Labile iron pool dynamics do not drive ferroptosis in colorectal cancer cells.

Results

Exogenous iron sensitizes CRC cells to JKE-induced ferroptosis
The selenoprotein glutathione peroxidase 4 (GPX4) is a central regulator of ferroptosis (23). Although inhibition of GPX4 is widely accepted as a mechanism to induce ferroptosis, we have previously shown that the commonly used GPX4 inhibitor RSL3 exhibits off-target effects and non-specific inhibition across the selenoproteome, including thioredoxin peroxidases (24). We therefore utilized a more specific, next-generation GPX4 inhibitor, JKE1674, for our experiments.
Among the CRC cell lines, HCT116 and SW480 have been reported to be relatively resistant to ferroptosis inducers, whereas RKO cells are highly sensitive (17). We found that JKE1674-mediated GPX4 inhibition was less potent in HCT116, SW480, and RKO compared to other ferroptosis inducers (24) (Figs. 1A, S1A). To determine whether manipulating the labile iron pool could sensitize cells to JKE1674 (0.625, 1.25, 2.5 μM), we cotreated cells with ferric ammonium citrate (FAC) at increasing concentrations (0.25, 0.5, and 1.0 mM). The fibrosarcoma cell line HT1080, a well-established ferroptosis sensitive cell line (13), was included as a positive control. Exogenous iron significantly reduced cell viability in the ferroptosis-resistant CRC lines, HCT116 and SW480, particularly at the highest FAC concentration (Fig. 1B). Addition of ferroptosis inhibitor, liproxstatin-1, rescued cell viability (Fig. 1, A and B). Consistent with these findings, assessment of cell death by SYTOX green revealed that iron supplementation also robustly increased cell death (Fig. 1, C and D).
We treated cells with combinations of FAC and JKE1674, and performed synergy analysis using the Bliss independence dose-response surface model (25). This model assumes additivity when two agents act independently, with deviations reflecting synergy or antagonism. Synergy scores >10 indicate synergy, scores between −10 and 10 indicate additivity, and scores <–10 indicate antagonism. The mean synergy score represents the average of all scores across concentration combinations found within the plot. Across all four cell lines tested, cell viability assays revealed robust synergistic interactions between FAC and JKE1674 over a range of concentrations in HT1080 cells, and additive interactions in HCT116, SW480, and RKO cells (Fig. 1E).

Exogenous iron potentiates RSL3 and imadazole ketone erastin -induced ferroptosis in CRC cells
Next, we tested whether FAC could potentiate two other ferroptotic inducers, RSL3 and the SLC7A11 inhibitor, imadazole ketone erastin (IKE) (Figs. 2A and 3A). Co-treatment of HCT116 and RKO cells with RSL3 (1, 0.5, 0.25 μM) and FAC resulted in a dose-dependent decrease in cell viability (Figs. 2A, S2A, and S2B) and increased cell death (Fig. 2, B and C). In contrast, SW480 and HT1080 cells did not exhibit enhanced sensitivity to RSL3 upon FAC addition. Ferroptosis can also be triggered by inhibition of the cystine/glutamate antiporter SLC7A11, which increases lipid peroxidation (9, 26). Notably, FAC potentiated IKE-induced cell death and decreased cell viability in the HCT116, SW480, and RKO cell lines (Fig. 3, A–C).
To quantify these interactions, we again applied the Bliss independence dose-response surface model. Cotreatment with RSL3 and FAC demonstrated strong synergy in HCT116 cells and additive effects in SW480 and RKO cells, and antagonistic affects in HT1080 cells (Fig. 2D). For IKE, synergistic effects were observed in SW480 and HT1080 cells, additive responses in HCT116, and antagonistic effects in RKO (Fig. 3D). Together, these results show that increasing the labile iron pool through exogenous iron supplementation can potentiate ferroptosis both through GPX4-dependent and GPX4-independent mechanisms.

