The ferroptosis inducing compounds RSL3 and ML162 are not direct inhibitors of GPX4 but of TXNRD1

Abstract

Ferroptosis is defined as cell death triggered by iron-dependent lipid peroxidation that is preventable by antioxidant compounds such as ferrostatin-1. Endogenous suppressors of ferroptosis include FSP-1 and the selenoprotein GPX4, the latter of which directly enzymatically reduces lipid hydroperoxides. Small molecules that trigger ferroptosis include RSL3, ML162, and ML210; these compounds are often used in studies of ferroptosis and are generally considered as GPX4 inhibitors. Here, we found that RSL3 and ML162 completely lack capacity of inhibiting the enzymatic activity of recombinant selenoprotein GPX4. Surprisingly, these compounds were instead found to be efficient inhibitors of another selenoprotein, TXNRD1. Other known inhibitors of TXNRD1, including auranofin, TRi-1 and TRi-2, are also efficient inducers of cell death but that cell death could not be suppressed with ferrostatin-1. Our results collectively suggest that prior studies using RSL3 and ML162 may need to be reevaluated in the context of ferroptosis with regards to additional enzyme targets and mechanisms of action that may be involved.

Fig. 1

Ferroptosis inducers do not inhibit pure GPX4 but are direct inhibitors of TXNRD1 in vitro. A) Inhibitory activity of class I FINs, RSL3 (top left), ML162 (top middle), ML210 (top right), and known TXNRD1 inhibitors TRi-1 (bottom left), TRi-2 (bottom middle left), Auranofin (bottom middle right), and Aurothioglucose (bottom right) against GPX1, GPX4, GR, and TXNRD1 in activity assays using pure enzyme systems in vitro; Four-parameter dose-response curve fit to n = 2 technical replicates; B) nanoDSF experiments showing no effect on thermal stability of GPX4 by 100 μM RSL3, ML162, nor ML210 (left); Thermal stabilization of TXNRD1 by 100 μM TRi-1 and TRi-2, as well as RSL3, ML162, but not ML210 (middle left); Dose-dependent thermal stabilization of TXNRD1 by RSL3 (middle right); and a summary of triplicate thermal shifts with GPX4 and TXNRD1 in nanoDSF experiments. Data are presented as mean ± s.d. of n = 3 technical replicates.

Fig. 2

Ferroptosis inducers inhibit TXNRD1 in cells. A) Cellular inactivation of TXNRD1 measured with the RX1 activity probe at 12 h; B) cellular inactivation of TXNRD1 measured with RX1 activity probe at 24 h; C) cellular inactivation of TXNRD1 measured with RX1 activity probe at 4 h. Similar suppression in RX1 fluorescence signal was seen at all timepoints when comparing incubation of cells with TRi-1 (C, right panel), RSL3 (A-C, left panels), or ML162 (A-C, middle left panel), but not with ML210 (A-C, middle right panel), at several doses as indicated; D) summary of activity as controlled for viability is shown for 4 h (left panel), 12 h (middle panel), and 24 h (right panel) incubation times. Viability data is presented in Extended DataFig. 3. Data are presented as mean ± s.d. of n = 3 biological replicates. Unpaired, two-tailed t-test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 in comparison to DMSO control; non-significant differences are not noted.

Fig. 3

RSL3, TRi-1, and TRi-2 have differential cellular cytotoxicity profiles. A) Cytotoxic effects of RSL3 (pink curves, left panels), but not TRi-1 (pink curves, middle panels) or TRi-2 (pink curves, right panels), are suppressed by Fer-1 (green curves) but not by MitoTEMPO (black curves) in either A549 (top panels) or H1975 cells (bottom panels) in media not supplemented with selenium as determined 24 h after addition of compounds; four-parameter dose-response curve fit to n = 4 technical replicates; B) A549 cells (left panels), H1975 cells (middle panels) and HT1080 cells (right panels) grown with (black bars) or without (grey bars) 100 nM selenium supplementation in the growth medium were treated for 24 h as indicated whereupon viability was determined, here shown as percent of DMSO-treated controls. The additional treatments are a known inhibitor of ferroptosis (Ferrostatin-1), a strong antioxidant (NDGA), iron chelators (DFO and DPD50), an inhibitor of necrosis (Necrostatin-1), or a caspase inhibitor/apoptosis blocker (z-VAD-fmk). Data from n = 4 technical replicates are shown. Unpaired, two-tailed t-test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 in comparison to DMSO control of same Se condition; non-significant differences are not noted. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4

FINs, but not TRi compounds, affect GPX4 migration. A) A549 (left panel) or H1975 cells (middle panel) were seeded 18 h (with or without 2 μM Ferrostatin-1 pre-treatment for 14 h) before being treated with 3 μM of compound for 24 h. Proteins were resolved by reducing SDS-PAGE (10 mM DTT) and probed with the ab125066 anti-GPX4 antibody. GAPDH and Ponceau stains acted as the loading control; blots are representative of n = 3 (left panel), or n = 2 (middle panel) biological replicates. The 24-hour cytotoxicity profiles of the compounds in H1975 cells under these conditions are also presented (right panel). Cytotoxicity data are presented as mean ± s.d. of the average of n = 3 technical replicates; B) H1975 cells were seeded 18 h (with or without 2 μM Ferrostatin-1 pre-treatment for 14 h) before being treated with ∼IC₅₀ concentrations of compound for 6 h. Proteins were resolved by reducing SDS-PAGE (10 mM DTT) and probed with either ab125066 anti-GPX4 antibody (left panel), or sc-166120 anti-GPX4 antibody (right panel). GAPDH and Ponceau stains acted as the loading control; blots are representative of n = 1 (right) biological replicate; C) Recombinant GPX4 was incubated with DMSO, 8 μM RSL3 or 0.5 μM TRi-1 for 30 min (right panel) in presence of full reaction mixture. H1975 lysates were prepared as in A. Proteins were resolved by reducing SDS-PAGE (10 mM DTT) and probed with the ab125066 anti-GPX4 antibody. Ponceau staining was the loading control; blot is representative of n = 2 technical replicates D) A549 cells were seeded 18 h before being treated with 3 μM compound for 24 h. Together, the left and central panel were from the same biological replicate, the right panel from a separate replicate. Proteins were resolved by SDS-PAGE with samples pre-treated with varying concentrations of DTT (0–10 mM in right panel, or otherwise indicated) and probed with the ab125066 anti-GPX4 antibody. Note the sharper slower migrating GPX4-immunoreactive band in samples from cells treated with FINs but not TRi compounds. n = 2 biological replicates from A549 (A, left) were used, left and middle blots from the same replicate, and right from a second.