Activation of lysosomal iron triggers ferroptosis in cancer

Iron catalyses the oxidation of lipids in biological membranes and promotes a form of cell death referred to as ferroptosis1–3. Identifying where this chemistry takes place in the cell can inform the design of drugs capable of inducing or inhibiting ferroptosis in various disease-relevant settings. Whereas genetic approaches have revealed underlying mechanisms of lipid peroxide detoxification1,4,5, small molecules can provide unparalleled spatiotemporal control of the chemistry at work6. Here, we show that the ferroptosis inhibitor liproxstatin-1 (Lip-1) exerts a protective activity by inactivating iron in lysosomes. Based on this, we designed the bifunctional compound fentomycin that targets phospholipids at the plasma membrane and activates iron in lysosomes upon endocytosis, promoting oxidative degradation of phospholipids and ferroptosis. Fentomycin effectively kills primary sarcoma and pancreatic ductal adenocarcinoma cells. It acts as a lipolysis-targeting chimera (LIPTAC), preferentially targeting iron-rich CD44high cell-subpopulations7,8 associated with the metastatic disease and drug resistance9,10. Furthermore, we demonstrate that fentomycin also depletes CD44high cells in vivo and reduces intranodal tumour growth in an immunocompetent murine model of breast cancer metastasis. These data demonstrate that lysosomal iron triggers ferroptosis and that lysosomal iron redox chemistry can be exploited for therapeutic benefits.


Introduction
Iron reacts with hydrogen peroxide to produce oxygen-centred radicals, initiating a chain reaction that leads to oxidised organic products, a process broadly de ned as the Fenton reaction 11 .Chemically reactive lipids in biological membranes are ideal substrates for such reactions.Accumulation of damaged phospholipids can eventually cause loss of membrane integrity, altered organelle functions and release of components in the cell that can further promote cellular damage and cell death 12 .Ferroptosis has been shown to involve various organelles 13 including peroxisomes 14 , mitochondria 15 , the endoplasmic reticulum (ER) 16 and endolysosomes 17 .However, it is currently unclear whether individual organelles contribute to ferroptosis via altered signalling, metabolism and biosynthesis, or whether membrane lipids of these compartments are also direct substrates for iron-mediated oxidation of membrane lipids leading to cell death [18][19][20][21] .Therefore, where in the cell and the extent to which ironmediated oxidation of membrane lipids promotes ferroptosis remain elusive.

Lysosomal iron triggers ferroptosis
We set out to determine the subcellular sites of action of liproxstatin-1 (Lip-1), a small molecule that protects cells against ferroptosis induced by inactivation of oxidised lipid-detoxi cation systems, including the cystine/glutamate antiporter (SLC7A11/SLC3A2)/glutathione (GSH)/glutathione peroxidase 4 (GPX4) and NAD(P)H/ferroptosis suppressor protein 1 (FSP1)/quinone nodes 22 .In-cell labelling of an alkyne-containing synthetic analogue of Lip-1, we named cLip-1, in HT-1080 brosarcoma cells using click chemistry 6 revealed lysosomal targeting using uorescence microscopy (Fig. 1a and Extended Data Fig. 1a,b).In vivo, cLip-1 delayed death in mice undergoing acute renal failure as a result of genetic deletion of Gpx4 4 .Labelling cLip-1 in mouse tissues revealed its accumulation in the liver and kidney, colocalizing with a lysosomal marker in renal proximal tubules (Fig. 1b,c and Extended Data Fig. 1c).In vitro, cLip-1 prevented oxidation of membrane lipids and protected cells from genetic depletion of Gpx4 or pharmacological inhibition of GPX4 with RSL3, essentially recapitulating the biological activity of Lip-1 (Extended Data Fig. 1d-f).These data validated cLip-1 as a suitable surrogate of Lip-1 to investigate ferroptosis.It is noteworthy that labelled cLip-1 was predominantly found to co-localize with a lysosomal marker at concentrations higher than the lowest effective dose, supporting the notion that Lip-1 exerts an anti-ferroptotic activity speci cally in this organelle.
