B and C) suggesting that the disruption of ferritinophagy leads to lower basal levels of labile iron. Interestingly, NCOA4 KO and TAX1BP1 KO show similar reductions in IG1-FM fluorescence when compared with WT cells, indicating that both are essential for the regulation of labile iron via ferritinophagy. We then treated all three cell lines sequentially with FAC to stimulate ferritin production under iron-replete conditions, followed by DFO to stimulate ferritin degradation under iron-deficient conditions, in order to assess whether ferritinophagy could release labile iron from ferritin in response to iron deficiency (Fig. 3A). While FAC/DFO treated WT HeLa cells had significantly lower IG1-FM fluorescence compared to controls, the fluorescence for FAC/DFO treated WT cells was still elevated as compared to either of the untreated KO cell lines. These data indicate that ferritinophagy can mitigate iron deficiency by mediating release of sequestered iron stores. Additionally, IG1-FM fluorescence was further reduced in the KO cells upon treatment with FAC and DFO, indicating that the chelator treatment, even in the presence of high ferritin, can promote even further reductions in iron bioavailability.

Interestingly, while we observed complementary decreases in total cellular iron in all three cell lines upon cotreatment with FAC and DFO, as measured by inductively coupled plasma mass spectrometry (ICP-MS), HeLa cells with NCOA4 or TAX1BP1 KO showed reciprocal increases in total iron versus WT HeLa cells (SI Appendix, Fig. S21). These findings suggest that a surplus of ferritin results in an overall higher cellular buffering capacity for iron, with a lower level of labile iron. In a key set of control experiments, imaging of HeLa WT cells with the IG1 control probe shows minimal intracellular staining and no significant response upon treatment with FAC and DFO, consistent with data obtained in HEPG2 and U2OS cells (SI Appendix, Fig. S22) and highlighting the need for the tandem ABS and labeling in IG1-FM to achieve requisite signal-to-noise responses. Collectively, these data provide direct evidence that NCOA4 and TAX1BP1 modulate intracellular labile Fe(II) pools through ferritinophagy.

Having established the utility of IG1-FM to image ferritinophagy-induced changes in intracellular labile Fe(II) levels, we next sought to apply this sensor to biological applications of broader impact and turned our attention to study how cellular antioxidant regulation affects labile iron pool status. In this context, NRF2 is an important transcription factor that up-regulates a variety of detoxifying gene elements, known as AREs, to combat oxidative stress, including several that are associated with iron regulation (64–66). Under basal conditions, NRF2 interacts with its native regulator Kelch-like erythroid cell-derived protein with CNC homology-associated protein 1 (KEAP1), which promotes NRF2 ubiquitination via the Cullin 3‐based E3 ligase (Fig. 4A) (67). NRF2 is subsequently degraded by the 26S proteasome and maintained at low levels in the cytoplasm (68). In many cancers, NRF2 is aberrantly activated to combat oxidative stress associated with the higher energy demand necessary for the rapid cell division (69–72), while dysregulation of iron homeostasis is also a prominent marker of cancer (73) due to the high demand of iron for cellular processes such as DNA synthesis/repair and energy generation (11, 14, 38). While it has long been speculated that NRF2 activation leads to decreased iron bioavailability as a means to reduce the more damaging effects of oxidative stress, the number of studies that have measured this directly has remained limited (66, 74, 75).

To address this question, we first applied IG1-FM to measure basal intracellular labile Fe(II) levels in lung adenocarcinoma A549 cells with NRF2 KO (NRF2 KO^+V0), generated via CRISPR-Cas9 methods (76), versus A549 cells with NRF2 genetic rescue (NRF2 KO^+NRF2). A549 cells harbor a loss-of-function mutation in the NRF2 regulator KEAP1 at D236H, and thus genetic rescue of NRF2 should also lead to its activation (77). Interestingly, we observed a marked decrease in labile Fe(II) in NRF2 KO^+NRF2 cells compared with NRF2 KO^+V0 control cells (Fig. 4 B and C), suggesting that NRF2 activation directly regulates intracellular labile iron bioavailability, in agreement with previous findings (66). The decrease in fluorescence signal was supported by complementary ICP−MS measurements of iron to phosphorus ratios, which show a decrease in total cellular iron for A549 cells with NRF2 rescue (SI Appendix, Fig. S23). Along these lines, NRF2 would be expected to promote iron export and storage, as ferroportin and ferritin are both ARE-containing genes (65). To better understand how each of these elements may regulate the labile iron pool, we performed RNA sequencing (RNA-seq) on the two NRF2 KO A549 cell lines. Comparing the changes in the NRF2-dependent expression profile, we see elevated gene expression of the two subunits of ferritin (FTH1, ferritin heavy chain 1; FTL, ferritin light chain) and of ferroportin (SLC40A1) in A549 cells with NRF2 KO^+NRF2 (Fig. 4D). The largest changes in expression occur for ferroportin which may explain the overall lower levels of total iron in NRF2 rescue cells, as measured by ICP-MS. Notably, the expression of transferrin receptor (TFRC), which promotes the import of the Fe(III)-bound transferrin complex into cells through receptor-mediated endocytosis (1) but is not an ARE-containing gene, did not change significantly upon genetic rescue of NRF2.

