Abstract

   The polynucleotide kinase/phosphatase (PNKP) is a DNA repair enzyme
   possessing bifunctional DNA 3′-phosphatase and DNA 5′-kinase
   activities. It plays an important role in the rejoining of single- and
   double-strand DNA breaks and is considered as a potential therapeutic
   target for different cancer types. Here we show that PNKP is highly
   expressed in triple negative breast cancer (TNBC) and associated with
   poor prognosis and chemoresistance. Targeting of PNKP enhanced
   ferroptosis in TNBC, which was associated with increased labile iron
   pool and ROS and concomitantly decreased in intracellular glutathione,
   SCD1 and GPX4 levels. Transcriptomic profiling and mechanistic data
   indicate that PNKP targeting robustly enhances the lysosomal and the
   autophagic machinery by activating STING and concurrently inhibiting
   STAT3, thereby increasing ferritinophagy, intracellular iron level and
   modulating the expression of key ferroptosis regulators. Importantly,
   PNKP and STAT3 are rapidly phosphorylated, colocalize, and interact
   upon ferroptosis induction or doxorubicin treatment, the first line
   treatment for TNBC patients. Targeting PNKP together with doxorubicin
   synergistically inhibited the growth of TNBC in an animal model and of
   TNBC-patients derived organoids. These results offer a promising
   therapeutic combination for TNBC and highlight the clinical potential
   of PNKP targeting and ferroptotic death for TNBC therapy.

   Keywords: Ferroptosis, TNBC therapy, PNKP, STING, STAT3

1. Introduction

   Despite the recent progress in targeted- and immuno-therapy,
   chemotherapy remains the mainstay treatment for triple negative breast
   cancer (TNBC), a highly aggressive and heterogenous breast cancer
   subtype [[37]1]. TNBC is defined by the absence of estrogen and
   progesterone receptors and of HER2 amplification and is characterized
   by poor prognosis and high metastatic rate. Standard chemotherapeutic
   agents such as taxanes (mitotic inhibitors) and anthracyclines (DNA
   intercalators) are typically applied either in neoadjuvant or adjuvant
   settings [[38]2]. Although a subset of TNBC patients can benefit from
   these chemotherapy treatments, approximately 30–50 % of the patients
   evolve chemoresistance and overall poor survival [[39]3,[40]4],
   highlighting the clinical need of new therapeutic strategies and potent
   combination therapies.

   Doxorubicin, an anthracycline antibiotic, is possibly the most common
   anti-cancer chemotherapeutic agent used against a wide range of human
   cancers, including TNBC [[41]5]. It is a quinone compound- implicated
   in reactive oxygen species (ROS) production, which can damage -
   cellular membranes, DNA and proteins. In addition, it has the capacity
   to intercalate into the DNA and interfere with topoisomerase II
   activity [[42]6]. These activities not only damage cancer cells but are
   also associated with dosage-dependent cardiotoxicity and heart failure
   [[43]7]. Extensive studies on doxorubicin-induced cancer cytotoxicity
   or cardiotoxicity suggest that doxorubicin may trigger various forms of
   regulated cell death (RCD), depending on the cell type, on the applied
   dose, and on the exposure time [[44]8]. It appears that in addition to
   apoptotic cell death, doxorubicin can induce pyroptosis, necroptosis,
   and ferroptosis [[45]9].

   Ferroptosis is an RCD pathway driven by iron-dependent lipid
   peroxidation and severe damage of cellular membranes [[46]10]. ROS,
   glutathione (GSH), labile iron pool (LIP) and polyunsaturated fatty
   acids (PUFAs) are fundamental for lipid peroxidation and ferroptosis
   execution [[47]11,[48]12]. Lipid peroxides are commonly metabolized
   into reactive aldehydes, such as 4-hydroxynonenal (4-HNE) and
   malondialdehyde (MDA), which react with cellular proteins and DNA to
   generate a variety of covalent adducts. The DNA adducts are associated
   with oxidative DNA damage, production of 8-hydroxydeoxyguanosine
   (8-OHdG) [[49]13], and a variety of DNA lesions that engage the DNA
   damage response [[50]13,[51]14].

   The major ferroptosis protective pathway in mammalian cells is the
   cyst(e)ine-GSH-GPX4 (Glutathione peroxidase 4) axis [[52]15]. GPX4, a
   key ferroptosis protector, utilizes GSH as a cofactor to reduce lipid
   hydroperoxides, while GSH synthesis requires cystine import via the
   cystine/glutamate System Xc-antiporter. Two other key defense axes are
   the coenzyme Q10 (CoQ10)-FSP1(Ferroptosis suppressor protein 1)
   [[53]16] and the tetrahydrobiopterin (BH4)-GCH1 (GTP cyclohydrolase-1)
   [[54]17], which suppress lipid peroxidation independent of GPX4
   activity. On the other hand, lipoxygenases (LOXs), which catalyze the
   dioxygenation of PUFAs and generate phospholipid peroxides (PLOOH)
   [[55]18], ACSL4 (acyl-CoA synthetase long chain family member 4) and
   LPCAT3 (lysophosphatidylcholine acyltransferase 3), which are required
   for production of arachidonic acid (AA) and adrenic acid
   (AdA)-containing phosphatidylethanolamine (PE), the primary targets of
   lipid peroxidation, are all promoting ferroptosis execution.

   Growing evidence suggests that ferroptosis vulnerability can be used
   for cancer therapy [[56]19], and previously we showed that TNBCs are
   highly vulnerable to ferroptosis [[57]20,[58]21]. To exploit the
   ferroptotic susceptibility of TNBC, we developed a powerful
   computational pipeline to predict ferroptosis repressors as potential
   targets for TNBC therapy and identified PNKP as one of the top
   candidates [[59]22]. PNKP is a DNA repair enzyme exhibiting dual
   phosphatase/kinase activities, which are required for repair of single-
   and double-strand breaks (SSBs and DSBs) by non-homologous end joining
   [[60]23]. It is present both in the nucleus and in the mitochondria and
   involved in repair of oxidative DNA damage generated by ROS, ionizing
   radiation and certain chemotherapeutic agents [[61]24,[62]25].

   Here we show that PNKP acts as a potent repressor of ferroptosis in
   TNBC. We identified the PNKP-STING-STAT3 axis as a regulator of
   ferroptotic cell death by converging the autophagic-lysosomal machinery
   with the DNA damage response. Our findings show that PNKP depletion
   simultaneously modulates the activities of cGAS-STING and STAT3 in
   opposing directions, potentially amplifying their individual effects
   through feedback loops. We also found that PNKP is a powerful
   therapeutic target for TNBC in combination with doxorubicin, a commonly
   used chemotherapeutic drug for TNBC patients, and propose that this
   synergistic combination could be a powerful approach to increase
   doxorubicin efficacy and possibly reduce its toxicity.

2. Results

2.1. Targeting of PNKP in TNBC induces lipid peroxidation and ferroptotic
death

   Considering the vulnerability of TNBC to ferroptosis, we developed
   different approaches, including a computational pipeline to identify
   potential ferroptosis repressors as candidates targets for TNBC therapy
   [[63]21,[64]22,[65]26]. PNKP was one of the top candidates, and further
   analysis of its expression in the TCGA (The Cancer Genome Atlas)
   datasets revealed a significant higher expression in breast cancer
   patients compared to normal women ([66]Fig. 1A), which was associated
   with poor prognosis in TNBC as demonstrated by the Kaplan–Meier curve
   (data from METABRIC, [67]Fig. 1B).

Fig. 1.

   [68]Fig. 1
   [69]Open in a new tab

   PNKP suppresses ferroptosis in TNBC.