Iron is predominantly internalised by endocytosis in cancer cells 7,23 , regulating cell-state transitions and proliferation 7,24,25 .The observed accumulation of cLip-1 in lysosomes raised the prospect of ferroptosis inhibition by iron inactivation in this organelle, contrasting with a free radical-trapping activity 26 .Indeed, the iron chelator deferoxamine (DFO) has been shown to protect cells against erastin-induced ferroptosis 2 and Lip-1 contains an iron-chelating o-phenyldiamine core 27 , further supporting this hypothesis.Nuclear magnetic resonance (NMR) spectroscopy and visual inspection indicated that Lip-1 forms complexes with iron(III) that are stable under acidic conditions, such as those found in lysosomes (e.g.pH<5), but which dissociate at higher pH (pH>12) (Fig. 1d and Extended Data Fig. 2a).Cyclic voltammetry indicated that Lip-1 and DFO impair iron redox properties (Fig. 1e).As controls, we investigated two other synthetic analogues of Lip-1, namely metcLip-1 and alcLip-1.In metcLip-1, aromatic amines are methylated, adversely impacting on radical-trapping capacity 26 .In contrast, alcLip-1 is an aliphatic analogue characterised by higher pKa of amines.It is expected to be already protonated at physiological pH and thus to exhibit a reduced propensity to accumulate in lysosomes and to form tight complexes with iron (Extended Data Fig. 2b).Labelling of metcLip-1 in cells revealed a similar staining pattern as that of labelled cLip-1, showing co-localisation with a lysosomal marker, whereas a weaker uorescence of labelled alcLip-1 was detected (Extended Data Fig. 2c).Cyclic voltammetry further indicated that metcLip-1 retained some level of redox inactivation capacity towards iron, whereas alcLip-1 did not exhibit any measurable effect (Extended Data Fig. 2d).Consistently, metcLip-1 protected cells against RSL3-induced oxidation of membrane lipids and cell death, although to a lesser extent than Lip-1.This re ected the reduced capacity of this analogue to inactivate iron, presumably due to steric hindrance of the methyl substituents.In comparison, alcLip-1 was biologically inactive in this context (Extended Data Fig. 2e,f).In-cell labelling of an alkyne-containing derivative of DFO (cDFO) 7,8 revealed nuclear and lysosomal uorescence signals (Extended Data Fig. 2g).Both DFO and Lip-1 induced degradation of the iron storage protein ferritin and iron regulatory protein 2 (IRP2) in primary human pancreatic ductal adenocarcinoma (PDAC) cells, indicating that upon binding to lysosomal iron, these compounds deplete the available cellular iron pool 28 , providing a rationale for their toxicity at high concentrations (Fig. 1f).Treatments with hydroxychloroquine (HCQ) or ba lomycin-A1 (Baf-A1), which raise the lysosomal pH and prevent iron(III) unloading from its endocytic carriers, led to reduced pools of free lysosomal iron(III) and protected cells against RSL3-induced oxidation of membrane lipids.Baf-A1 also protected cells against RSL3-induced death (Fig. 1g,h and Extended Data Fig. 2h).Upon treatment with RSL3 for 1 h, membrane lipid oxidation was predominantly detected in lysosomes according to BODIPY uorescence.By contrast, treatment with RSL3 for 4 h led to a staining indicative of oxidised membrane lipids that co-localised mainly with a uorescently-labelled biological marker of the ER (Fig. 1i and Extended Data Fig. 3a,b).Our data suggest that initiation of the radical chain reaction takes place early in the lysosomal compartment, where redox-active iron(II) can be found 18 .This radical chain reaction then propagates to membrane lipids of other proximal organelles including the ER.In support to this, treating cells with well-established ferroptosis inducers led to a depletion of glutathione (GSH) and an increase of hydroxyl radicals in lysosomes (Extended Data Fig. 3c-f).Together, these data illuminate the central role of lysosomal iron as a ferroptosis trigger.