Given that IG1-FM is capable of detecting differences in intracellular labile Fe(II) in cells with complete NRF2 depletion versus NRF2 activation, we next sought to determine whether IG1-FM could be used to profile the relative extent of NRF2 regulation of labile Fe(II) pools across multiple cell lines. While NRF2 activation can result from a surplus of ROS, it can also occur due to mutations in NRF2 or its regulator KEAP1 that impair NRF2-KEAP1 binding and subsequent NRF2 degradation (Fig. 4A). These mutations that disrupt the NRF2-KEAP1 interaction are especially frequent in non–small cell lung cancers (NSCLC) and are associated with poor patient prognosis (69, 78, 79). As such, we applied IG1-FM to measure relative intracellular labile Fe(II) levels across a panel of NSCLC cell types with varying degrees of NRF2-KEAP1 interaction. In particular, H1975 and H1299 cells harbor no mutations in KEAP1. In contrast, PC-9 have NRF2 amplification (80), while H460 and H1944 cells harbor G333C and R272L KEAP1 loss-of-function mutations, respectively, that should also promote NRF2 activation. Interestingly, IG1-FM imaging showed a significantly elevated fluorescence signal in H1975 and H1299 cells versus PC-9, H460, and H1944 cells, indicating an overall higher basal level of labile Fe(II) in these cell lines (Fig. 5 A and B and SI Appendix, Fig. S24). Additionally, treatment with the KEAP1 inhibitor KI-696 resulted in decreases in IG1-FM signal only in H1975, H1299, and PC-9 cells, all of which have WT KEAP1, while KI-696 produced no significant difference in H460 and H1944 cells with KEAP1 mutations. To confirm these measurements, we performed analogous imaging with RhoNox-4, an ABS probe which utilizes different reaction chemistry for detection of labile Fe(II) (81), and TRX-Puro, a histochemical TRX probe whose signal can be normalized via treatments with free puromycin (33). Both RhoNox-4 and TRX-Puro showed similar trends in labile Fe(II) across the NSCLC panel with higher signals for H1975 and H1299 cells, and signal decreases upon treatment with KI-696 only for H1975, H1299, and PC-9 cells (SI Appendix, Figs. S25 and S26). These results suggest that NRF2 activity can directly correlate with the bioavailability of intracellular labile Fe(II). Unlike the NRF2 KO A549 cells lines, all NSCLC cell lines showed no significant changes in total iron upon KEAP1 inhibition via KI-696 aside from H1299 cells, which exhibited a slight increase (SI Appendix, Fig. S27). Additionally, whereas cell lines with NRF2 amplification or KEAP1 mutations had higher levels of total iron compared to KEAP1 WT cells, these differences were modest. Given this disparity with expression changes seen for A549 cells, we suspected that these cell lines possess different ARE gene expression profiles with greater promotion of proteins involved in iron storage and/or uptake. To confirm this hypothesis, we performed RNA-seq on H1975 cells and compared the expression levels of FTH1, FTL, SLC40A1, and TFRC between the two conditions. Upon treatment with KI-696, the most marked changes in gene expression occurred for FTH1 and FTL (SI Appendix, Fig. S28), while TFRC also witnessed a small increase in expression, likely due to stabilization of iron-responsive-element-binding proteins (82). These expression profiles suggest that there is a finely tuned balance between iron export, import, and storage in these cells. This homeostatic balance produces overall smaller changes in total iron concentrations, while still decreasing labile iron pools. Indeed, an elevation in either or both iron storage and export will lower its bioavailability and likely outweigh the increased contributions of labile iron from increased TFRC. These results are also in agreement with those of HeLa cells where inhibition of ferritinophagy via NCOA4 KO or TAX1BP1 KO leads to reduced labile iron but elevated total iron.