   (A) The expression level of PNKP in non-TNBC and TNBC patients compared
   to healthy women was extracted from the TCGA datasets (n = 114 normal,
   879 non-TNBC, 180 TNBC). (B) Kaplan-Meier curve for TNBC patients,
   stratified by their PNKP expression level (high and low PNKP indicates
   the 10 % patients with the highest and lowest PNKP expression,
   respectively) (n = 33 patients per group). (C–D) Knocking down (KD) of
   PNKP expression by the constitutive (C) or the Tet-inducible (D) shPNKP
   in the indicated TNBC cell lines were assessed by qPCR for PNKP
   transcripts (n = 3 in C, n = 2 in D, ∗∗∗∗p < 0.0001), or by Western
   Blotting (WB) for PNKP protein using anti-PNKP - antibodies. GAPDH was
   used as a loading control. p-values were determined by one-sample
   t-test. (E) KD of PNKP by siRNA (SMART pool) in MDA-MB-468 TNBC cell
   line was assessed by qPCR and compared to control (siCTRL) (n = 2
   experiments in triplicates, ∗∗∗∗p < 0.0001). (F–H) Increased lipid
   peroxidation in PNKP-depleted cells. The indicated TNBC cell lines were
   depleted of PNKP either by siRNA, shRNA, or by the Tet-shRNA as
   indicated. Where indicated,The cells were treated with doxycycline
   (DOXY) (1 μg/ml) for 96 hr, and lipid peroxidation (lipid ROS) was
   assessed by BODIPY-C11 oxidation (F–G) or by WB to detect 4-HNE protein
   adducts (H). Representative confocal images of BODIPY‐C11 fluorescence
   are shown (F). Scale bar, 10 μm. Fluorometric quantification of
   lipid‐ROS (G) was calculated relative to control (0; DMSO) and the mean
   values ± SD from two independent experiments are shown. (I) Cells
   transfected with the indicated siRNA were grown in the absence or
   presence of ferrostatin-1 (3 μM) and cell death was measured by CellTox
   Green. Fluorometric quantification of cell death was calculated
   relative to control, and the mean values ± SD from two independent
   experiments are shown. (J) The indicated TNBC cell lines transfected
   with siCTRL or siPNKP were grown in the absence or presence of
   ferrostatin-1 (3 μM) and cell viability was measured 72 h later by MTT
   assay. Percentage (%) of cell viability was calculated compared to
   siCTRL. Shown are the mean values ± SD (n = 3). (K) MDA-MB-468 cells
   were treated with the indicated concentrations of PNKP inhibitor
   (A12B4C3, SIGMA) and lipid-ROS was assessed using BODIPY-C11 as
   described in G. Shown are the mean values ± SD from four independent
   experiments. (L) The indicated TNBC cell lines were treated with 20 μM
   of PNKP inhibitor in the presence or absence of a 1:1 mixture of
   liproxstatin/ferrostatin (3 μM) and cell viability was measured 72 hr
   later by MTT assay. Percentage (%) of cell viability was calculated
   compared to DMSO. Shown are the mean values ± SD from at least three
   independent experiments.

   To determine if PNKP indeed suppresses ferroptosis in TNBC, we depleted
   its expression by shRNA (constitutive and Tet-inducible shRNA)
   ([70]Fig. 1C and D) or by siRNA (SMARTpool) ([71]Fig. 1E–[72]S1A) in
   several TNBC cell lines of different molecular subtypes (MDA-MB-468,
   BT549, Hs578, HCC70) and assessed lipid peroxidation by the fluorescent
   sensor BODIPY-C11 ([73]Fig. 1F–G and S1B–F). As seen in the
   representative confocal microscopy images and by the fluorometric
   quantification - - ([74]Fig. 1G–[75]S1C, F), an increase in the
   oxidized form and the corresponding ratio between the oxidized (green)
   and reduced (red) forms were observed in PNKP knockdown (KD) cells.
   These results were further supported by an increase in 4-HNE protein
   adducts as assessed by Western Blot (WB) analysis using an antibody
   against 4-HNE, an aldehydic product of lipid peroxidation ([76]Fig.
   1H). The increased in lipid peroxidation led us to examine the
   influence of PNKP KD on ferroptotic death using cell death ([77]Fig.
   1I) and cell viability ([78]Fig. 1J) assays in the absence or presence
   of the ferroptosis inhibitors ferrostatin and/or liproxstatin. As
   shown, PNKP KD reduced cell viability by ∼40–60 %, while ferrostatin
   and liproxstatin significantly increased the viability of PNKP-depleted
   TNBC cells ([79]Fig. 1J–[80]S1G, S1I). Likewise, depletion of PNKP
   significantly increased cell death as measured by CellTOX™ Green
   cytotoxic assay (Promega), while ferrostatin rescued the death response
   ([81]Fig. 1I). Collectively, these results suggest that depletion of
   PNKP in TNBC cells induces lipid peroxidation and ferroptotic death.

   To further validate the effects of PNKP KD, we investigated the impact
   of A12B4C3, a noncompetitive inhibitor of the phosphatase activity of
   PNKP , on lipid peroxidation ([82]Fig. 1K) and ferroptotic death
   ([83]Fig. 1L). While this inhibitor has been previously reported as a
   potent and selective for PNKP, with an IC[50] of ∼60 nM in several
   human cancer cell lines [[84]27,[85]28], our findings show it to be
   less effective, exhibiting an apparent IC[50] of ∼20 μM ([86]Fig. S1H)
   in multiple TNBC cell lines. Nevertheless, treatment of TNBC cells with
   A12B4C3 (20 μM) for 12 h significantly increased lipid peroxidation as
   measured by BODIPY-C11 fluorescence ([87]Fig. 1K), and enhanced
   ferroptotic cell death as measured by reduced cell viability 72 hr post
   treatment and rescue with ferrostatin and liproxstatin ([88]Fig. 1L).

2.2. Depletion of PNKP modulates ROS, GSH and intracellular iron levels

   To further characterize the influence of PNKP on ferroptosis, we
   examined the effects of PNKP depletion (by siRNA and/or shRNA) on key
   ferroptosis modulators, including ROS, GSH, LIP and major regulatory
   proteins. To monitor intracellular ROS levels, we used the
   nonfluorescent H[2]DCFDA (2′,7′-dichlorodihydrofluorescein diacetate)
   probe, which is cleaved upon oxidation and converted into the highly
   fluorescent DCF (2′,7′-dichlorofluorescein) [[89]29]. As seen in
   [90]Fig. 2A and B, a significant increase in DCF fluorescence was
   obtained in PNKP KD compared to control cells. Since PNKP is also
   localized in the mitochondria ([91]Fig. 2D–[92]S2A), we examined the
   influence of its depletion on mitochondrial ROS using the MitoSOX Red
   Reagent and observed a significant increase as measured by flow
   cytometry ([93]Fig. 2C–[94]S2B). The increase in intracellular and
   mitochondrial ROS was accompanied by an increase in mitochondrial lipid
   peroxidation ([95]Fig. 2E) and by a decrease in intracellular GSH level
   in PNKP depleted cells, as measured by colorimetric assay ([96]Fig.
   2F). We next examined the influence of PNKP depletion on intracellular
   iron level and LIP, using a colorimetric assay ([97]Fig. 2G and H) and
   the FerroOrange probe ([98]Fig. 2I–K, [99]S2C). As shown, PNKP
   depletion significantly increased total iron and LIP in the different
   TNBC cell lines. Furthermore, the iron chelator 2,2′-Bipyridyl
   partially restored cell viability of PNKP KD cells ([100]Fig. 2L),
   consistent with the rescue effects of ferrostatin or liproxstatin.
   These results indicate that PNKP KD elevated ROS levels both in the
   cytosol and in the mitochondria, decreased GSH and increased LIP
   levels.

Fig. 2.

   [101]Fig. 2
   [102]Open in a new tab

   PNKP depletion modulates the levels of intracellular ROS, glutathione,
   iron, and key ferroptosis regulators.

