Ferroptosis is an iron-dependent, non-apoptotic form of cell death initiated by oxidative stress and lipid peroxidation. Recent evidence has linked ferroptosis to the action of the

transcription factor Nuclear factor erythroid-2 derived,-like-1 (NFE2L1). NFE2L1 regulates proteasome abundance in an adaptive fashion, maintaining protein quality control to secure cellular

homeostasis, but the regulation of NFE2L1 during ferroptosis and the role of the ubiquitin-proteasome system (UPS) herein are still unclear. In the present study, using an unbiased

proteomic approach charting the specific ubiquitylation sites, we show that induction of ferroptosis leads to recalibration of the UPS. RSL3-induced ferroptosis inhibits proteasome activity

homolog 2 (DDI2) is a critical step for the ferroptosis-induced feed-back loop of proteasome function. Cells lacking DDI2 cannot activate NFE2L1 in response to RSL3 and show global

hyperubiquitylation. Genetic or chemical induction of ferroptosis in cells with a disrupted DDI2-NFE2L1 pathway diminishes proteasomal activity and promotes cell death. Also, treating cells

with the clinical drug nelfinavir, which inhibits DDI2, sensitized cells to ferroptosis. In conclusion, our results provide new insight into the importance of the UPS in ferroptosis and

highlight the role of the DDI2-NFE2L1 as a potential therapeutic target. Manipulating DDI2-NFE2L1 activity through chemical inhibition might help sensitizing cells to ferroptosis, thus

enhancing existing cancer therapies.

Regulated cell death is initiated by intracellular and extracellular perturbations that trigger tightly orchestrated molecular programs [1]. Ferroptosis is a form of non-apoptotic cell death

mediated by iron-dependent lipid peroxidation and loss of plasma membrane integrity [2, 3]. Recent studies have implicated ferroptosis in several pathologies, such as neurodegeneration and

cancer [4]. Execution of ferroptosis is tightly linked to lipid and glutathione metabolism [5] and several chemical compounds have been shown to induce ferroptosis. Glutathione peroxidase 4

(GPX4) is a critical enzyme protecting from lipid reactive oxygen species (ROS) formation and, thus, from ferroptosis by most prominently using glutathione for its antioxidative activity

[6]. While depletion of the reduced glutathione pool predisposes cells to ferroptosis, the compound RSL3 directly inhibits GPX4 [2, 7, 8]. Thus, oxidative stress is sensed and mitigated by

GPX4 and its inactivation leads to lipid peroxidation and cell death in cells and mouse models [8]. Interestingly, ferroptosis is linked to adaptive changes in protein homeostasis, as

ferroptosis initiation is associated with diminished proteasomal activity and restoration of proteasomal activity protects cells from ferroptotic cell death [9]. The ubiquitin-proteasome

system (UPS) manages the degradation of unwanted, obsolete, or damaged proteins [10]. The UPS controls almost all cellular processes by precision proteolysis, enabled by specific ubiquitin

ligases [11]. Attachment of ubiquitin to these proteins is the prerequisite for the ATP-dependent degradation by the proteasome. The 26S proteasome is a multi-subunit protease complex, which

consists of the barrel-shaped 20S proteolytic core particle capped with one or two 19S regulatory particles for binding to the polyubiquitylated proteins [12]. While it is accepted that the

UPS is a critical pillar of cellular health, the roles of UPS remodeling and the nature of these changes in ferroptosis remain unclear. We and others have shown that the transcription

factor Nuclear factor erythroid-2, like-1 (NFE2L1, also known as TCF11 or NRF1) regulates ferroptosis [9, 13]. NFE2L1 upregulates the expression of genes encoding proteasome subunit genes

[14, 15], thus restoring proteasomal activity and protecting from ferroptosis. NFE2L1 mediates the adaptive component of proteasomal activity and is subject to complex posttranslational

regulation: NFE2L1 is a cap’n’collar transcription factor tethered in the endoplasmic reticulum membrane. In a simplified model, NFE2L1 undergoes deglycosylation by N-glycanase 1 (NGLY1,

also known as PNGase) and proteolytic cleavage by the protease DNA damage-inducible 1 homolog 2 (DDI2) [16, 17]. However, the resulting fragment containing the bZIP-DNA-binding

domain-containing is virtually absent in most cells, as it is being continuously ubiquitylated and degraded by the proteasome [18, 19]. Upon treatment with chemical proteasome inhibitors or

if the levels of ubiquitylated proteins exceed the capacity of the available proteasomes, NFE2L1 escapes degradation and restores proteasomal activity [14]. However, the impact of

DDI2-mediated activation of NFE2L1 on UPS in ferroptosis remains unknown. Here, we determine the impact of ferroptosis on UPS and global ubiquitylation using an unbiased MS approach.