Iron homeostatic gene expression is not altered following RSL3, JKE, and IKE-induced ferroptosis
In the intestine, iron is imported and exported through the key transporters DMT1, TFRC, and FPN (the only known mammalian iron exporter). Ferritin serves as the primary intracellular iron storage protein, sequestering excess iron from the LIP (1, 27). The expression of these iron homeostasis genes is regulated by iron regulatory proteins (IRPs), a class of RNA-binding proteins that control translation by binding to iron-responsive elements within the untranslated regions of target mRNAs (28, 29). Depending on the location of iron-responsive elements in the 3′ or 5′ untranslated region, IRP binding can upregulate or downregulate the expression of DMT1, TFRC, FPN, and ferritin heavy chain 1 (FTH1) during systemic iron deficiency or overload (30, 31).
In addition to IRPs, iron homeostasis genes are transcriptionally regulated by transcription factors. Hypoxia-inducible factor-2α and nuclear factor erythroid two–related factor-2 bind to hypoxia response elements and antioxidant response elements, respectively, particularly under conditions of hypoxia or oxidative stress (14, 32, 33, 34, 35).
It is widely assumed (18, 19), but not rigorously tested, that ferroptosis is accompanied by an endogenous increase in the LIP, which would sensitize cells to iron-dependent lipid peroxidation. To investigate this, we examined the expression of iron homeostasis genes (DMT1, TFRC, FTH1, and FPN) following induction of ferroptosis. Quantitative RT-PCR was performed 16 h after treatment with three distinct ferroptosis inducers (Fig. 4). Ferroptosis inducer concentrations were selected based on their ability to robustly decrease cell viability after 16 h (Fig. 4, A–C). Moreover, this time point captures early shifts in the LIP while still maintaining sufficient cell viability. Surprisingly, ferroptosis induction did not result in widespread transcriptional changes in iron homeostasis genes, with the exception of SW480 cells treated with RSL3 (Fig. 4A) and in HT1080 cells treated with IKE (Fig. 4B). Increased TFRC1 expression after treatment with RSL3 in SW480 cells and increased FTH1 expression after treatment with IKE in RKO and HT1080 cells may point toward transcriptional remodeling of the LIP through exogenous uptake and storage of iron during ferroptosis. However, these changes in gene expression are not consistent across dosages (Fig. S4C). These findings suggest that transcription of iron homeostasis genes is not altered and do not drive changes in LIP during ferroptosis.

The labile iron pool is not altered during ferroptosis
Previous studies suggested that the LIP increases following the initiation of ferroptosis, primarily through the degradation of ferritin via ferritinophagy (18, 19). These studies measured the LIP using Calcein AM (36, 37, 38) and PhenGreen SK (39), chelation-based iron probes that, while still widely used, have inherent limitations (40, 41). Specifically, these probes are dependent on cellular esterases to allow for membrane permeability and therefore uptake and distribution can be variable. Moreover, iron binding to these probes quenches fluorescence, necessitating the use of a second, cell-permeable iron chelator to generate a turn-on signal through competitive binding. Consequently, the observed fluorescence response reflects a complex interplay among probe concentrations, the extent of ester hydrolysis and intracellular sequestration, and the relative affinities of both probes and comepeting chelator for Fe2+ and Fe3+. Therefore accurate measurement requires careful calibration experiments that are rarely performed in practice (36, 39).
To assess LIP levels after ferroptosis induction with RSL3, JKE, or IKE, we employed the reactivity-based probe TRX-PURO (20), which detects LIP via the Fenton-type reactivity that is associated with labile Fe2+ but not with Fe3+ in storage nor with protein-bound iron co-factors. TRX-PURO is an endoperoxide-caged conjugate of puromycin that is uncaged upon reaction of the endoperoxide function with labile iron. Uncaged puromycin is incorporated into nascent polypeptides at the ribosome, where it can be readily detected by flow cytometry, western blotting, or immunofluorescence staining using α-puromycin antibodies (Fig. 5A). We validated the iron-dependent response of TRX-PURO in HEK293 cells cotreated with the ferric iron source FAC (10–50 μM). FAC is only competent for TRX-PURO activation following its internalization and reduction to the ferrous LIP. Puromycin incorporation increased in a dose-dependent manner with FAC, consistent with enhanced labile iron availability (Fig. 5B). Co-treatment with the iron chelator deferoxamine reduced puromycin incorporation to baseline levels, confirming the probe’s specificity for iron.
To determine whether the LIP increases endogenously during ferroptosis, we assessed TRX-PURO activation via puromycin incorporation over 16 h of continuous treatment with probe in the presence of RSL3, JKE, or IKE (Fig. 5, C–E). Maximal activation of TRX-PURO requires ∼8 h due to a rate-determining retro-Michael reaction involved in puromycin release (20) and so these studies are necessarily limited in their ability to report on fluctuations in LIP at very early timepoints. Nevertheless, we found that increasing concentrations of RSL3 led to increased LIP in HCT116 and SW480 cells while RKO and HT1080 cells showed a more stochastic TRX-PURO that was not dose-responsive (Figs. 5C and S5A). Conversely, JKE treatment did not affect the LIP in HCT116 nor SW480 cells at any of the doses studied while RKO and HT1080 cells showed elevated LIP at higher doses (Figs. 5D and S5B). Finally, IKE increased the labile iron pool concentration in SW480 cells marginally, but this increase was not consistent over multiple dosages. Furthermore, IKE failed to increase labile iron over untreated controls in any other cell lines examined (Figs. 5E and S5C).
To overcome the inability of TRX-PURO to assess timepoints earlier than eight hours we used FerroOrange, another reactivity-based probe to detect changes in the LIP at four hours after induction of ferroptosis (Fig. 6). FerroOrange is a rhodamine derivative attached with an N-oxide group. Rhodamines are inherently fluorescent, although the addition of the N-oxide group quenches this fluorescence. After reacting with Fe2+, the N-oxide group is deoxygenated allowing the probe to be “turned on” and activate fluorescence (21, 22). We validated the response of FerroOrange to iron in HCT116 cells treated with FAC (25–100 μM). Incubation of the cells with FerroOrange for 30 min after treatment of FAC for 16 h robustly increased fluorescence of FerroOrange and cotreatment with deferoxamine (50 μM) attenuated the response to baseline levels (Fig. 6A).
Previous studies have shown that the LIP increases after treatment with erastin within the first six hours after initiation of ferroptosis (19, 42). This suggests that an increase in the LIP may be a part of a feedforward mechanism whereby the LIP increases first before the execution of ferroptosis and detectable decreases in cell viability are observed (Fig. 6, B–E). Using FerroOrange we probed the concentration of the LIP after four hours of treatment with RSL3, JKE, and IKE. We found that treatment with RSL3 did not change the LIP in HCT116 or RKO cell lines. In contrast, SW480 and HT1080s showed a decrease in LIP concentration. Treatment with JKE did not cause any increase in labile iron in any colorectal cancer cell line and decreased the LIP size in HT1080 cells. Lastly, IKE did not change LIP levels in any cell line. The decrease in LIP size may be due to limitations of FerroOrange as a live-cell stain. As ferroptosis progresses in the cell, changes in membrane permeability may cause probe leakage resulting in a misleading signal decline. Nevertheless, even accounting for these limitations, induction of ferroptosis did not cause any increase in the LIP in any of the cell lines used. Taken together, these data indicate that elevation of the LIP is not a consistently conserved or essential feature of pharmacologically-induced ferroptosis.