   (A–C) Increased levels of cytosolic (A–B) and mitochondrial (C) ROS in
   PNKP-depleted cells. The indicated TNBC cells expressing the Tet-shPNKP
   were treated with either DMSO or doxycycline (DOXY) (1 μg/ml) for 4–5
   days. Cells were then incubated either with CM-H[2]DCFDA to measure ROS
   as described in Methods or with MitoSOX to measure mitochondrial ROS
   using flow cytometry. Percentage (%) of ROS (A, B) and MitoSOX (C) in
   DOXY-treated cells were calculated relative to DMSO control cells.
   Shown are the mean values from 3 independent experiments done in
   duplicates (A), three experiments (B) and two experiments (C). (D) PNKP
   is localized in the mitochondria of TNBC cells. Shown are
   representative confocal images of the indicated TNBC cells stained with
   MitoTracker (red) and PNKP (green). Scale bar, 5 μm. (E) Increased
   mitochondrial lipid-ROS in BT549 cells depleted of PNKP (siPNKP)
   compared to control (siCTRL) cells was detected by MitoPeDPP staining.
   Shown are representative confocal images. Scale bar, 5 μm. (F–H) PNKP
   KD decreased intracellular levels of GSH (F), while increased total
   iron (G) and labile Fe^+2 levels (H) as measured by colorimetric assays
   described in Methods. Shown are the mean values ± SD relative to
   control from 2 experiments done in duplicates (in F), and 2 or 4
   independent experiments in G and H. (I) Representative confocal images
   of cells stained with the FerroOrange probe. Scale bar, 10 μm. (J–K)
   TNBC cells expressing either the constitutive KD (shPNKP) (J) or an
   inducible Tet-shPNKP (K) were treated with FerroOrange, fluorometric
   quantification was calculated as fold increase relative to control
   cells and the mean values ± SD are shown (n = 3 in J or 2 in K). (L)
   TNBC cells depleted for PNKP were grown in the absence or presence of
   the iron chelator 2,2-BP (8 μM) and 72 hr later, cell viability was
   measured relative to control cells (siCTRL). Shown are the percentage
   (%) of the mean values ± SD from three experiments. (M–N) The levels of
   the ferroptosis-related proteins GPX4, SCD1 and ferritin heavy chain
   (FTH) from cells transfected either with siCTRL/pLKO (−) as control or
   with siPNKP/shPNKP (+) were assessed by WB. GAPDH was used as a loading
   control. The blot marked with “∗” (PNKP) appears also in [103]Fig. 5C,
   as both panels originate from the same Western blot experiment. (O)
   Transcripts of PNKP, GPX4 and SCD1 were measured by real time PCR, the
   mean values ± SD from three independent experiments are shown.

   We then assessed the influence of PNKP on key ferroptosis regulators by
   WB and real time PCR (qPCR). As seen in [104]Fig. 2MN, S2D, PNKP KD
   reduced the protein level of GPX4, the major ferroptosis repressor as
   well as of ferritin heavy chain (FTH) and stearoyl-CoA desaturase-1
   (SCD1), the rate limiting enzyme of monounsaturated fatty acids (MUFAs)
   synthesis. SCD1 markedly contributes to ferroptosis resistance by
   increasing MUFA levels, which exhibit antioxidant properties and
   indirectly influence PUFA levels [[105]30]. The transcripts levels of
   GPX4 and SCD1 were also significantly reduced ([106]Fig. 2O).
   Interestingly, we found a positive correlation between PNKP and GPX4
   expression in all types of breast cancer including TNBC patients
   ([107]Fig. S2E). Collectively, these results suggest that PNKP
   depletion leads to ferroptosis execution by modulation the levels of
   ROS, GSH, LIPs, GPX4, SCD1 and FTH.

2.3. PNKP depletion modulates the transcription of ferroptosis-associated
pathways

   To gain molecular insight into the ferroptotic death induced by PNKP
   KD, we profiled the transcriptomic response of PNKP KD in MDA-MB-468 by
   RNAseq. Differential gene expression (DGE) analysis revealed that 143
   genes were significantly upregulated, while 184 genes were
   downregulated in PNKP-depleted cells compared to control. Many of the
   DEGs are related to ferroptosis ([108]Fig. 3A), including critical
   lipid modifying enzymes, such as LPCAT3 (upregulated) and LPCAT1
   (downregulated), which positively and negatively regulate PUFAs
   synthesis respectively [[109]31]. Pathway enrichment analysis revealed
   upregulation of the IRF3 (Interferon regulatory factor 3)-IRF
   (Interferon)-STING (stimulator of interferon genes) pathway, consistent
   with previous reports [[110]32,[111]33]. Notably, we also observed
   upregulation of lysosomal and autophagy pathways ([112]Fig. 3B), as
   well as oxidative phosphorylation and various lipid metabolism related
   pathways, particularly of phospholipids, including PE. The expression
   of selected genes related to these pathways, such as the lysosomal
   proton pumps ATP6V0E1 and ATP6V1E1 and the sulfotransferases SULT1A1/2,
   were further validated by qPCR ([113]Fig. S3A). Downregulation of
   mitotic and cell cycle pathways, possibly due to accumulation of
   damaged DNA, were detected along with significant reduced levels of MYC
   targets, IL6/STAT3 signaling, selenium network, and doxorubicin
   resistance genes ([114]Fig. 3B). Gene set enrichment analysis plots of
   selected pathways with the corresponding normalized enrichment score
   (NES) are shown in [115]Fig. 3C–[116]S3B.

Fig. 3.

   [117]Fig. 3
   [118]Open in a new tab

   Transcriptomic profile of PNKP depleted cells and activation of the
   autophagy-lysosomal machinery.

   (A–C) Transcriptomic analysis of PNKP-depleted cells. Transcriptomic
   profiles of control (pLKO) and PNKP depleted (shPNKP) MDA-MB-468 cells
   were determined by RNAseq analysis as described in Methods.
   Differential gene expression between samples was quantified by Limma.
   Shown are some significantly differential genes related to ferroptosis
   (∗p-value <0.05). (B) Gene set enrichment analysis (GSEA) was performed
   based on the RNAseq results. Shown are selected pathways with
   significant (p-value <0.05) normalized enrichment score (NES). (C) GSEA
   plots for selected pathways. Normalized enrichment score (NES) and
   p-values are indicated. (D) The levels of LAMP1 in control (siCTRL, -)
   or PNKP depleted (siPNKP, +) TNBC cell lines were assessed by WB. GAPDH
   served as a loading control. (E, F). The levels of LC3B (I-cytosolic,
   II-lipidated, autophagosome) were assessed by WB (E) or by IF (F) in
   control cells or cells that were either treated with the indicated
   concentrations of PNKP inhibitor or depleted for PNKP. Shown are
   representative confocal images of LC3B staining following PNKP deletion
   (F). Scale bar, 5 μm. (G) Representative confocal images of MITF
   localization in control and PNKP depleted BT549 cells. Scale bar,
   10 μm. (H) The levels of PNKP, MITF, TFEB and LAMP1 transcripts in
   control (siCTRL) and PNKP depleted (siPNKP) MDA-MB-468 cells were
   measured by qPCR. The mean values ± SD from two independent experiments
   are shown. (I) Increased co-localization of FTH with lysosomes after
   PNKP depletion. MDA-MB-468 cells transfected with the indicated siRNA
   were co-stained with lysotracker and FTH. Shown are representative of
   confocal images. Scale bar, 5 μm. (J) Correlation between PNKP
   expression and the enrichment of the autophagy pathway (measured by
   GSVA) in BC patients (n = 1904) and TNBC patients (n = 299) from
   MetaBric data.

   Among the significantly upregulated pathways in PNKP depleted cells,
   the enrichment in the autophagy and lysosomal signatures led us to
   investigate the impact of PNKP depletion/inhibition on lysosomal
   activity and autophagy, both being involved in ferroptosis execution
   [[119]34]. We observed upregulation in the lysosomal marker LAMP1
   (lysosomal associated membrane protein 1) ([120]Fig. 3D–[121]S3D) and
   the autophagic protein LC3B (Microtubule-associated protein 1 light
   chain 3B), as shown by the immunofluorescence (IF) ([122]Fig. 3F) and
   by the WB ([123]Fig. 3E) analysis. We also observed increased intensity
   of LysoTracker ([124]Fig. 3I–[125]S3C), and its co-localization with
   ferritin (FTH) ([126]Fig. 3I). These results suggest that PNKP KD
   enhances ferritinophagy, a process that promotes autophagic degradation
   of ferritin and the subsequent release of iron to upregulate LIP
   [[127]35]. Consistent with these findings, we observed an upregulation
   in the mRNA levels of LAMP1 as well as of the autophagy-lysosomal
   transcription factors MITF and TFEB, both known to activate the
   autophagic pathway [[128]36], and enhanced nuclear staining of MITF in
   various TNBC cell lines ([129]Fig. 3G,H, [130]S3E). The elevated
   activity of the lysosomal/autophagy pathway in PNKP-depleted cells is
   also reflected by the expression profiles of breast cancer patients,
   particularly of TNBC, where a negative correlation exists between PNKP
   expression and the autophagic pathway ([131]Fig. 3J). Notably,
   stratification of breast cancer patients based on PNKP expression
   levels revealed a negative correlation between PNKP and MAP1LC3B, which
   encodes the LC3B autophagy protein ([132]Fig. 3SF).

