p21 can be a barrier to ferroptosis independent of p53

The directionality of regulation of ferroptosis by p53 is highly context specific

We analyzed the response of several human cancer cell lines to the ferroptosis inducer erastin that belongs to the class I ferroptosis inducers (FINs) [41] and categorized them based on the degree of response (Figure 1A). In line with previous reports, even cell lines having the same tissue of origin varied in their response to ferroptosis [42]. The main aim of the current study was to identify if p53 or its targets could be responsible for the dichotomy between at least some of the resistant and sensitive cells. Although we had previously surmised that p53 status was not always predictive of the ferroptosis sensitivity of a given cancer cell line [19], we wanted to determine if the loss of p53 in a given cancer type would then alter its sensitivity to ferroptosis.

We chose two colon cancer cell lines with varying ferroptosis sensitivities- RKO and HCT-116 (Figure 1A) for which isogenic derivatives with respect to their p53 status were already available. These isogenic cell lines were created by the deletion of a functional domain of p53 [43]. In both cell lines, the loss of p53 made them less sensitive to the chemotherapeutic doxorubicin (Left panels of Figure 1B, 1C), which is thought to elicit at least part of its effects on cancer cell survival through the activation of p53 [44]. On the other hand, the loss of p53 only slightly decreased the ferroptosis sensitivity of HCT-116 cells, while the RKO cells actually became more sensitive upon the loss of p53 (Right panels of Figure 1B, 1C). These results highlight the complexity in defining a set direction of regulation of ferroptosis by p53. Our findings are in line with the current literature in the field showing that p53 can either promote or block ferroptosis [5].

p21 is differentially regulated between cells that are sensitive and resistant in response to ferroptosis

We reasoned that the nuanced roles of p53 in ferroptosis might be indirect and perhaps based on one or more p53 targets being activated in response to ferroptosis induction. To this end, we sought to examine the protein levels of p21, as it is one of the key downstream targets of p53. In fact, one key difference between the HCT-116 and RKO cells used above was their relative p21 protein abundance (Figure 1D).

We found that upon the induction of ferroptosis using two class 1 FINs (erastin or IKE), three different ferroptosis-sensitive cell lines (HT-1080, SK-HEP1 and U2OS) showed decreased levels of p21 protein (as well as p53) as a function of erastin concentration (Figure 2A–2C). On the other hand, there was an increase in p21 protein levels in two ferroptosis-resistant cell lines (HCT-116, H1299) (Figure 2D, 2E). This increase in the levels of p21 was p53-independent since it was observed even in the p53-null H1299 cell line and in HCT-116 cells that were engineered to lose p53 (p53 KO HCT116).

We also evaluated the effect of FINs on the p21 protein levels of p53 KO derivatives of the sensitive cells, HT-1080 and SK-HEP1. As we had reported previously, the loss of p53 impairs the ferroptosis-response of these cells to some extent [19]. While the HT-1080 p53 KO cells still fall within the ferroptosis-sensitive category defined in Figure 1A, the increase in ferroptosis-resistance caused by the loss of p53, places SK-HEP1 p53 KO cells on the upper edge of the moderate class. Accordingly, p21 levels were decreased in the ferroptosis-sensitive HT-1080 p53 KO upon treatment with FINs, while they were enhanced in the SK-HEP1 p53 KO cells, which moderately resist ferroptosis (Supplementary Figure 1). These results further support that the resistance to ferroptosis and the ability of the cell line to promote FIN-dependent augmentation of p21 protein levels are linked independent of the p53 status.

We then wanted to understand the nature of regulation of p21 upon ferroptosis induction. To this end we compared ferroptosis-sensitive HT-1080 (p53 wild-type) cells and ferroptosis-resistant H1299 and HCT116 cells. To our surprise, we found that in both sensitive and resistant cells, p21 mRNA expression was upregulated at the mRNA level (Figure 3). As controls, increases in the levels of chac1 and ptgs2, known to be induced during ferroptosis [6], were documented as well. Note that there was not a universal reduction in protein levels upon ferroptosis induction as evidenced by constant levels of our loading control, as well as the additional control of expected increase in levels of ferritin in ferroptosis [45]. This result indicates that the process of ferroptosis induces p21 gene expression in a p53-independent manner and that the subsequent loss of p21 protein in the sensitive cells is most likely a consequence of a post-transcriptional event. It also suggests that this differential regulation of p21 protein may then determine the extent of death achieved.

Altering p21 protein levels changes the sensitivity of cells to ferroptosis

The above results indicated a potential role for p21 in determining the sensitivity of cells to ferroptosis. To validate this hypothesis, we experimentally altered p21 levels and examined the changes in ferroptosis sensitivity of both resistant and sensitive cells.