Furthermore, we demonstrate a critical role of DDI2-mediated activation of the NFE2L1 pathway in calibrating the UPS for ferroptosis protection.

Human EA.hy926 cells and mouse WT1 cells were used for cellular experiments. EA.hy926 DDI2 KO cells were created previously [19]. Cells were cultured in DMEM Glutamax (Thermo), supplemented

with 10% v/v fetal bovine serum (FBS, Sigma) and 1% v/v Penicillin-Streptomycin (10.000 U/ml, Thermo). Cells were incubated at 37 °C, 5% v/v CO2 and were passaged two to three times a week.

For experiments 300,000, 100,000, and 25,000 cells were seeded at 6, 24, and 96-well cell culture plates, respectively. Primary GPX4 mutant fibroblasts (derived from a patient with a

homozygous mutation c.647 G > A in exon 6 of GPX4) and control cells were kindly provided by Sanath K. Ramesh (curegpx4.org). These cells were incubated in DMEM GlutaMax supplemented with

10% v/v FBS and 1% v/v PS, supplemented with 10 µM ferrostatin-1 (SelleckChem). Reverse transfection with SMARTpool siRNA (Dharmacon) was performed using 30 nM of siRNA in RNAiMAX (Thermo).

Cells were incubated for 48 h after transfection, and then treatments were performed. For overexpression experiments, 250 ng of each plasmid construct (Empty vector, DDI2 WT, DDI D252,

NFE2L1 WT, and NFE2L1-8ND) was added to TransIT-X2® Transfection Reagent (Mirus) and after 10 min incubation, cells were added to the medium. For treatments, different concentrations of RSL3

(SelleckChem), FIN56 (SelleckChem) and ferrostatin-1 (SelleckChem) were used as indicated. Inhibition of proteasome and DDI2 was performed using bortezomib (SelleckChem) and nelfinavir

mesylate (Sigma), respectively. Cell viability was assessed by AquaBluer (MultiTarget Pharmaceuticals), by changing the medium of cells to 1:100 of AquaBluer in phenol red-free DMEM

GlutaMax, after 20 h of treatment. Cell plates were incubated for 4 h at 37 °C incubator and using a Spark Reader (Tecan) fluorescence was measured at emission of 590 nm with an excitation

of 540 nm.

Cells were lysed in RIPA buffer (150 mM NaCl (Merck), 5 mM EDTA (Merck), 50 mM Tris pH 8 (Merck), 1% v/v IGEPAL CA-630 (Sigma), 0.5% w/v sodium deoxycholate (Sigma Aldrich), 0.1% w/v SDS

(Roth),1 mM protease inhibitors (Sigma)) for 3 min in TissueLyser II (30 Hz; Qiagen). Cell lysates were centrifuged for 15 min (4 °C, 21,000 g) and the concentration of proteins in

supernatant was determined by Pierce BCA Protein Assay (Thermo) according to the manual. Proteins were denatured for 5 min at 95 °C in Bolt LDS Sample buffer with 5% v/v 2-mercaptoethanol

(Sigma). 10-30 µg of proteins were loaded in Bolt 4-12% Bis-Tris gels (Thermo) followed by transferring onto a 0.2 mm PVDF membrane (Bio-Rad) at 25 V, 1.3 A for 7 min. Membranes were blocked

in TBS-T (200 mM Tris (Merck), 1.36 mM NaCl (Merck), 0.1% Tween-20 (Sigma)) containing 5% w/v milk powder for 1 h at room temperature after staining in Ponceau-S (Sigma Aldrich). Incubation

by primary antibodies (Supplementary Table 1) was performed overnight at 4 °C followed by washing membranes three times for 10 min with TBS-T and incubation with secondary antibodies for 1 

h at room temperature. SuperSignal West Pico PLUS (Thermo) was used for developing blots in a ChemiDoc MP imager (Bio-Rad). All uncropped blots can be found in Supplementary Figs. S1-S3.