2.4. PNKP depletion induces DNA damage and STING-mediated autophagy

   Considering the strong effects of PNKP on the lysosomal-autophagic
   activities ([133]Fig. 3B and 3D-I, S3A, C-E), its known DNA repair
   function [[134]32,[135]33] and the enrichment in IFN-STING pathway in
   PNKP KD cells ([136]Fig. 3B and C), we assumed that PNKP may modulate
   the autophagic activity through the cGAS (cyclic GMP-AMP
   synthase)-STING pathway. To explore this possibility, we first
   demonstrated that PNKP depletion/inhibition indeed increased DNA damage
   using WB and IF analysis of phospho-histone H2AX (pγH2AX/Ser139)
   ([137]Fig. 4A and B, E, F, [138]S4A), as a marker for DNA double-strand
   breaks [[139]37]. Elevated level of RAD51, which is involved in DNA
   DSBs repair [[140]38], were also detected in PNKP KD cells ([141]Fig.
   4C), further demonstrating the impact of PNKP on DNA damage.

Fig. 4.

   [142]Fig. 4
   [143]Open in a new tab

   Increased DNA damage and STING-dependent autophagy upon PNKP deletion.

   (A–C) WB analysis demonstrating the influence of PNKP depletion
   (siPNKP) in different TNBC cell lines on DNA damage using antibodies
   against p-γH2AX (A) and RAD51 (C). GAPDH (in A) or Tubulin (in C) were
   used for loading control. (B) Representative confocal images for
   p-γH2AX staining in control and PNKP depleted MDA-MB-468 cells. Scale
   bar, 20 μm. (D) Increased mitochondrial DNA (mtDNA) damage in PNKP
   depleted (shPNKP) cells. Total DNA (genomic and mitochondria) was
   purified from the indicated TNBC cells and mtDNA fragments were
   amplified by qPCR as described in Methods. Shown are the mean
   values ± SD as fold change of control from four independent
   experiments. P-values were determined by one-sample t-test. (E–G)
   Increased DNA damage is associated with upregulation of STING and
   autophagy upon PNKP depletion. The indicated TNBC cell lines were
   either depleted of PNKP by siRNA (E, G) or treated with the PNKP
   inhibitor (PNKPi) A12B4C3 (F) and were analyzed by WB with the
   indicated antibodies. GAPDH was used for loading control. (H) Increased
   levels and co-localization of STING and LC3B following PNKP knockdown.
   Control (siCTRL) and PNKP depleted (siPNKP) MDA-MB-468 cells were
   co-stained for STING (red) and LC3B (green). Shown are representative
   IF images. Scale bar, 10 μm. (I–J) Inhibition of STING suppressed
   ferritinophagy in PNKP KD cells. Control (siCTRL) or PNKP depleted
   (siPNKP) MDA-MB-468 cells were incubated in the presence or absence of
   STING inhibitor (STINGi) C-170 (0.1 μM, 24 hr) and then analyzed either
   for ferritin accumulation in lysosomes (I) by co-staining with
   LysoTtracker and FTH as shown in the confocal images (Scale bar, 5 μm),
   or by qPCR (J) for autophagy/lysosomal-related transcripts. Shown are
   the mean values ± SD from two independent experiments. (K–Q) Inhibition
   of STING reduces LIP, lipid-ROS, autophagy and rescued cell viability
   of PNKP KD cells. Control (siCTRL) and PNKP depleted (siPNKP)
   MDA-MB-468 cells treated with STINGi were stained with FerroOrange (K,
   M), BODIPY-C11 (L, N) and processed by confocal microscopy (Scale bar,
   10 μm), assessed for LC3B (O) and for 4HNE protein adducts (P) by WB,
   or for cell viability by MTT assay (Q). Cell viability (Q) and
   fluorometric quantification of K (in M) and of L (in N) were calculated
   relative to control (siCTRL) and the mean values ± SD from n = 3 (in M,
   N) and n = 2 (in Q) experiments are shown. (R–S) STING silencing
   suppresses the lysosomal activity and rescues cell viability of PNKP
   depleted cells. Control (pLKO) or STING depleted (shSTING)
   MDA-MB-468 cells were transfected with siCTRL or siPNKP and 72 hr later
   were stained with LysoTracker (R) and processed for confocal microscopy
   (Scale bar, 10 μm), or assessed for cell viability (S) by MTT. Cell
   viability was calculated relative to control (siCTRL) and the mean
   values ± SD from 2 experiments are shown.

   As PNKP localizes both in the nucleus and the mitochondria, and is
   involved in maintaining mitochondrial DNA (mtDNA) stability and
   integrity [[144]25], we employed a qPCR-based approach as previously
   described [[145]25,[146]39] to monitor mtDNA damage. As shown in
   [147]Fig. 4D, depletion of PNKP significantly reduced the levels of
   intact mtDNA. The elevated DNA damage observed in PNKP-depleted cells
   was accompanied by increased cytosolic double-stranded DNA (dsDNA)
   fragments ([148]Fig. S4B). This was evidenced by enhanced anti-dsDNA
   antibody staining and the reduced co-localization with MitoTracker,
   suggesting leakage of DNA into the cytosol.

   Previous studies have demonstrated a functional link between DNA damage
   and activation of the cGAS-STING pathway, which is involved in
   inflammation and an innate immune response and can also activate
   autophagy through different mechanisms, including direct interaction
   with LC3 [[149][40], [150][41], [151][42], [152][43]]. Indeed, we found
   that KD of PNKP ([153]Fig. 4E) or inhibition of its activity by A12B4C3
   ([154]Fig. 4F) increased the level of STING1 in different TNBC cell
   lines and its colocalization with LC3B ([155]Fig. 4H) as well as the
   level of LC3B-II, a marker of autophagosomes. We also observed elevated
   level of phospho-TBK1 (TANK-binding kinase 1) ([156]Fig. 4G), which is
   recruited to activated STING and phosphorylates the transcription
   factor IRF3, leading to transcription of type I IFN, particularly IFN-β
   as previously reported [[157]33]. These results suggest that the
   cGAS-STINGSTING pathway may contribute to the enhanced autophagic
   response observed in PNKP depleted cells, thereby promoting ferroptosis
   death. To explore this possibility, we examined the influence of C-170,
   a potent and covalent STING inhibitor [[158]44] on autophagy/lysosomal
   activity and ferroptotic death in PNKP-depleted cells. As shown in
   [159]Fig. 4I, C-170 significantly reduced the signal of LysoTracker
   staining and its co-localization with FTH in PNKP KD cells, suggesting
   reduced ferritinophagy. These results were accompanied by reduced
   levels of LAMP1, TFEB and MITF transcripts ([160]Fig. 4J), in agreement
   with previous reports demonstrating the impact of cGAS-STING pathway on
   TFEB and lysosomal biogenesis [[161]45]. C-170 also dimmed the
   FerroOrange signal of PNKP-depleted cells ([162]Fig. 4K–M), suggesting
   a STING-dependent elevation of LIP in PNKP-KD cells. Furthermore, STING
   blockade significantly reduced lipid peroxidation in PNKP-depleted
   cells, as demonstrated by the level of oxidized BODIPY-C11 ([163]Fig.
   4L–N) and by the 4-HNE protein adducts ([164]Fig. 4P). C-170 also
   decreased LC3B lipidation ([165]Fig. 4O) and significantly restored
   cell viability of PNKP-KD cells ([166]Fig. 4Q). Similarly, silencing of
   STING1 expression by shRNA suppressed the ferroptosis-associated
   features of PNKP deficient cells. This was reflected by reduced
   lysosomal activity ([167]Fig. 4R and [168]S4D), decreased lipid
   peroxidation and LIP (Fig. [169]S4E–F and [170]S4G, H respectively) and
   a partial rescue in cell viability ([171]Fig. 4S). Collectively, these
   results suggest that PNKP depletion elevates DNA damage, thereby
   activating the cGAS-STING pathway, which in turn modulates the
   autophagic machinery and enhances ferritinophagy and ferroptotic death.