Development of a lipolysis-targeting chimera
Therapy-resistant cancer cells have been shown to be vulnerable to ferroptosis 18,[29][30][31] .These cells can overexpress the cancer stem cell marker and iron transporter CD44, a membrane glycoprotein associated with tumourigenesis and cancer metastasis 9,[32][33][34] .By doing so, these cells upregulate iron endocytosis to promote the activity of iron-dependent demethylases, enabling speci c transcriptional programs underlying cell-state transitions and acquisition of a drug-tolerant persister cancer cell phenotype 7 .Thus, lysosomal iron has emerged as a tractable druggable target to promote ferroptosis in a cell statedependent manner 10 .With this in mind, and the knowledge that lysosomal iron can trigger oxidation of membrane lipids, we rationally designed a small molecule to target lipids at the plasma membrane, which upon endocytosis activates iron(II) in lysosomes and promotes Fenton-like chemistry, exploiting membrane lipids as substrates for oxidation.To this end, we designed a chimera of the uorescent lipophilic natural product marmycin A and the synthetic White-Chen ligand, which we named fentomycin (Fig. 2a and Extended Data Fig. 4a-e).Marmycin A has been shown to accumulate at the plasma membrane of cells and to be internalised by endocytosis 35 , whereas the White-Chen ligand is commonly used in chemical synthesis to oxidize organic substrates by activating iron(II) [36][37][38] .In the presence of hydrogen peroxide and under mild acidic aqueous conditions, such as those found in lysosomes, the White-Chen iron catalyst is thought to form a reactive iron-oxo intermediate, which, like hydroxyl and hydroperoxyl radicals, is able to abstract a hydrogen atom from organic substrates, including fully saturated and least reactive ones, to produce reactive carbon-centred radicals, leading to oxidation products 39,40 .Fentomycin is therefore reminiscent of bifunctional molecular glues able to induce proximity de novo to manipulate protein function or to induce degradation [41][42][43] .We hypothesised that by using the abundant reactive iron(II) in the drug-tolerant cell state, such a chimera would form an active catalyst in situ susceptible to promote oxidative degradation of proximal membrane lipids in lysosomes, ultimately triggering ferroptosis.
In a cell-free system, fentomycin accelerated the oxidation of a liposome-forming unsaturated phospholipid under experimental conditions comparable to that found in lysosomes including acidic pH, the presence of hydrogen peroxide and a water soluble iron(II) salt (Fig. 2b).The intrinsic uorescence of fentomycin revealed its localisation at the plasma membrane together with CD44, when experiments were conducted at low temperature to reduce endocytic ux (Fig. 2c), recapitulating the photophysical properties of the parental marmycin A. In contrast, at a physiological temperature, fentomycin and marmycin A were found to target the lysosomal compartment, whereas chemical labelling of an alkynecontaining White-Chen (cWhite-Chen) ligand revealed a weak pan-cellular staining (Fig. 2d and Extended Data Fig. 4f).
Fentomycin induced the oxidation of membrane lipids in HT-1080 cells, comparing favourably with wellestablished ferroptosis inducers, as shown by mass spectrometry-based lipidomics (Fig. 2e, Extended Data Fig. 5a and Supplementary Table 1).Furthermore, fentomycin induced oxidation of membrane lipids and altered cell viability in a series of human and murine PDAC cell lines and primary cells, whereas the biological activities of marmycin A and cWhite-Chen ligand were marginal (Extended Data Fig. 5b,c).
Consistent with the oxidation of membrane lipids, sublethal doses of fentomycin led to an increase of the ferroptosis gatekeepers GPX4 and SLC7A11 in HT-1080 cells (Fig. 2f), and membrane lipid oxidation was prevented by the lipophilic antioxidant tocopherol (Toc), the iron chelator deferiprone (Def) and Lip-1 (Fig. 2g,h, Extended Data Fig. 5d and Supplementary Table 2).Fentomycin further induced the production of 4hydroxynonenal (4-HNE) (Fig. 2i), which is characteristic of peroxidation and breakdown of fatty acid chains 44 .Interestingly, 4-HNE can induce cellular damage and its production is a hallmark of ferroptosis 45 .Longer treatment of cells with fentomycin led to the upregulation of hormone sensitive lipase (HSL) and an increase of lysophospholipids and glycerol (Fig. 2j-l and Supplementary Table 3), which suggests that oxidised phospholipids trigger a membrane remodelling response 46 .Finally, fentomycin-induced cell death was antagonised by well-established ferroptosis inhibitors, which included iron chelators and antioxidants, but not by apoptosis or necroptosis inhibitors (Fig. 2m and Extended Data Fig. 5e,f).It is noteworthy that fentomycin exhibits a residual toxicity that ferroptosis inhibitors cannot fully overcome.Together, these data indicate that pharmacological activation of lysosomal iron can trigger ferroptosis, with fentomycin acting as a LIPolysis-TArgeting Chimera (LIPTAC).