Discussion
Ferroptosis is a form of regulated cell death that depends on iron and is characterized by the accumulation of lipid peroxides (10). Unlike apoptosis or necroptosis, ferroptosis lacks well-defined molecular markers. Additionally, whether endogenous mediators regulate ferroptosis is poorly understood. Among these regulators is thought to be iron and particularly the LIP. The LIP plays a central role by triggering the Fenton reaction and initiating lipid peroxidation. However, the extent to which endogenous fluctuations in iron availability through the LIP directly contribute to ferroptosis remains unknown.
Our data confirms that iron remains functionally essential for ferroptosis. When we increased intracellular iron levels through exogenous supplementation with FAC, CRC cells became markedly more sensitive to ferroptosis inducers as seen by a decrease in cell viability and increase cell death. These phenotypes were rescued by addition of ferroptosis inhibitor, liproxstatin - 1. This synergy was observed in multiple ferroptosis-sensitive and -resistant CRC cell lines and the ferroptosis-sensitive HT1080 fibrosarcoma cell line.
Despite the reliance on iron for triggering ferroptosis, our results suggest CRC cells may not potentiate ferroptosis through iron mobilization but instead exhibit a relatively stable LIP. Our quantification of the LIP using TRX-PURO indicates that higher concentrations of RSL3 endogenously increase the LIP in CRC cells. We have previously reported (21) that RSL3 has several off targets independent of GPX4 which may contribute to an increase in LIP, which may confound the results. When using JKE, a more specific GPX4 inhibitor, we observed no change in labile iron in ferroptosis-resistant cell lines, HCT116 and SW480 and increase in the LIP in ferroptosis-sensitive cell lines, RKO and HT1080 at higher JKE concentrations. IKE-induced ferroptosis showed little to no change to the LIP in any of our tested cell lines. Furthermore, the minimal transcriptional response of iron homeostasis genes (TFRC, DMT1, FTH1, FPN) suggests that ferroptosis initiation in these cells does not trigger broad changes in iron homeostasis, at least within the early time points examined. Taken together these results highlight an important distinction: while ferroptosis in CRC cells requires iron, the endogenous regulation of the LIP may not be the primary determinant of ferroptotic potentiation. This is in contrast to a study that showed erastin-induced ferroptosis increases the LIP in breast cancer cells (43). The variation among cell lines implies context-specific regulation of iron metabolism, influenced by cell line, tissue origin, and mechanism of ferroptosis induction (GPX4-independent versus dependent).
The compartmentalization of iron in subcellular compartments may also contribute to ferroptosis in ways not captured by bulk cytosolic LIP measurements. It has previously been shown that erastin-induced ferroptosis increases the LIP in the endoplasmic reticulum and lysosome in HT1080 cells (44). Additionally recent work by Cañeque et al. demonstrates that activating lysosomal iron triggers ferroptosis in cancer cells, particularly affecting drug-tolerant persister populations (45). This finding suggests that compartment-specific iron pools, such as lysosomal iron, play a pivotal role in ferroptosis regulation. Our observations are consistent with the idea that the cytosolic LIP remains largely stable during ferroptosis and could suggest that iron compartmentalization or redistribution, rather than changes in total cytosolic iron levels, plays a critical role in ferroptosis sensitivity. Many tumors including CRC are characterized by elevated iron accumulation (14, 15, 16, 17) which suggests mechanisms to buffer excess redox-active iron to avoid cytotoxicity. CRC may uniquely have an adaptive mechanism to attenuate LIP size to limit oxidative damage. Identifying and targeting these iron-buffering pathways could represent an approach to increase the tumor sensitivity to ferroptosis-inducing therapies. Further studies are needed to determine whether iron compartmentalization within subcellular compartments plays a role in CRC.
In summary, our study reveals that while exogenous iron can synergize with ferroptosis inducers in CRC, the LIP does not increase endogenously during ferroptosis. Future studies should investigate whether targeting iron compartmentalization or buffering systems can enhance ferroptosis sensitivity in CRC and other iron-dependent cancers.