2.5. STAT3 cooperates with PNKP to modulate autophagy-lysosomal and
ferroptosis

   In addition to STING, it was previously shown that STAT3 also modulates
   autophagy in different cancer types through transcriptional regulation
   of autophagic-related genes and/or interaction with autophagic-related
   proteins [[172]46]. It was also shown that STAT3 is highly active in
   TNBC and that TNBC are highly sensitive to STAT3 inhibition [[173]47].
   Furthermore, recent studies suggest that STAT3 inhibition may induce
   ferroptosis in certain (e.g. gastric cancer), but not all (e.g.
   pancreatic ductal adenocarcinoma) cancer types [[174]48]. In line with
   these reports, and the downregulation of STAT3 pathway in the RNAseq
   data of PNKP-depleted cells ([175]Fig. 3B and C), we examined the
   influence of PNKP depletion on STAT3 activation by WB and by IF
   analysis using anti-pSTAT3 (pY705) antibody. As shown in [176]Fig.
   5A–C, [177]S5A the levels of pSTAT3 were markedly reduced in different
   PNKP-KD TNBC cell lines (MDA-MB-468, BT549, Hs578, HCC1937, HCC38)
   without an obvious effect on the total STAT3 level, suggesting that
   PNKP depletion impairs STAT3 activation. These observations led us to
   examine the impact of STAT3 inhibition on autophagy/lysosomal activity,
   using the small molecule STAT3 inhibitor Stattic [[178]49]. As shown in
   [179]Fig. 5D and E, treatment with Stattic reduced the levels of pSTAT3
   and FTH proteins but increased the levels of LAMP1 and TFEB proteins,
   as well as pγH2Ax, STING and LC3B levels ([180]Fig. S5F). The
   transcripts of LAMP1 and TFEB were also increased ([181]Fig. 5F) as
   well as the intensity of LysoTracker staining ([182]Fig. 5G) and of
   FerroOrange signal ([183]Fig. 5H). These results indicate that, similar
   to the effects of PNKP depletion/inhibition, STAT3 inhibition also
   enhances lysosomal activity, ferritinophagy and increases LIP in TNBC.
   We, therefore, examined whether Stattic can also enhance lipid
   peroxidation in TNBC cells. Indeed, we found that treatment with
   Stattic (0.25 μM, 24 h) induced lipid peroxidation as determined by
   BODIPY-C11 staining ([184]Fig. 5I) and by the level of 4-HNE-protein
   adducts as shown in the WB analysis of Stattic-treated cells
   (0.25–2 μM, 24 h) ([185]Fig. 5J). In line of these results, we found
   that Stattic (0.5 μM) could sensitize TNBC cells to ferroptosis induced
   by RSL3, a canonical GPX4 inhibitor. As shown by the RSL3 dose response
   curve, Stattic significantly reduced the IC[50] of RSL3 in
   MDA-MB-468 cells and even more profoundly in BT549 cells ([186]Fig.
   5K). Likewise, depletion of PNKP sensitized TNBC cells to RSL3 as well
   as to FIN56, which induces ferroptosis by triggering degradation of
   GPX4 [[187]50] and by inhibiting GPX4 translation and Coenzyme Q10
   production [[188]51]([189]Fig. 5K–[190]S5D).

Fig. 5.

   [191]Fig. 5
   [192]Open in a new tab

   PNKP modulates STAT3 activity to enhance autophagy and ferroptosis.

   (A–B) Depletion of PNKP reduced the level of phospho-STAT3 (pY705). The
   levels of pSTAT3 in control (siCTRL) and PNKP depleted (siPNKP) cells
   were assessed either by cell fractionation (A) and WB using Histone 3
   (H3) as loading control, or by IF (B) for pSTAT3. Scale bar, 5 μm. (C)
   Total cell lysates from control (siCTRL) and PNKP-depleted (siPNKP)
   TNBC cells were assessed by WB with the indicated antibodies. The blot
   marked with “∗” (PNKP) appears also in [193]Fig. 2M, as both panels
   originate from the same Western blot experiment. (D–J) STAT3 inhibition
   enhances lysosomal activity, LIP and lipid ROS. Cells were treated with
   the STAT3 inhibitor Stattic (0.25 μM, 24 hr) and assessed by WB for
   STAT3 activation using anti-pSTAT3 (D) or the indicated
   autophagy-lysosomal markers (E). GAPDH was used as loading control (D,
   E). (F) The transcript levels of LAMP1 and TFEB were measured by qPCR.
   Shown are the mean values ± SD from two independent experiment
   (p-values measured by one-sample t-test). Cells were treated with
   Stattic (0.25 μM, 24 hr) and stained with LysoTracker (G), FerroOrange
   (H), or with BODIPY-C11 (I). Shown are representative confocal images.
   Scale bars, 20 μm (G, H), 10 μm (I). Fluorometric quantification of
   lipid-ROS was calculated relative to control (DMSO) and the mean
   values ± SD from three experiments are shown. (J) MDA-MB-468 cells were
   treated with increasing concentrations of Stattic (0.1–2 μM) and 20 hr
   later assessed by WB analysis for 4-HNE protein adducts. β-tubulin
   served as loading control. (K) Dose-response curve of RSL3 in the
   indicated control (pLKO), PNKP depleted (shPNKP), or Stattic-treated
   (0.5 μM) TNBC cell lines. MTT assay was used to measure cell viability
   72 hr post treatment. Shown are mean percentage relative to DMSO ± SD
   from two independent experiments. (L–M) Co-inhibition of PNKP and STAT3
   induced ferroptotic cell death. MDA-MB-468 cells (L) were treated with
   the indicated drugs in the absence or presence of iron chelator
   (2,2′BP, 8 μM) or UAMC-3203 (UAMC, 2.5 μM) and assessed 72 hr later for
   cell viability by MTT. Shown are percentage (%) of control with the
   mean values ± SD from three experiments. The indicated TNBC cells (M)
   were treated with the PNKP and/or STAT3 inhibitors for 24 hr and then
   assessed by WB using anti-SCD1 antibody. GAPDH was used for loading
   control. (N, O) PNKP-STAT3 interaction is enhanced during ferroptosis
   and/or doxorubicin treatment. Cells were treated either with RSL3 (N)
   or doxorubicin (O) were co-stained for PNKP (red) or STAT3 (green).
   Scale bar, 10 μm. (P) Increased phospho-PNKP after treatment with
   doxorubicin. Cell fractionation (left) and IF (right) were carried out
   24 h post doxorubicin treatment and the levels of phospho-PNKP
   (pS114/T118) were assessed by WB. C-Myc and GAPDH were used as loading
   controls for nuclear (Nuc.) and cytoplasmic (Cytopl.) fractions
   respectively. Representative IF images of pPNKP staining (red) are
   shown, scale bar 10 μm. (Q, R) Co-Immunoprecipitation (Co-IP) of PNKP
   and STAT3. The indicated TNBC cells were subjected to
   immunoprecipitation (IP) with or without anti-PNKP (±) in steady state
   conditions (Q) or 24 hr post doxorubicin treatment (R). Shown are WB
   with the indicated antibodies demonstrating Co-IP of STAT3 with PNKP.
   (S) Enhanced SHP-2 activation in PNKP KD cells. Control and PNKP
   depleted (siRNA/shPNKP) TNBC cells were assessed for SHP-2
   phosphorylation (pY542) by WB analysis. GPADH was used for loading
   control.