Fentomycin induces ferroptosis in cancer
We next evaluated the effect of lysosomal iron activation in disease-relevant models.To this end, we investigated the iron content of primary tumour tissues of distinct cancer types, including human PDAC, various human sarcoma subtypes and a murine model of spontaneous breast cancer metastasis.These were chosen for their refractory nature to standard-of-care treatments and capacity to form metastases, contributing to poor clinical outcomes.Furthermore, these indications have been shown to be vulnerable to ferroptosis 18,20,29,30,47 .
Inductively coupled plasma-mass spectrometry (ICP-MS) showed the total iron content to be higher in cancer compared to adjacent non-cancerous tissues of the same patients (Fig. 3a and Extended Data Fig. 6a), and the cellular iron load was found to be higher in subpopulations of cancer cells overexpressing CD44 (Fig. 3b).Studying cells from freshly dissociated human primary PDAC and sarcoma tissues showed higher levels of redox-active lysosomal iron(II) in the CD44 high subpopulations of cancer cells compared to their CD44 low counterparts (Fig. 3c and Extended Data Fig. 6b).This was consistent with prior ndings that showed that CD44 mediates iron endocytosis in cancer cells acquiring a drug-tolerant persister phenotype 7 .In cells obtained from freshly dissociated human primary tumours, fentomycin induced oxidation and lipolysis of membrane lipids, an effect that was antagonised by ferroptosis inhibitors (Fig. 3d,e, Extended Data Fig. 6c,d and Supplementary Table 4).Remarkably, fentomycin also reduced the number of CD44 high cells in PDAC and undifferentiated pleomorphic sarcoma (UPS) and this was also antagonised by ferroptosis inhibitors (Fig. 3f and Extended Data Fig. 6e,f).In primary PDAC cells and human PDAC-derived organoids, fentomycin exhibited a more pronounced effect on cell viability compared to standard-of-care drugs, including irinotecan, 5-FU and oxaliplatin (Fig. 3g and Extended Data Fig. 6g,h).
Regional lymph node cancer lesions are important predictors of distant metastases and mortality.
Recently, it has been shown that lymph protects metastasizing cells from ferroptosis 48 .ICP-MS showed lower total iron levels in lymph uids compared to blood and serum (Extended Data Fig. 7a).Thus, we evaluated the effect of fentomycin on the viability of cancer cells directly in the lymphatics.To this end, we used the 4T1 immunocompetent murine model of spontaneous triple negative breast cancer metastases in Balb/c mice.Quantifying iron in cells isolated from intranodal 4T1 tumours showed a higher iron load in CD44 high cancer cells compared to the CD44 low subpopulations (Fig. 3h).These CD44 high cells also exhibited a higher iron(II)-redox activity in lysosomes and fentomycin altered cell viability in vitro, which was antagonised by Lip-1 (Extended Data Fig. 7b,c).Treating mice bearing intranodal 4T1 tumours with fentomycin by intranodal administration every-other-day led to an inhibition of tumour growth (Fig. 3i,j and Extended Data Fig. 7d,e).Further analyses of residual tumours indicated that fentomycin induced oxidation of membrane lipids and production of lysophospholipids, exhibiting its activity preferentially against CD44 high over CD44 low cancer cell subpopulations (Fig. 3k,l, Extended Data Fig. 7f,g and Supplementary Table 5).Together, these data support the idea that fentomycin elicits ferroptosis in vivo by exploiting the higher abundance of lysosomal iron(II) in CD44 high cell subpopulations.