Finally, we hypothesized that NSCLC cells that possess lower basal labile iron due to NRF2 amplification or KEAP1 mutations would have heightened sensitivity to iron chelation compared to cells with WT KEAP1. Indeed, NRF2 expression has been previously shown to correlate positively with resistance to ferroptosis in a panel of different ovarian cancer cell lines (66). To test this notion, we treated this panel of NSCLC cells with DFO and monitored cell death over 72 h. While H1299 cells showed modest cell death upon chelation treatment, PC-9, H1944, and H460 had significantly elevated cell death compared to both H1975 and H1299 cells (Fig. 5C and SI Appendix, Fig. S29). Taken together, the data show that IG1-FM can be used as a tool to profile intracellular labile Fe(II) levels and reveal a reciprocal relationship between NRF2-ARE activation and labile Fe(II) pools that can be exploited as a unique metal-dependent vulnerability to selectively target cancer cells with hyperactivated NRF2 via iron chelation treatment (Fig. 5D).

We have presented the design, synthesis, and biological evaluation of a first-generation probe for fluorescence imaging of intracellular labile iron pools using a tandem ABS and labeling strategy. We leveraged the combination of a selective and sensitive TRX-based trigger for ABS of Fe(II) in combination with ortho-quinone-methide chemistry for protein labeling to generate IG1-FM, a green fluorescent probe for imaging intracellular labile iron pools. By installing a self-immolative linker between the analyte-responsive element and the fluoromethyl dye reporter, we anticipate that this quinone-methide chemistry will be broadly applicable for other ABS strategies featuring scaffolds with limited cellular retention. Such strategies may include further tuning of the chemical reporter to enable development of probes with red-shifted fluorescence emission, different imaging modalities, enrichment handles for -omics studies, and/or therapeutic payloads.

IG1-FM achieves a selective and sensitive Fe(II)-dependent fluorescence turn-on response, enabling the detection of changes in intracellular labile Fe(II) in live cells upon iron supplementation, iron chelation, and/or iron-dependent hormone (hepcidin) treatments. Experiments with the control reagents IG1 or hydroxymethyl 4 that do not possess the latent quinone methide group show the essential nature of the dual sensing/labeling moiety to trap the probe inside the cell to obtain robust signal-to-noise responses. Moreover, the enhanced cellular retention obtained using IG1-FM advances the study of foundational aspects of iron biology. In one example, we established that intracellular labile Fe(II) pools are directly regulated by ferritinophagy. Genetic depletion of the ferritin-binding protein NCOA4 or the autophagy receptor TAX1BP1 both resulted in decreased IG1-FM fluorescence, suggesting that degradation of ferritin can directly contribute to the native intracellular labile Fe(II) pool. Further investigations may explore how inhibition of ferritinophagy may affect subcellular localization of labile Fe(II) pools, or whether overexpression of ferritinophagy-associated proteins may impact basal iron levels. Additionally, elucidating how NCOA4 may regulate other forms of labile iron, such as that incorporated into labile heme, would be of additional future interest.

We also used IG1-FM to reveal that NRF2-ARE activation regulates the bioavailability of intracellular labile Fe(II) pools in cellular models of cancer, where we identify a reciprocal relationship in which a greater extent of NRF2 activation drives lower levels of intracellular labile Fe(II) pools across various lung cancer cell lines. We speculate that upregulation of ferroportin and ferritin, which can contribute to iron export and sequestration, respectively, in NRF2-ARE pathways play important biochemical roles, but because NRF2 can also regulate TFRC, heme oxygenase-1, both of which can increase labile iron, and can indirectly control ferritin levels through HECT and RLD domain containing E3 ubiquitin protein ligase 2 and vesicle-associated membrane protein 8 (66), this situation is likely to be mechanistically complex. As such, the outstanding question of how all AREs may impact labile Fe(II) independently remains to be fully deciphered. More broadly, our findings presage that cross talk between labile iron and NRF2-ARE signaling may be an important prognostic marker and a therapeutic target in cancer initiation and progression. Notably, we show that cells with NRF2 amplification or KEAP1 loss-of-function mutations have higher rates of cell death when treated with iron chelator than cells with WT KEAP1. This finding identifies a promising therapeutic strategy to target this iron-dependent disease vulnerability, as cancers with heightened NRF2 activity may be selectively treated while minimizing off-target effects against healthy tissue where NRF2 is maintained at low concentrations with minimal activation. Altogether, this work provides a starting point for further investigations into how iron as a transition metal signal (84) may contribute to metal-dependent disease vulnerabilities by spatial, temporal, and redox control.

A.T.P., R.L.G., T.O., C.D.M., N.P.W., G.M.D., A.R.R., and C.J.C. designed research; A.T.P., R.L.G., T.O., S.G.M., and N.P.W. performed research; A.T.P., R.L.G., T.O., C.D.M., L.T., and K.C. contributed new reagents/analytic tools; A.T.P., R.L.G., S.G.M., M.R., N.P.W., G.M.D., A.R.R., and C.J.C. analyzed data; and A.T.P., R.L.G., A.R.R., and C.J.C. wrote the paper.

The authors declare no competing interest.