   As both PNKP and STAT3 inhibition increased autophagic/lysosomal
   activity, lipid peroxidation, and sensitized TNBC cells to ferroptosis
   ([194]Fig. 5K–[195]S5D), we examined the impact of their co-inhibition
   on cell viability. We applied a dose of the PNKP inhibitor A12B4C3
   (10 μM), which has no significant effect on cell viability ([196]Fig.
   S1H), together with ∼IC[50] concentration of Stattic, and found that
   the drugs combination drastically reduced cell viability in both
   MDA-MB-468 and BT-549 cells, which could be partially restored by the
   ferroptosis inhibitor UAMC-3203 and the iron chelator 2,2′-Bipyridyl
   ([197]Fig. 5L and [198]S5G, respectively). This strong effect of STAT3
   and PNKP co-inhibition on cell viability was accompanied by an increase
   in the transcription levels of several autophagic genes, including
   MAP1LC3B, TFEB, ATG14, ATG4B and BCN1, which are essential for a
   autophagosome formation [[199]52,[200]53] ([201]Fig. S5B). On the other
   hand, we observed a substantial decrease in the protein level of SCD1
   ([202]Fig. 5M) and in the transcription levels of key ferroptosis
   protecting genes, including GPX4 and SCD1 in response to PNKP and STAT3
   co-inhibition compared to inhibition of either PNKP or STAT3 alone
   ([203]Fig. S5B). In summary, although PNKP inhibition reduced STAT3
   activation, we found that co-inhibition of STAT3 and PNKP provokes
   ferroptotic death. These results suggest that PNKP cooperates with
   STAT3 to protect TNBC cells from ferroptosis, possibly through their
   effects on autophagic/lysosomal activity and/or DNA damage. In line
   with this hypothesis, we found that ferroptosis inducers (RSL3,
   Erastin) enhanced the nuclear staining of pPNKP ([204]Fig. S5E) shortly
   after treatment (5–10 hr), and concurrently the colocalization between
   PNKP and STAT3 ([205]Fig. 5N). The elevated levels of nuclear pPNKP and
   PNKP-STAT3 colocalization were also observed in response to doxorubicin
   treatment for 24 hr ([206]Fig. 5O–P, [207]S5C), implying that the two
   proteins may interact. Indeed, co-immunoprecipitation (Co-IP)
   experiments ([208]Fig. 5Q and R) suggest a physical interaction between
   STAT3 and PNKP. This interaction might protect STAT3 from
   dephosphorylation, thereby reducing STAT3 phosphorylation in
   PNKP-depleted cells ([209]Fig. 5A–C, [210]S5A). The interaction might
   also play a role in transcriptional regulation of specific targets
   genes and/or in DNA damage response [[211]54]. Regardless the exact
   role of PNKP-STAT3 interaction, we found a significant increase in
   SHP-2 phosphorylation in PNKP KD TNBC cell lines ([212]Fig. 5S). These
   results suggest that PNKP depletion leads to activation of the SHP-2
   phosphatase [[213]55], which in turn dephosphorylates its substrate
   pSTAT3 [[214]56]. Collectively, we found that PNKP cooperates with
   STAT3, possibly through physical interaction and/or inhibition of SHP-2
   activation, to protect TNBC cells from ferroptosis by regulating the
   autophagy/lysosomal machinery and the DNA damage response. This newly
   identified physical and functional interaction between PNKP and STAT3,
   together with previous reports on the crosstalk between the STAT3 and
   the cGAS-STING pathways [[215][57], [216][58], [217][59]], introduce
   the PNKP-STING-STAT3 axis as a new ferroptosis regulator, integrating
   the autophagy-lysosomal activities with the DNA damage.

2.6. Combination of PNKP targeting and doxorubicin reduced tumor growth in a
xenograft mouse model and TNBC patients-derived organoids

   Although co-inhibition of STAT3 and PNKP significantly reduced cell
   viability ([218]Fig. 5L–[219]S5G), the high toxicity of Stattic
   [[220]60] limits its clinical use for TNBC therapy. Doxorubicin, on the
   other hand, has long been used as a first line treatment for TNBC
   patients. In addition, the RNAseq data ([221]Fig. 3B and C) indicates
   that the expression of doxorubicin resistance-related genes was reduced
   in PNKP KD cells, while PNKP expression was upregulated in
   chemoresistant TNBC patients ([222]GSE20194) ([223]Fig. 6A). These
   observations prompted us to examine the combined effects of PNKP
   inhibition and doxorubicin treatment on TNBC cells viability. We
   selected two fixed doses of doxorubicin (60 and 100 nM) that inhibit
   cell viability by 5–45 % across the different TNBC cell lines, and
   effectively low dose of PNKP inhibitor (10 μM). As shown in the
   dose-response matrix ([224]Fig. 6B), the drugs combination
   substantially reduced cell viability, and exhibit strong synergy as
   reflected by the calculated combination index (CI)([225]Fig. 6C) using
   the Bliss independence method [[226]61]. This synergistic effect was
   associated with elevated levels of DNA oxidation as demonstrated by IF
   analysis with anti-8-hydroxy-2-deoxyguanosine (8-OHdG) antibody
   ([227]Fig. 6D). Notably, the synergy was obtained only for doxorubicin
   and not for paclitaxel or cisplatin ([228]Fig. S6C), two
   chemotherapeutic drugs which are commonly used to treat the TNBC
   patients.

Fig. 6.

   [229]Fig. 6
   [230]Open in a new tab

   Elevated levels of PNKP in chemoresistant TNBC patients and its
   targeting together with doxorubicin suppress TNBC growth both in vitro
   and in vivo.

   (A) PNKP expression in TNBC patients, stratified by their chemotherapy
   response: pCR (complete response, n = 45) and RD (residual disease,
   n = 165). Data taken from GEO record [231]GSE20194. (B) Heatmap
   representation of cell viability in response to the indicated drugs or
   drugs combinations in several TNBC cell lines. Cell viability was
   measured 72 hr post treatment and is shown as percentage of control
   (DMSO). Experiment done twice in duplicates, shown are means of the
   duplicates in one representative repeat. (C) Combination index (CI) for
   the combination between 10 μM PNKPi (A12B4C3) and 60–100 nM doxorubicin
   in the indicated cell lines was calculated according to the Bliss
   independence method. (D) Inhibition of PNKP (10 μM) potentiates the
   effect of doxorubicin (60 nM) on DNA oxidation as assessed by IF
   staining with anti-8OHdG antibody (red) 24 hr post treatment. Shown are
   representative confocal imagines. Scale bar, 5 μm. (E) BT549 or 4T1
   cells grown as spheroids were either treated with PNKP inhibitor
   (60 μM) or with doxorubicin (360–600 nM) alone or with combination of
   the two drugs for 10 days. Shown are representative brightfield images.
   Scale bar, 100 μm. (F) Control or BT549 cells expressing Tet-shPNKP
   were grown as spheroids and treated either with DMSO as control,
   doxycycline (6 μg/ml) to induce PNKP KD, doxorubicin (360–600 nM), or
   doxycycline and doxorubicin in the absence or presence of ferrostatin-1
   (18 μM). Brightfield images were acquired 15 days post treatment. Scale
   bar 100 μm. (G, H) Effects of single and drugs combination on TNBC
   patients derived organoids (PDOs) (G) were evaluated for viability
   assay 72 hr post treatment using the ATPlite luminescence assay as
   described in Methods. Shown are the mean percentage of viable organoids
   relative to control (DMSO) ± SD from two different PDOs (top and
   bottom), with at least 3 replicates per PDO. (H) Representative
   brightfield images showing the morphology of organoids after different
   treatments, scale bar 100 μm. (I–L) Depletion of PNKP together with
   doxorubicin treatment reduced tumor burden and increased lipid-ROS.
   Naïve or Tet‐shPNKP expressing MDA‐MB‐468 cells were implanted - into
   the mammary fat pad of female nude mice. When tumors reached ∼100 mm^3,
   mice were randomized into four groups: Naïve or
   Tet‐shPNKP ± doxorubicin, n = 6–8 tumors/group. The two groups
   expressing the Tet-shPNKP were treated with doxycycline to knockdown
   PNKP. (I) Graph shows the mean tumor volume ± SEM in the indicated time
   points from non-induced (-doxycycline; naïve ± doxorubicin) and induced
   (+doxycycline; Tet‐shPNKP ± doxorubicin) tumors. p-values shown were
   measured by t-test comparing the volumes on day 26. 2-way ANOVA was
   performed to verify the significant difference between the groups
   (p-value = 5.11e-4) and the time points (p-value = 6.67e-31). (J–K)
   Image of excised tumors showing the reduced size after combining effect
   of PNKP depletion and doxorubicin treatment (J) along with end point
   (day 26) tumors weight (K) of the indicated groups. Shown are the mean
   mass of excised tumors ± SD (n = 6–8 per group). (L) Western blot
   analysis of control (naïve) and PNKP depleted tumors
   (Tet-shPNKP) ± doxorubicin (n = 3–4/group). Tumors were homogenized,
   lysed and analyzed for 4HNE‐protein adducts, and GAPDH was used as a
   loading control.