Discussion
Ferroptosis is a form of cell death resulting from the uncontrolled oxidation of membrane lipids.Whether this process is enzymatically driven, and where it is triggered in the cell had remained unclear.Iron can react with hydroperoxides to initiate or propagate a radical chain reaction independently of enzymes.The lysosomal compartment is a key regulator of cellular iron homeostasis 7,28,49 .Its acidic nature together with the presence of reactive iron and hydrogen peroxide provide the ideal chemical environment to catalyse the oxidation of membrane phospholipids (Fig. 4).Speci c cancer cell subpopulations upregulate the iron transporter CD44 to promote oxidative demethylation of repressive chromatin marks and unlock the expression of genes involved in cancer progression 7 .Thus, while higher iron levels enable these cells to acquire a drug-tolerant pro le, this also inexorably confers vulnerability to ferroptosis 10 .The fact that CD44 also marks development, self-renewal, wound healing and immune cell activation raises a putative contribution of lysosomal iron in promoting ferroptosis in these biological settings.Enhancing the redox activity of lysosomal iron by means of genetic intervention to eradicate cancer cells is challenging.Here, we have developed a lipolysis-targeting chimera that takes advantage of the higher iron load of speci c cancer cell subpopulations to induce ferroptosis by activating lysosomal iron(II).We demonstrate that manipulating the redox activity of lysosomal iron provides control over membrane lipid oxidation, supporting the contention that lysosomal iron is a trigger of ferroptosis.Fentomycin exhibits a unique chemotype to investigate ferroptosis and provides a new paradigm to target drug-tolerant persister cancer cells 50  Antibodies.Antibodies are annotated below as follows.WB, western blot; FCy, ow cytometry; FM, uorescence microscopy.Hu, used for human samples.Ms, used for mouse samples.Dilutions are indicated.Any antibody validation by manufacturer is indicated and can be found on the manufacturers' websites.Our antibody validation knockdown (KD) and/or KO strategies as described here for relevant antibodies is indicated.Subsequently, the dissociated tumour suspension was applied to a MACS SmartStrainer (30 µm) (Miltenyi).Samples were diluted with 1× PBS (Phosphate-buffered saline) and centrifuged at 300× g.The cell pellet was resuspended in RPMI (10%FBS, penicillin/streptomycin) and cells were counted using an automated cell counter (Entek) Establishment of xenograft derived primary cell cultures (XDPCC).XDPCC models were originally derived from PDX patient models.The PDX fragments designated for cell culture were processed in a biosafety chamber.After ne mincing they were treated with collagenase type V (Sigma-Aldrich, C9263) and trypsin/EDTA (Gibco, 25200-056) and were suspended in Dulbecco's modi ed Eagle's medium supplemented with 1% w/w penicillin/streptomycin and 10% FBS.After centrifugation, cells were resuspended in serum-free ductal media adapted from previous protocols 52 at 37 °C in a 5% CO 2 incubator.Ampli ed cells were stored in liquid nitrogen.Cells were weaned from antibiotics for more than 48 h before testing.This protocol was used to establish the cells designated as PDAC053T, PDAC090T, PDAC211T and PDAC030T.
Chemosensitivity pro ling of XDPO and XDPCC.For chemosensitivity pro ling, XDPO were plated into 96well plates and then subjected to incrementally increasing concentrations of drugs.Cell viability was measured 72 h after treatment using CellTiter-Glo 3D (Promega, G9683).Doubling times (DT) of XDPO viability for untreated control conditions were calculated on days 0 and 3.The ratio of day 3 over day 0 corresponds to the replication rate (RR) of the cells at 72 h.Doubling time was calculated with the formula 72 × 2/RR.Fluorescence and luminescence values were quanti ed using the plate reader Tristar LB941 (Berthold Technologies).Each experiment was performed at least 3 times with at least 3 replicates.
Isolation of blood, serum and lymphatic uid from mice.Balb/C mice (25-week-old adult female mice) were purchased from Charles River and grow in CRCM animal core facility.Mice were housed under sterile conditions with sterilised food and water provided ad libitum and maintained on a 12 h light/dark cycle.Mice were not subjected to any procedures prior to the lymph, blood and serum samples collection.Lymph sample collection:Thirty min before the beginning of the experiment, buprenorphine (Buprecare), an analgesic, was administered by intraperitoneal injection (0.5 mg/kg).Mice were euthanised by intraperitoneal injection of a ketamine/xylazine combination (ketamine 100 mg/kg (Imalgène)/xylazine 10 mg/kg (Rompum); 20 µl/g).After cutaneous and peritoneal incisions, the lymph has been collected in the intestinal lymph trunk 53 with a glass capillary.The lymph collection was placed in cryotubes, frozen at -20 °C and stocked at -80 °C.Blood and serum sample collection: After lymph collection, we performed a terminal cardiac puncture (23G needle with 1 to 2 μL syringe) with thoracotomy to collect a large volume of blood without anticoagulants.100 to 200 µL of blood sample were placed in a vial, frozen at -20 °C and stocked at -80 °C.The rest of the blood collection was left to stand at room temperature for 30 min, then centrifuged at 2000× g for 15 min.The supernatant (serum) was collected in a vial, frozen at -20 °C and stored at -80 °C.