In the resistant cell lines, HCT-116 and H1299, our goal was to determine if ferroptosis resistance can be lowered upon loss of p21. We used RNA interference against p21 in these resistant cells and indeed observed a reduction in the resistance to ferroptosis (Figure 4A, 4B). We tested the possibility that a more complete and non-transient loss of p21 might be required to further enhance the sensitivity of these cells, as it was reported that p21 can alter the metabolic pathways involved in ferroptosis [35]. Indeed, the HCT-116 derived p21 -/- cell line [46], had a much-enhanced sensitivity to ferroptosis compared to its wild-type counterpart (Figure 4C).

As a reciprocal approach we increased p21 levels in the HT-1080 cell line that is ferroptosis-sensitive. Overexpression of a construct expressing wild-type p21 did suppress this form of cell death in the HT-1080 cells (Figure 4D).

Thus, the results in Figure 4A–4D demonstrate that the capacity of cells to regulate p21 when treated with FINs crucially impacts the extent of response to ferroptosis. However, it is unclear if the ferroptosis-associated reduction of p21 in sensitive cells takes place as a consequence of translational or post-translational signals. The results in Figure 4D suggests that it is unlikely for the loss of p21 protein abundance to be a result of enhanced degradation of p21, as then the ectopic p21 protein would have also faced the same fate and should then have been rendered incapable of altering the degree of ferroptotic death. In further support of this hypothesis, we observed that proteasome inhibitor MG132 is unable to block any part of the reduction in p21 expression caused by erastin (Supplementary Figure 2A left panel). Note that MG132 treatment did slightly increase the sensitivity of ferroptosis, although this was found to be p21 independent (Supplementary Figure 2A right panel). Thus, based on these results we posit that p21 is more likely to be translationally regulated in response to ferroptosis.

The experiment in Figure 4D, further allowed us to determine which interactions of p21 may aid its role in ferroptosis. For this, we used two key mutant versions of p21, which disable either its CDK binding or PCNA binding domains [47, 48] and found that they differed in their ability to suppress ferroptosis. Specifically, the CDK binding-defective version of p21 was unable to block ferroptosis, while mutating the PCNA binding region impaired p21 to a much lesser extent in that regard (although the levels of expression of this mutant were slightly lower). This result suggests the possibility that p21 alters sensitivity to ferroptosis by affecting CDK-mediated functions.

Taken together, we demonstrate that differential regulation of p21 protein can serve as a determining factor of ferroptosis sensitivity in many human cancer cells and that this regulation is independent of p53. Our results further emphasize the importance of p53-targets in the regulation of cell survival, even in the absence of p53.

Cells

HCT116, H1299, SK-HEP1, and U2OS cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% heat-inactivated fetal bovine serum (Gemini Bioproducts, cat# 900-108). HT-1080 cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% heat-inactivated fetal bovine serum (Gemini Bioproducts, cat# 900-108), and 1% non-essential amino acids (Sigma-Aldrich, cat# M7145). RKO cells were grown in McCoy’s 5A modified medium (Gibco, cat# 16600-082) supplemented with 10% heat-inactivated fetal bovine serum (Gemini Bioproducts, cat# 900-108). The HCT116 and RKO isogenic cell lines that were created by the deletion of a functional domain of p53 [43] were a gift from Dr. Vogelstein. The HT-1080 p53 KO, HT-1080 p21 KO and SK-HEP1 p53 KO cell lines were genetically engineered using CRISPR technology [19]. All other cell lines were obtained from ATCC.

Drugs and chemicals

The commercially available compounds used were: erastin (Selleckchem, cat# S7242), doxorubicin (Sigma-Aldrich, cat#D1515) and MG132 (Selleckchem, cat# S2619). IKE (imidazole ketone erastin) was synthesized as in Larraufie MH et al., by Yan Zhang in Stockwell lab [51].

All the compounds were dissolved in DMSO (Sigma-Aldrich, cat# D8418).

Quantitative reverse transcription PCR

RNA was isolated from cells using the Qiagen RNeasy minikit. cDNA was generated using the Qiagen Quantitect reverse transcription kit with 0.5 μg of input RNA as measured with a NanoDrop (Thermo Scientific). Real-time PCR was carried out on an ABI StepOne Plus machine using the power SYBR Green dye (Thermo Scientific). Transcript levels were assayed in triplicate and normalized to L32 mRNA levels. Relative changes in cDNA levels were calculated using the comparative Ct method (ΔΔC_T method).

Primer sequences

L32 F: TTCCTGGTCCACAACGTCAAG, L32 R: TGTGAGCGATCTCGGCAC; p21 F: GGCGGCAGACCAGCATGACAGATT, p21 R: GCAGGGGGCGGCCAGGGTAT; chac1 F: GAACCCTGGTTACCTGGG, chac1 R: CGCAGCAAGTATTCAGGTGT; ptgs2 F: TAAGTGCGATTGTACCCGGAC, ptgs2 R: TCTCCAAAGGAGGTTACCTGC.