   To further demonstrate the potency of this combination, we compared the
   IC[50] values of doxorubicin in control TNBC cell lines to those of the
   PNKP KD (Tet-inducible shRNA) cells. The IC[50] values were extracted
   from the doxorubicin dose-response curves and were reduced by ∼2.5 fold
   ([232]Fig. S6A–B). Furthermore, this combination was also potent in 3
   dimensional (3D) spheroids ([233]Fig. 6E and F). We either used A12B4C3
   ([234]Fig. 6E) or the inducible PNKP KD ([235]Fig. 6F) and observed
   robust effects of PNKP targeting together with doxorubicin on TNBC cell
   viability (4T1, BT549) 10 days post treatment as shown by
   representative brightfield images. Importantly, ferrostatin-1 restored
   cell viability of BT549 spheroids treated with doxorubicin together
   with doxycycline to downregulate PNKP ([236]Fig. 6F).

   To expand these observations, we examined the effects of the drugs
   combination on TNBC patients-derived organoids (PDOs). The organoids
   were established as described in Methods and treated either with
   doxorubicin (0.24 μM) or A12B4C3 (20 μM–60 μM) alone or with the drugs
   combinations. As shown, single treatments had no significant effects on
   cell viability, but the combined treatments of doxorubicin (0.24 μM)
   with PNKPi (40 and 60 μM) substantially suppressed the growth of PDOs
   from two different TNBC patients ([237]Fig. 6G and H). Finally, we
   established xenograft models by orthotopic injection of the human
   MDA-MB-468 cells, either Naïve cells or cells expressing the inducible
   Tet-shPNKP into the mammary fat pad of nude mice. The tumors were grown
   for approximately two weeks to reach ∼100 mm^3, and then half of the
   Naïve group was treated with vehicle, while the second half with
   doxorubicin to assess the influence of doxorubicin alone. Likewise,
   half of the Tet-shPNKP expressing group were treated with doxycycline
   to induce PNKP KD, while the second half with doxycycline and
   doxorubicin to assess the impact of the combination. The mice were
   treated for 28 days, and tumor volume was measured every 4–7 days by
   digital Vernier caliper. As shown in [238]Fig. 6I, doxorubicin alone
   decreased tumor growth but not significantly, while PNKP KD had a
   stronger effect and significantly attenuated tumor growth. The
   combination of PNKP KD and doxorubicin had very strong effects on tumor
   growth. The tumors remained almost in the same volume throughout the
   treatment ([239]Fig. 6I) and were significantly smaller at the end
   point as shown in the image of the excised tumors ([240]Fig. 6J) and by
   their average mass ([241]Fig. 6K). Interestingly, the excised tumors of
   PNKP-depletion or its combination with doxorubicin were enriched with
   4-HNE-protein adducts as shown by the WB of representative tumors with
   anti-4-HNE antibodies ([242]Fig. 6L), suggesting increase in
   ferroptotic death. Overall, this data indicates that PNKP targeting
   robustly increases the potency of doxorubicin, thereby introducing PNKP
   as a promising, clinically relevant, therapeutic target for TNBC
   therapy.

3. Discussion

   In this study we show that targeting of PNKP in TNBC induces
   ferroptotic death ([243]Fig. 1, [244]S1) and potentiates the efficacy
   of doxorubicin ([245]Fig. 6, [246]S6A, B), suggesting that combining
   doxorubicin with PNKP blockade could improve the clinical outcome of
   doxorubicin treatment and may reduce its associated toxicity.

   Targeting of PNKP was previously proposed to have a therapeutic
   potential in several human cancers, including lung, breast, colorectal,
   and prostate cancer, particularly in combination with γ-radiation or
   topoisomerase I inhibitor [[247]62,[248]63]. In prostate cancer cells,
   combination of PNKP inhibitor (A12B4C3) and carbon ion irradiation
   induced cell cycle arrest and apoptotic death [[249]62]. In these
   cells, similarly to our data ([250]Fig. 1K, L, [251]S1H), a relatively
   high concentration of A12B4C3 (10–20 μM) was used at a micro rather
   than a nano molar range as previously reported [[252]28]. These
   observations highlight the need for a highly potent and selective PNKP
   inhibitor.

   Although additional small molecule inhibitors were developed (such as
   A83B4C63), they exhibit high toxicity towards normal cells [[253]63].
   Nevertheless, for the in vitro studies, A12B4C3 had similar effects as
   the PNKP KD (shRNA/siRNA) and induced ferroptotic death in several TNBC
   cell lines ([254]Fig. 1L). The ferroptotic death induced by PNKP
   depletion/inhibition was accompanied by characteristic hallmarks of
   ferroptosis [[255]64], including increase in lipid peroxidation, ROS,
   and LIP, concomitant with a decrease in GSH, GPX4, and SCD1 levels
   ([256]Fig. 1F–L, [257]S1B-G, 2M−O, [258]S2D).

   Through transcriptomic analysis and mechanistic studies, we found that
   PNKP depletion robustly enhanced the lysosomal/autophagy activities
   ([259]Fig. 3B–I, [260]S3A, [261]S3C-E) and increased DNA damage
   response ([262]Fig. 4A–F, [263]S4A-B). The role of PNKP in DNA repair,
   especially of SSBs, and its implication in the pathology of several
   autosomal recessive neurodegenerative diseases, including AOA4
   (Ataxia-ocular motor apraxia 4), CMT2B2 (Charcot–Marie–Tooth disease)
   and MCSZ (Microcephaly, seizures, and developmental delay) have been
   described in various studies [[264]65].

   However, its impact on the lysosomal/autophagy machinery and on
   ferroptotic death as described in this study was found to be mediated
   by its dual effects on the cGAS-STING and the STAT3 pathways, which are
   both involved in autophagy and DNA damage response ([265]Fig. 7A).
   While the DNA damage response associated with STING function is
   mediated by cGAS and TBK axis to modulate antitumor and antiviral
   immunity [[266]66], recent studies suggest that STING regulates
   lysosomal biogenesis and autophagy through TFEB activation in a
   TBK-independent manner [[267]67]. TFEB activation and transcription of
   TFEB-targeted genes were also obtained in response to STAT3 inhibition
   in an ischemic stroke model [[268]68], consistent with our results
   shown in [269]Fig. 5E–F, [270]S5B.

Fig. 7.

   [271]Fig. 7
   [272]Open in a new tab

   The PNKP-STING-STAT3 axis integrates autophagy/lysosomal machinery and
   DNA damage (A) An illustration depicting the impact of PNKP targeting
   on the STAT3 and cGAS-STING pathways, highlighting their crosstalk and
   feedback loops to potentiate ferroptosis and chemotherapy.

   Under baseline conditions of cancer cells (No Perturbation), pSTAT3
   positively regulates tumor growth, while negatively regulates DNA
   damage response, the lysosomal-autophagic activities and ferroptosis.
   STAT3 activation is negatively regulated by the phosphatases SHP-1/2.
   We identified an interaction between PNKP and STAT3, which was
   transiently enhanced upon ferroptosis induction or doxorubicin
   treatment. This interaction might directly protect pSTAT3 from
   dephosphorylation and/or indirectly from ferroptosis through
   transcriptional modulation of ferroptosis protecting genes, such as
   GPX4, SCD1, FTH1, and of autophagic genes.

   Depletion/inhibition of PNKP disrupts the DNA repair machinery, leading
   to increased DNA damage and subsequent STING activation, which in turn
   enhances the autophagy-lysosomal activities. Simultaneously, the levels
   of pSTAT3 decrease, possibly due to SHP-2 activation and/or loss of
   physical protection. Decrease in pSTAT3 further enhances the
   autophagy-lysosomal machinery and the DNA damage, which potentiate the
   activity of STING to enhance ferritinophagy, lipid peroxidation, and
   ferroptotic death. This potentiation is mediated via positive feedback
   loops and contributes to the potent anti-tumor efficacy of PNKP
   targeting in combination with the chemotherapeutic drug doxorubicin.