Intranodal murine metastasis models and fentomycin treatment.Murine breast cells (4T1) were transplanted into 6 to 8-week-old female Balb/c mice (syngeneic with the 4T1 model).To perform injections into lymph nodes, the lymphatics were rst traced by injecting 2% Evans Blue dye (Sigma-Aldrich, E2129,) into the foot pedal 5 min before performing intranodal injections.After injecting Evans Blue dye, the mice were anesthetised using iso urane and a small (5-10 mm) incision was made in the region of the right popliteal lymph node.The lymph node was located based on Evans Blue staining, immobilised with forceps, and 20000 cells (Experiment 1) to 10000 cells (Experiments 2 and 3) suspended in 1× PBS were injected in a volume of 10 μL into the popliteal lymph node using a 27 G Hamilton syringe.Injection into the lymph node was con rmed by visible swelling of the lymph node.The incision was closed using surgical glue (3M VetBond Tissue Adhesive, 1469SB,) and the mice were closely monitored for signs of pain or distress.Once tumours were palpable in at least 75% of the mice (~1 week after injection), 10 μL of volume of fentomycin (0.003 mg per animal every-other-day) of vehicle delivered intralymphatically into the tumour-bearing lymph node every other day until the experimental endpoint.Intranodal tumour diameters were measured thrice weekly with calipers until any tumour in the mouse cohort reached 2.0 cm in its largest diameter which was the pre-determined experimental endpoint for these experiments.At that point, all mice in the cohort were killed, per approved protocol, for analysis of intranodal tumour diameter, tumour mass and mouse.Tumour samples were frozen in 10% DMSO in FBS (1 °C/ min until -80 °C) for subsequent cellular analyses.No formal randomisation techniques were used.However, animals were allocated randoμLy to treatment groups and specimens were processed in an arbitrary order.For all experiments, the maximum permitted tumour diameter was 2.0 cm and this limit was not exceeded in any experiment.For all experiments, mice were kept on a normal Chow diet and fed ad-libitum.
Fluorescence microscopy.Cells were plated on coverslips and treated as indicated.BODIPY 665/676 (Thermo Fisher Scienti c, B3932, 10 µM, 45 min), LysoTracker Deep red (Thermo Fisher Scienti c, L12492, 100 nM, 45 min), SQSS (in-house, 50 nM, 24 h) and 1-Red (in-house, 100 nM, 1 h before xation) were added to live cells before xation.For fentomycin and marmycin A treatments, cells were treated at the indicated temperature with 1 µM compound for 1 h.BacMam transduced cells (see transduction section) were treated with BODIPY 665/676 16 h after transduction.Cells were then washed three times with 1× PBS, xed with 2% paraformaldehyde in 1× PBS for 12 min and then washed three times with 1× PBS.For antibody staining, cells were then permeabilised with 0.1% Triton X-100 in 1× PBS for 5 min and washed three times with 1× PBS.Subsequently, cells were blocked in 2% bovine serum albumin/BSA, 0.2% Tween-20/1× PBS (blocking buffer) for 20 min at room temperature.Cells were incubated with the relevant antibody in blocking buffer for 1 h at room temperature, washed three times with 1× PBS and were incubated with secondary antibodies for 1 h.Finally, coverslips were washed three times with 1× PBS and mounted using VECTASHIELD containing DAPI (Vector Laboratories, H-1200-10).BODIPY 665/676 treated cells were xed using ice cold reagents and placed at 4 °C immediately after mounting on cover slips and imaged immediately.Fluorescence images were acquired using a Deltavision real-time microscope (Applied Precision) or a thunder microscope (Leica).40×/1.4NA,60×/1.4NAand 100×/1.4NAobjectives were used for acquisitions and all images were acquired as z-stacks.Images were deconvoluted with SoftWorx (Ratio conservative -15 iterations, Applied Precision) and processed with FIJI 2.0.0-rc-69/1.52n.Images were taken in black and white and colouring was applied with FIJI.Fluorescence intensity is displayed as arbitrary units (AU) and is not comparable between different panels.Colocalisation quanti cation was calculated using FIJI 2.0.0-rc-69/1.52n.Nuclei were detected using DAPI or Hoechst uorescence as indicated.