Cells lysis was performed in lysis buffer (40 mM Tris pH 7.2 (Merck), 50 nM NaCl (Merck), 5 mM MgCl2(6H2O) (Merck), 10% v/v glycerol (Sigma), 2 mM ATP (Sigma), 2 mM 2-mercaptoethanol

(Sigma). Proteasome Activity Fluorometric Assay II kit (UBPBio, J41110) was used to measure proteasome activity. BCA Protein Assay (Bio-Rad) was used to normalize the results to protein

levels.

Cells were lysed in lysis buffer (50 mM Tris/HCl pH 7.5, 2 mM DTT, 10% v/v glycerol, 5 mM MgCl2, 0.05% v/v Digitonin, 2 mM ATP) containing phosphatase inhibitor (PhosphoStop, Roche

Diagnostics) as described previously [20]. Samples were incubated on ice for 20 min and centrifuged twice. Concentration of proteins was determined with Bio-Rad Protein Assay Kit II. 15 µg

of protein were loaded in NuPAGE 3-8% Tris-Acetate gels (Thermo) and run at constant voltage of 150 V for 4 h. Gels were kept for 30 min at 37 °C in activity buffer (1 mM MgCl2, 50 mM Tris,

1 mM DTT) with 0.05 mM substrate Suc-Leu-Leu-Val-Tyr-AMC (Bachem). ChemiDoc MP (Bio-Rad) was used to measure the fluorescent signal. Afterwards, to prepare samples for blotting gel was

incubated in a solubilization buffer (2% w/v SDS, 1.5% v/v 2-Mercaptoethanol, 66 mM Na2CO3) for 15 min. Samples were transferred to a PVDF membrane at 40 mA through tank transfer. The

membrane was kept for 1 h in the ROTI-block (Roth) and overnight in primary antibody (1:1000). The day after, the membrane was incubated for 3 h in the secondary antibody (1:10,000) at room

temperature and developed as described above.

HEK293a cells stably expressing short half-life firefly luciferase driven by upstream activator sequence (UAS) promoter and a chimeric NFE2L1 in which the DNA-binding domain was replaced by

the UAS-targeting Gal4 DNA-binding domain [21]. The assay measures nuclear translocation and its transactivation by binding of NFE2L1-UAS to a luciferase promoter [21]. 30,000 cells were

seeded in 96-well plates and after treatment with RSL3, cells were lysed, and luciferase emission was measured using Dual-Glo Luciferase Assay System (Promega) according to the

manufacturer’s instructions.

RNA extraction was performed using Nucleospin RNA kit (Macherey Nagel), based on manufacturer’s instruction, and the concentration of RNAs were measured with a NanoDrop spectrophotometer

(Implen). To prepare complementary DNA (cDNA), 500 ng RNA were added to 2 µL of MaximaTM H Master Mix 5x (Thermo Fischer) and the total volume was adjusted to 10 µL with H2O. The cDNA was

diluted 1:40 in H2O, and 4 µL of cDNA, 5 µL of PowerUpTM SYBR Green Master Mix (Thermo), and 1 µL OF 5 µM primer stock (Supplementary Table 2) were used to measure Relative gene expression.

Cycle thresholds (Ct) of gene of interest were measured using a Quant-Studio 5 RealTime PCR system (Thermo). Relative gene expression was normalized to TATA-box binding protein (TBP) levels

by the ddCt-method.

Protein digestion was performed as described previously [22,23,24]. Cells were lysed in SDC buffer (1% v/v SDC in 100 mM Tris-HCl, pH 8.5) followed by boiling for 5 min at 95 °C while

shaking at 1000 rpm. Protein concentrations of lysates were determined after 15 min of sonication (Bioruptor, Diagenode, cycles of 30 s) using the Pierce BCA Protein Assay (Thermo). CAA and

TCEP (final concentrations: 40 mM and 10 mM respectively) were added to 5 mg protein. After 10 min incubation of samples at 45 °C in the dark shaking at 1000 rpm, Trypsin (1:50 w/w) and LysC

(1:50 w/w) were added. Samples then were kept overnight at 37 °C while shaking at 1000 rpm for digestion. For proteome analysis, sample aliquots (~15 µg) were desalted in SDB-RPS StageTips