The p53 primer was obtained as premixed solution from Qiagen (Quantitech primer, HS_TP53_1_SG, cat# QT00060235) and the rest were individually ordered from Invitrogen.

Immunoblot

Cells were lysed with TEB lysis buffer (10mM Tris HCL pH 7.5-8, 137 mM sodium chloride, 10% glycerol, 1% NP-40) supplemented with 1mM magnesium chloride, 1mM calcium chloride and protease inhibitors (Roche). Protein concentrations were assayed using the Bio-Rad protein assay dye reagent and results were read using a spectrophotometer.

Protein extracts were run on in-house made Tris-Glycine SDS Polyacrylamide gels. Proteins were then electro-transferred at 360 mA for 70 min onto a nitrocellulose or PVDF membrane. Membranes were blocked with 5% milk in PBST (Phosphate-Buffered Saline with Tween) for 30 min, prior to being incubated overnight with primary antibodies (1:100-1:1000 dilution according to the specific antibody). The membranes were then washed three times with PBST and incubated with secondary antibody (1:5000 dilution) for 1 hour at room temperature. After three more washes with PBST, the membranes were imaged using ECL (Thermo Fisher, Pierce, cat# 32106 or EMD Millipore, Immobilon, cat# WBKLS0050). The primary and secondary antibodies were diluted with 1% milk in PBST.

The following primary antibodies were used: p53 (mAb 1801/mAb DO.1, in-house produced); p21 (C-19, Santa Cruz biotech, cat# sc-397); actin (Sigma-Aldrich, cat# A2066); ferritin/FTH1 (Cell Signaling Technology cat# 3998). Actin was used as loading control for all the blots.

Transfection: RNA interference

siRNA (15 nM) was used for each well in a 6-well plate. Lipofectamine RNAiMAX (Thermo Scientific) was used as the transfection reagent for all siRNA experiments (according to the manufacturer's instructions). After 18 hours, the media was changed and cells were treated with drugs 24 hours post transfection. Cells were plated prior to transfection such that they were only 80% confluent by the end of the drug treatment period.

The following siRNAs were used: siLuciferase [52], sip21 #1 (HS_CDKN1A_6 Flexitube siRNA from Qiagen), sip21 #2 (HS_CDKN1A_7 Flexitube siRNA from Qiagen).

Transfection: Ectopic expression of proteins

Plasmids were transfected into cells using Lipofectamine 3000 (Thermo Scientific) according to the manufacturer's instructions, with a ratio of 1 μg:1.7 μl lipofectamine reagent. After 18 hours, the media was changed and cells were treated with drugs 24 hours post transfection. The cells were plated prior to transfection such that they are only a maximum of 80% confluent by the end of the drug treatment period. The plasmids for full length and mutants of p21 were a kind gift from Dr. Vanessa Gottifredi and have been previously described [47, 48].

Note

Cells became more resistant to ferroptosis inducers post transfection. In order to obtain cell death post transfection, three key factors need to be controlled: cell density must be lower than normal, lipofectamine reagent needs to be washed off as soon as possible, and a three to four-fold higher dose of FINs must be used to induce ferroptosis.

Cell viability assay

For the dose response curves, 1800 cells were plated in 36 μl per well of a 384 well plate on day one. Drugs were dissolved in DMSO and a 12 point, two-fold series was prepared. The drugs were then dissolved 1:33 in media and 4 μl was added to each well of the plates on day two. After 24-48 hours of drug treatment (based on the cell line), the viability of cells was measured using a 1:1 dilution of the CellTiter-Glo Luminescent reagent (Promega, cat# G7573) with media, which was read on a Victor 5 plate reader after 10 minutes shaking at room temperature. The intensity of luminescence was normalized to that of the DMSO control. Experiments were performed twice in duplicates each time.

For viability assays when the experiment was performed in 6-well plates, cells were harvested using trypsin (0.5 ml per well) and the media was saved from each well. The trypsinized cells were resuspended with the saved media and 2-3 aliquots (0.05 ml each) sampling different regions of this suspension were transferred into a 96-well plate to serve as technical replicates for the measurement. CellTiter-Glo Luminescent Viability assay was used to measure the viability of these aliquots. The rest of the cultures were used to extract protein to be analyzed using western blots.

Statistical analysis

Prism (version 8, GraphPad) was used to make all the graphs in the paper and for performing all the statistical analysis shown. The GraphPad style (0.1234(ns), <0.0332(*), < 0.0021(**), <0.0002(***), <0.0001(****),) was used to represent the p values. The p values were calculated by ANOVA and appropriate multiple testing correction was done where required.