Figure 1
Figures

Figure 2 Development
Figure 2

.
Declarations 4 .Ubellacker, J.M. et al.Lymph protects metastasizing melanoma cells from ferroptosis.Nature 585, 113-118 (2020).49.Rizzollo, F., More, S., Vangheluwe, P. & Agostinis, P. The lysosome as a master regulator of iron metabolism.Trends Biochem.Sci.46,960-975(2021).50.Marine, J.C., Dawson, S.J. & Dawson, M.A. Non-genetic mechanisms of therapeutic resistance in cancer.Nat.Rev.Cancer 20, 743-756 (2020).Fresh tumour samples were obtained from patients undergoing surgery at Institut Curie and Paul Brousse hospitals.All patients provided written informed consent for use of tumour samples.The study was μm 30 × 150 mm).NMR spectroscopy was performed on Bruker 300, 400 or 500 MHz instruments.Spectra were run in methanol-d 4 , dimethylsulfoxide-d 6 , methylene chloride-d 2 or chloroformd, at 298 K. 1 H chemical shifts δ are expressed in ppm using the residual non-deuterated solvent as internal standard and the coupling constants J are speci ed in Hz.The following abbreviations are used: bs, broad singlet; s, singlet; d, doublet; dd, doublet of doublets; ddd, doublet of doublet of doublets; dt, doublet of triplets; dq, doublet of quadruplets; q, quadruplet; t, triplet; quint., quintet; m, multiplet. 13mical shifts δ are expressed in ppm using the residual non-deuterated solvent as internal standard.The purity of nal compounds was determined to be >98% by UPLC-MS.Low-resolution mass spectra (LRMS) were recorded on a Waters Acquity H-class equipped with a Photodiode array detector and SQ Detector 2 (UPLC-MS) tted with a reverse phase column (Acquity UPLC BEH C18 1.7 μm, 2.1x50 mm).HRMS were recorded on a Thermo Scienti c Q-Exactive Plus equipped with a Robotic TriVersa NanoMate Advion.Procedures for the synthesis of small molecules are detailed in the Supplementary Information.For all experiments we used a 0.1 M nBu 4 NBF 4 in MeCN (32.9 mg/μL stock solution).1mMFeCl3solutionswere prepared with 50 µL of 20 mM FeCl 3 solution in milliQ water and 950 µL of MeCN.Then, portions of 10 µL (0.2 eq.) of 20 mM stock solution of Lip-1 or analogues (solubilised in MeCN or MeOH) were addeduntil 1.0 eq. was reached.Above 1.0 eq., 50 µL (1.0 eq.) of 20 mM stock solution of the analogues were added.After each addition, the solution was stirred for a few seconds and voltammograms were recorded.Liposome preparation.Liposomal structures were prepared using the traditional lipid lm hydration method: 100 μL of a stock solution (1mg/μL chloroform) of 18:1 (Δ9-cis) PC (DOPC, Avanti Polar Lipids)were dissolved in 400 μL of chloroform and transferred into a round-bottom ask.The organic solvent was removed under reduced pressure in a rotary evaporator for 15 min at 200 rpm at 37 ºC in a water bath.Afterwards, the lipid lm was dried with a vacuum pump overnight.Then was hydrated with 1 μL of 0.1 mM sodium acetate buffer (pH 4.5) and vortexed every 5 min for 20 min.Liposomes were extruded by passing the suspension through 2 polycarbonate membranes (pore size 0.2 mm) 20 times.Lipid oxidation in vitro.Control experiment: 200 μL of the liposome solution were added into an Eppendorf tube and heated at 37 ºC with agitation at 800 rpm.Then 5 μL of an aq.solution of Fe(OTf) 2(1.4 mg/1.5 μL) and 13 μL of 0.1 mM acetate buffer (pH 4.5) were added.At t = 0 min 13 μL of an aq.solution of H 2 O 2 (10 μL H 2 O 2 (30%)/1 μL) were added.Fentomycin experiment: 200 μL of the liposome solution were added into an Eppendorf tube and heated at 37 ºC with 800 rpm.Then 13 μL of a solution 1 mM of fentomycin in DMSO and 5 μL of an aq.solution of Fe(OTf) 2 (1.4 mg/1.5 μL) were added.At t = 0 min 13 μL of an aq.solution of H 2 O 2 (10 μL H 2 O 2 (30%)/1 μL) were added.The kinetic process of DOPC oxidation was recorded with a QExactive mass spectrometer (Thermo Fisher Scienti c) equipped with a TriVersa NanoMate ion source (Advion Biosciences).Samples were injected at 0.5 h, 1 h, 2 h, 3 h, 4 h, 7 h and 24 h reaction time.