(Empore). Briefly, samples were diluted with 1% TFA in isopropanol to a final volume of 200 µl, loaded onto StageTips, and sequentially washed with 200 µl of 1% v/v TFA in isopropanol and

200 µl 0.2% v/v TFA/2% v/v ACN. Peptides were eluted with freshly prepared 60 µl of 1.25% v/v ammonium hydroxide (NH4OH)/80% v/v ACN and dried using a SpeedVac centrifuge (Eppendorf). Dried

peptides were resuspended in 6 µL buffer A (2% v/v ACN/0.1% v/v TFA). Di-Gly enrichment samples were diluted with 1% v/v TFA in isopropanol (1:5). For peptide cleanup, SDB-RPS cartridges

(Strata™-X-C, 200 mg/6 ml, Phenomenex Inc.) were equilibrated with 8 bed volumes (BV) of 30% v/v MeOH/1% v/v TFA and washed with 8 BV of 0.2% v/v TFA. Samples were loaded by gravity flow and

sequentially washed twice with 8 BV 1% TFA in isopropanol and once with 8 BV 0.2% v/v TFA/2% v/v ACN. Peptides were eluted with 2 × 4 BV 1.25% v/v NH4OH/80% v/v ACN and diluted with ddH2O

to a final of 35% v/v ACN. Samples were dried via a SpeedVac Centrifuge overnight.

We used the PTMScan® Ubiquitin Remnant Motif (Cell Signaling). The peptides were resuspended in 500 µL immunoaffinity purification (IAP) buffer and sonicated (Bioruptor) for 15 min. BCA

Protein Assay (Thermo) was used to determine protein concentration. The antibody-coupled beads were cross-linked as previously described [22, 24]. One vial of antibody coupled beads was

washed with cold cross-linking buffer (2000 g, 1 min). Subsequently the beads were incubated in 1 mL cross-linking buffer at room temperature for 30 min under gentle agitation. The reaction

was stopped by washing twice with 1 mL cold quenching buffer (200 mM ethanolamine, pH 8.0) and incubating for 2 h in quenching buffer at room temperature under gentle agitation. Cross-linked

beads were washed three times with 1 mL cold IAP buffer and directly used for peptide enrichment. For DiGly enrichment, 3 mg of peptide is used with 1/8 of a vial of cross-linked antibody

beads. Peptides are added to the cross-linked beads and the volume is adjusted to 1 mL with IAP buffer and incubated for 2 h at 4 °C under gentle agitation. The beads are washed twice with

cold IAP buffer and twice ddH2O via centrifugation. The enriched peptides were eluted by adding 200 µL 0.2% v/v TFA onto the beads, incubating 5 min at 1400 rpm and centrifuging for 1 min at

100 g. The supernatant was then transferred to SDB-RPS StageTips and the peptides washed, eluted and dried as previously described for total proteome samples.

Liquid chromatography of total proteome and ubiquitome samples was performed on an EASYnLCTM 1200 (Thermo) with a constant flow rate of 10 µL/min and a binary buffer system consisting of

buffer A (0.1% v/v formic acid) and buffer B (80% v/v acetonitrile, 0.1% v/v formic acid) at 60 °C. The column was in-house made, 50 cm long with 75 µm inner diameter and packed with C18

ReproSil (Dr. Maisch GmbH, 1.9 µm). The elution gradient for the ubiquitome started at 5% buffer B, increasing to 25% after 73 min, 50% after 105 min and 95% after 110 min. The gradient for

the proteome started at 5% buffer B and increased to 20% after 30 min, further increased at a rate of 1% per minute to 29%, following up to 45% after 45 min and to 95% after 50 min. The MS

was performed on a Exploris480 with injection of 500 ng peptide (Thermo). Fragmented ions were analyzed in Data Independent Acquisition (DIA) mode with 66 isolation windows of variable

sizes. The scan range was 300–1650 m/z with a scan time of 120 min, an Orbitrap resolution of 120,000 and a maximum injection time of 54 ms. MS2 scans were performed with a higher-energy

collisional dissociation (HCD) of 30% at a resolution of 15,000 and a maximum injection time of 22 ms. The MS measurement of the proteome was performed equivalently, yet including HighField

Asymmetric Waveform Ion Mobility (FAIMS) with a correction voltage of −50 V and a scan time of 60 min. The injection time for the full scan was 45 s and the MS2 injection time was set to 22 

s.