Experimental procedures

Drugs
RSL3, JKE1674, and IKE were prepared fresh from powder prior to use and diluted in DMSO. RSL3, JKE1674, and IKE were stored with a stock concentration of 10 mM in DMSO at −20 °C. TRX-PURO was stored at a stock concentration of 4 mM in DMSO at −20 °C. FerroOrange was stored at stock concentration of 1 mM in DMSO at −20 °C and protected from light. SYTOX was stored at stock concentration of 0.5 mM at −20 °C. Liproxstatin - 1 was stored at stock concentration 10 mM at −20 °C. All final concentrations for experiments were prepared in cell culture media.

Cell lines
Human intestinal colon cancer cell lines HCT116, SW480, RKO along with human embryonic kidney cells HEK293T, and fibrosarcoma cells HT1080 were utilized. All cell lines were cultured in complete DMEM medium supplemented with 10% fetal bovine serum and 1% antibiotic/antimycotic solution, maintained at 37 °C in an atmosphere of 5% CO2 and 21% O2.

Cell viability, death, and synergy assays
wCells were plated between 250 to 500 cells/well and were allowed to adhere overnight, imaged for a day 0 reading and then immediately treated as indicated in the figure legend. Images were acquired 72 h after treatment for final reading. Imaging was done on the Cytation 5 Imaging Multi-Mode reader with attached BioSpa from Agilent BioTek. Cytation software was used to quantify the cell count (https://www.agilent.com/en/product/cell-analysis/cell-imaging-microscopy/cell-imaging-microscopy-software/biotek-gen5-software-for-imaging-microscopy-1623226?srsltid=AfmBOorT7lf_g02lvH9YFAc_kjAE7Ilbd7gPVzoSNSVicKF-WPz6X8eX and https://www.agilent.com/en/product/microplate-instrumentation/microplate-instrumentation-control-analysis-software/imager-reader-control-analysis-software/biotek-gen5-software-for-detection-1623227?srsltid=AfmBOop41J80rMcTojBwnYjYiR9BUHDs2drSNuJX4x6f8hIy4Ge1-utV).
Analysis was performed by normalization to cell number at first reading (day 0) - all wells plated were averaged for analysis. Graphs were plotted using Prism with error bars representing mean ± standard deviation.
To determine synergy between two different compounds, technical replicates of control and drug treatment wells within a plate were averaged. Fold changes relative to the averaged vehicle treatment group were calculated and then converted to a percentage to represent percent viability. The percent viability of the control groups (Vehicle, 0.25 mM, 0.5 mM, 1 mM FAC) were then averaged to use as control values in the synergy analysis. These values were then used to determine synergy by the Bliss independence dose-response surface model using the SynergyFinder + web application (14).