approved by institutional regulatory boards (DATA190160).All cLip-1 in vivo experiments were performed in compliance with the German Animal Welfare Law and have been approved by the Institutional Committee on Animal Experimentation and the Government of Upper Bavaria (approved no.ROB-55.2-2532.Vet_02-18-13).All intranodal injection mouse experiments complied with all relevant ethical regulations and were performed according to protocols approved by the Institutional Animal Care and Use Committee at Harvard T.H. Chan School of Public Health (protocol IS00003460).For mouse lymph and blood collection, animal experiments were performed in accordance with the European Community guiding in the care and use of animals.Animal experiments were performed in agreement with the French Guidelines for animal handling and approved by local ethics committee (Agreement no.16487-2018082108541206 v3).Lead contact.Further information and requests for resources and reagents should be directed to the lead contact Raphaël Rodriguez (raphael.rodriguez@curie.fr).Materials availability.Please contact the lead author Raphaël Rodriguez for in-house reagents, which can be made available under a material transfer agreement with Institut Curie.Data availability.Lipidomics data are presented in Supplementary Tables1-5.potassiumpermanganatesolutions and heating.Reaction products were puri ed by ash column chromatography on silica gel 60 (230-400 mesh, Macherey Nagel) or aluminium oxide (activated neutral, Sigma-Aldrich), by Combi ash Rf, or by preparative HPLC Quaternary Gradient 2545 equipped with a Photodiode Array detector (Waters) tted with a reverse phase column (XBridge BEH C18 OBD Prep column 5 NMR of Lip-1-iron(III) complexes: 1 H NMR spectra were recorded on a 500 MHz Bruker spectrometer at 310 K, and chemical shifts δ are expressed in ppm using the residual non-deuterated solvent signals as internal standard.General procedure: From 0 to 1 equivalent of FeCl 3 (Alfa Aesar, 12357, lot E23Z042) , portions of 0.5 equivalent of a solution of FeCl 3 in methanol-d 4 were added up to a solution of 1 equivalent of liproxstatin-1 (Lip-1, Sigma-Aldrich, SμL1414, lot 0000152075) in methanol-d 4 into an NMR tube and NMR spectra were recorded.Then a drop of tri uoroacetic acid (TFA, Sigma-Aldrich, T6508) or sodium deuteroxide (NaOD, Eurisotop, D076Y) was added.0.94 mg of Lip-1 were dissolved in 600 µL of methanol-d 4 .From 0 to 1 equivalent of FeCl 3 , portions of 3.0 µL of 92 mM solution of FeCl 3 were added.Cyclic voltammetry.Cyclic voltammetry51experiments were performed with a three-electrode cell.A saturated calomel electrode (SCE) was used as reference, a steady glassy carbon (GC) electrode of diameter 3 mm was selected as working electrode and a platinum wire as counter-electrode.All cyclic voltammograms were recorded at room temperature with a μ-autolab III from Metrohm using Nova software with a scan rate of 2 V/s.MeCN and MeOH were used in HPLC grade (Carlo Erba).3 or Lip-1 were dissolved in water for DFO and FeCl 3 , and MeCN for Lip-1.For dilution of each compound, 15 µL of these stock solutions were added to 135 µL of MeCN to reach a concentration of 10 mM in each vial.For mixture of compounds, 15 µL of each compound were added to 120 µL of MeCN to reach a concentration of 10 mM.miliQ water and HPLC grade MeCN (Carlo Erba) were used.