DIA raw files were processed using Spectronaut (13.12.200217.43655) in directDIA mode. The FASTA files used for the search were: Uniprot Homo sapiens (29.03.2022) with 20609 entries, Uniprot

Homo sapiens isoforms (29.03.2022) with 77157 entries and MaxQuant Contaminants for filtering: 245 entries. Analysis was performed via Perseus (version 1.6.2.3). For the ubiquitome samples

the output was converted with the plugin “Peptide Collapse” (version 1.4.2). Values were log2-transformed and missing values were replaced by imputation from normal distribution with a width

of 0.3 and downshift 1.8 separately for each sample (For full data please see Supplementary Table 3). Ubiquitome was normalized to total proteome. Comparison between conditions of proteome

and ubiquitome was performed via Student’s T-test in R 4.2.2 (P value cutoff 0.5).

All data were analyzed with Microsoft Excel, GraphPad Prism, and R (4.2.2). Data were visualized in GraphPad Prism and shown as mean ± standard deviation (SD) with plotting individual data

distribution of technical replicates. Except for the proteomics, every experiment was replicated twice. Sample size was not based on power calculations. 1-way ANOVA with Dunnett’s Post-hoc

Test was used when comparing three or more groups and one variable, and 2-way ANOVA followed by Tukey’s Test was used for comparing two groups with two variables. P-values lower than 0.05

were considered significant and are as such indicated in the graphs with an asterisk between groups.

It remains largely unclear how the UPS is involved in the execution of ferroptosis. For two specific reasons this is important to understand: First, it is unclear if the regulation of UPS is

a ferroptosis-specific event or simply a consequence of proteasome dysfunction. Second, as critical regulators of ferroptosis such as GPX4 have been shown to be regulated by UPS [9],

understanding global and site-specific ubiquitylation will potentially aid the discovery of novel key players in ferroptosis. Treatment of EA.hy926 cells with the ferroptosis inducer RSL3

led to higher levels of ubiquitin (Fig. 1a and Supplementary Fig. S4). Interestingly, the effect of RSL3 on ubiquitin levels was somewhat comparable if not higher to treatment of cells with

the established chemical proteasome inhibitor bortezomib (BTZ) (Fig. 1a). We next asked if this effect also triggered the activation of NFE2L1, which an important part of UPS defense against

inhibition of the proteasome. Indeed, treatment with RSL3 led to higher protein levels of the cleaved fragment of NFE2L1 (ca. 95 kDa) and lower levels of the full-length protein in a time-

and dose-dependent fashion (Fig. 1b). This increase in cleaved fragment was associated with higher levels of NFE2L1 in the nucleus (Fig. 1c). Based on native PAGE analysis and measuring

degradation of fluorogenic peptides, RSL3 diminished proteasomal activity (Fig. 1d, e). Hence, we tested the possibility that RSL3 may have off-target effects directly on proteasomal

activity. At concentrations well-above what is used in our cell assays there was no direct inhibition of proteasomal activity in cell lysates, unlike what is seen with regular proteasome

inhibitors. (Fig. 1f). This diminishes the possibility that the RSL3-induced decrease in proteasome activity and increase in ubiquitin levels are consequences of off-target pharmacological

inhibition of the proteasome by direct binding of RSL3.

a Immunoblot of ubiquitin from EA.hy926 cells treated with 5 µM RSL3 and 100 nM bortezomib (BTZ) for 9 h and 3 h, respectively. b Immunoblot of NFE2L1 in EA.hy926 cells treated with

indicated time points and concentrations of RSL3 (FL: full-length form ca. 120 kD; CL: cleaved ca. 95 kD). c Immunoblot of NFE2L1 in nuclear and cytoplasmic fractions isolated from EA.hy926

cells treated as indicated for 3 h. d Native page of EA.hy926 cells treated with 5 µM RSL3 and 100 nM BTZ for 6 h with in-gel activity and immunoblot of the ɑ1-7 (20S) subunits. e

Proteasomal activity in EA.hy926 cells treated with 5 µM RSL3 for 3 h. f Proteasomal activity assay in EA.hy926 cells with the extracts being incubated with indicated concentrations and time

points of RSL3 and proteasome inhibitor MG132. g Volcano Plot of the ubiquitome of EA.hy926 cells treated for 9 h with 5 µM RSL3 Padj