qPCR
Cell lines were treated for 16 h and RNA was extracted using the Trizol reagent. RNA yield was then quantified using a Nanodrop. 1 μg of RNA was reverse transcribed to cDNA using the Invitrogen SuperScriptTM III First-Strand Synthesis System. After cDNA was collected Real time PCR reactions were done using three technical replicates for each sample. Then cDNA gene specific primers and SYBR green master mix were combined and then run on the Applied BioSystems QuantStudio 5 Real-Time PCR System. GAPDH was used as the housekeeping gene to calculate fold-change of the genes using the ΔΔCt method.
DMT1 F: CCTGTGGCTAATGGTGGAGTTGG.
DMT1 R: GGAGATTGATGGCGATGGCTGAC.
TFRC1 F: AGTTGAACAAAGTGGCACGAG.
TFRC1 R: GCAGTTGGCTGTTGTACCTC.
FTH1 F: TCCTACGTTTACCTGTCCATG.
FTH1 R: GTTTGTGCAGTTCCAGTAGTG.
FPN F: CACAACCGCCAGAGAGGATG.
FPN R: CACATCCGATCTCCCCAAGT.

Western blot
Cells were seeded in a 12-well plate each condition and allowed to adhere overnight. Cells were plated to reach ∼0.4x10ˆ6 cells/well at time of harvest. After adherence cells were treated for 16 h with RSL3, JKE1674, IKE, and with or without TRX PURO (1 mM). Cells were then lysed with RIPA assay buffer with added protease (1:100 dilution; MilliporeSigma) and phosphatase (1:100 dilution; Thermo Fisher Scientific) inhibitors. Lysates were quantified by BCA protein assay kit (Pierce – Thermo Fisher Scientific) and normalized for loading. Solubilized proteins were resolved on 10% SDS-polyacrylamide gels and transferred to nitrocellulose membrane, blocked with 5% milk in TBST, and immunoblotted with the indicated primary antibodies: Actin (1:1000) (Proteintech 66009-1-Ig), and α-puromycin (1:1000) (PMY-2A4). Horseradish peroxidase–conjugated secondary antibodies anti mouse (1:2500) were purchased from Cell Signaling (7076, 7074) and Thermo Fisher Scientific (A15999). Immunoblots were imaged using the iBright imaging system.

Flow cytometry for TRX-PURO and FerroOrange
After treatment with TRX-PURO (1 μM), RSL3 (0.25, 0.5, 1 μM), JKE (0.625, 1.25, 2.5 μM), or IKE (0.625, 1.25, 2.5 μM) for 16 h cells were harvested using Trypsin and quenched with FACS buffer (PBS + 3% BSA + 1 u. They were then pelleted, washed, and resuspended in PBS in a 96 well round-bottom plate. Cells were then stained with live fixable blue (1:2500; diluted in PBS) (Invitrogen L23105) for 10 min, quenched with FACS buffer, and centrifuged at 600g for five minutes. Cell pellets were washed once with PBS and then fixed with 4% paraformaldehyde on ice for 30 min. After fixation, the cells were stained overnight with Alexa647 anti-puromycin conjugated antibody (1:500 dilution) (BioLegend 381,508) in permeabilization buffer (Invitrogen 00-8333-56) while protected from light.
After treatment with RSL3 (1 μM), JKE (10 or 2.5 μM), IKE (10 or 2.5 μM) for four hours, cells were harvested using Trypsin and quenched with cell culture medium or FACS buffer). They were then pelleted, washed, and resuspended in PBS in a 96 well round-bottom plate. Cells were then stained with FerroOrange (1:1000) for 30 min, washed twice with PBS, and then resuspended in FACS buffer while protected from light.

Statistical analysis
Results are expressed as the mean ± standard deviation for all figures unless otherwise noted. Significance between two groups was tested using a two tailed unpaired t test. Significance among multiple groups was tested using a one-way ANOVA or two-way ANOVA. For all ANOVAs the Tukey-Kramer test was used for posthoc multiple comparisons. GraphPad Prism 10.0 (graphpad.com) was used for the statistical analysis. Statistical significance is described in the figure legends as: ∗p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001.

Data availability
Additional information or raw data will be shared upon request addressed to the contact authors.

Supporting information
This article contains supporting information.

Conflict of interest
The authors declare that they have no conflicts of interest with the contents of this article.