   (B) The interplay between STAT3 and cGAS-STING pathways. The cGAS-STING
   (stimulator of interferon genes) pathway is involved in innate immune
   response and is activated by accumulation of cytosolic DNA fragments
   produced by DNA damage or microbial infections. The DNA fragments are
   sensed by cGAS (cyclic GMP-AMP synthase), which synthesizes cGAMP
   (cyclic GMPAMP) from ATP and GTP to activate STING. Activation of STING
   at the ER triggers its translocation to the Golgi apparatus and
   concomitantly the recruitment of TBK1 (TANK-binding kinase 1) and
   IKKα/β (Inhibitory kappaB kinase). TBK1 phosphorylates the
   transcription factor IRF3 (interferon regulatory factor 3), which in
   turn induces transcription of type I interferons (INFβ, α), while
   IKKα/β phosphorylates IkBα (inhibitor of nuclear factor kappa B),
   leading to release and activation of NF-kB. Activated NF-kB induces the
   transcription of inflammatory cytokines, such as TNF, IL-1β, and IL-6.
   Secreted IFNα/β activates their receptors IFNAR1/2, leading to JAK
   (Janus kinase)/TYK2 activation and phosphorylation of STAT1/2. The
   STAT1/2 dimer recruits IRF9 (interferon regulatory factor), translocate
   to the nucleus and triggers the transcription of ISGs (IFN-stimulated
   genes) including cGAS, demonstrating the positive feedback loops
   (dashed brown lines with arrow ends) of this pathway. IL-6 produced in
   response to NF-kB activation, binds to a cell-surface receptors complex
   composed of IL-6R and gp130, which in turn activates JAK1/2, leading to
   STAT3 phosphorylation (Y705), dimerization, and activation of
   transcription of inflammatory and anti-autophagic genes and of SOCS
   (suppressor of cytokine signaling), thus highlighting the crosstalk
   between the STAT3 and STING pathways. SOCSs act as negative feedback
   (dashed brown lines with circle ends) regulators by inhibiting JAK
   activity, preventing excessive signaling. STAT3 induces the expression
   of SOC3, which inhibits IFNα-induced STAT1 phosphorylation and
   consequently ISG expression, while suppressing the expression of STAT1,
   IRF7, and IRF9, further demonstrating the interplay between the two
   pathways. In addition to the role of STING in immune response, it
   activates autophagy by direct interaction with LC3 [[273]40] and by
   activation of MITF [[274]67], while STAT3 indirectly regulates
   autophagy and can suppress MITF [[275]80].

   The STAT3 and STING pathways have also been implicated in ferroptosis
   regulation [[276]69]. STING regulates ferroptosis through multiple
   mechanisms in a cell context dependent manner, and was reported to
   either inhibit or enhance ferroptosis [[277]70,[278]71]. In line with
   our results ([279]Fig. 2M–O, [280]S2D-E), recent studies suggest that
   STAT3 is upregulated in chemoresistant gastric cancer cells, and it
   directly binds to the promoter sequences of key negative regulators of
   ferroptosis, including GPX4, SCL7A11 and FTH1, thereby modulating their
   expression and protecting cells from ferroptosis [[281]48].

   Moreover, these two pathways are connected through several feedback
   loops. Previous studies showed that activation of STING by DNA damage
   in TNBC induced the expression of IL-6 and an autocrine activation of
   STAT3 [[282]57]. However, our findings suggest that this feedback
   activation could be suppressed by PNKP depletion, which markedly
   reduces STAT3 activation ([283]Fig. 5A–C, [284]S5A). Other studies
   showed that STAT3 inhibition enhanced STING signaling induced by STING
   agonists and synergistically increased the anti-tumor immunity response
   in animal models of TNBC [[285]58]. STAT3 inhibition also synergized
   with inducers of type I IFN, a downstream effector of the cGAS-STING
   pathway, to effectively inhibit B cell lymphoma growth, possibly by
   alleviating the suppressive effects of STAT3 on IRF7, IRF9, STAT1, and
   STAT2 expression [[286]72]. On the other hand, the two pathways are
   also connected through negative feedback loops mediated by the
   Interferon-stimulated genes (ISGs) and the SOCS (Suppressors of
   cytokine signaling), SOC1S and SOCS3. SOCS can suppress STAT1 and STAT3
   activation, and consequently IFN-α mediated ISGs expression [[287]73].
   The interplay and crosstalk between STAT3 and cGAS-STING pathway is
   illustrated in [288]Fig. 7B and highlights the feedback loops between
   the cascades.

   The dual effect of PNKP depletion/inhibition on cGAS-STING activation
   and concurrently on STAT3 inhibition may increase the robustness of the
   response, possibly through inhibition/activation of feedback loops that
   eventually potentiate the response of STING activation and of STAT3
   inhibition as shown in [289]Fig. 7A. This may explain the synergistic
   effect of PNKP and STAT3 co-inhibition ([290]Fig. 5K–M, [291]S5B, G).
   We found that PNKP and STAT3 cooperate and possibly can protect cells
   from cell death induced by FINs or doxorubicin ([292]Fig. 5K and L).
   The two proteins colocalize shortly after FINs or doxorubicin
   treatments and can be found in the same immunocomplex, as shown by
   Co-IP assay ([293]Fig. 5N–O, [294]S5C, 5Q, 5R). This interaction might
   prevent inactivation of STAT3 by SHP-2 or play a different role that
   needs to be further defined. Yet, the inactivation of STAT3 in response
   to PNKP inhibition can contribute to the synergistic effect with
   doxorubicin, as doxorubicin by itself induced DNA damage response,
   increased DNA oxidation, most profoundly together with PNKP inhibition
   ([295]Fig. 6B–L, [296]S6A-B), which can further enhance STING
   activation and STAT3 inhibition. Furthermore, previous studies reported
   that STAT3 inhibition and doxorubicin synergistically inhibited TNBC
   cells growth in vitro, reduced the expression of PD-L1 and activated
   immune cell-mediated cancer cell death [[297]57], while other reports
   showed that SHP-2 activation enhances the activity of STING activation
   and chemotherapy response [[298]74].

   Altogether, we identified PNKP as a regulator of ferroptosis in TNBC
   and showed that PNKP is not only involved in DNA repair but also plays
   an important role in regulating the autophagic/lysosomal machinery by
   converging the cGAS-STING and STAT3 pathways, highlighting the impact
   of the PNKP-STING-STAT3 axis. In addition, we found that PNKP could be
   a promising target for TNBC therapy in combination with established and
   already approved chemotherapeutic drugs to increase effectiveness and
   reduce toxicity.

4. Materials & Methods

4.1. Cell culture

   TNBC cell lines and human embryonic kidney HEK293T cells were obtained
   from the American Type Culture Collection (USA). HCC38 was obtained
   from M. Virginie (Institute Curie, Research Centre, Paris, France;
   2014). MDA-MB-468, HCC1937, HCC70, HCC38, BT549, Hs578T and 4T1 cells
   were grown and maintained in RPMI medium (Gibco BRL, USA) containing
   10 % fetal bovine serum (Gibco BRL, USA) and penicillin/streptomycin
   (100 U/ml, 100 μg/ml), while HEK293T cells were grown in Dulbecco's
   modified Eagle medium (DMEM), supplemented as mentioned above. Cells
   were cultured at 37 °C in a humidified incubator of 5 % CO2. Cell lines
   were routinely (once a month) checked for mycoplasma using a PCR assay.

4.2. Chemicals and drugs

   STAT3 inhibitor (Stattic, sc-202818) was purchased from Santa Cruz
   Biotechnology. Liproxstatin-1 (17730), Ferrostatin-1 (SML0583),
   UAMC-3203 (26525), STING inhibitor C-170 (30157), doxorubicin (15007)
   and cisplatin (13119) were purchased from Cayman Chemical. Erastin
   (E7781), 2,2′-bipyridyl (D216305), A12B4C3 (PNKPi, A8736), doxycycline
   (D3072) and doxycycline hyclate (D9891), RSL3 (SML2234), FIN56
   (SML1740), paclitaxel (T7402) and thiazolyl blue tetrazolium bromide
   (MTT) (M5655) were purchased from Sigma.

4.3. Antibodies

   The antibodies list is provided in [299]Supplementary Table 1
   (Supporting Information).

4.4. Drugs treatments

   For chemotherapy experiments: cells were seeded in 96-well plates and
   24h later were treated with either PNKPi (10 μM), doxorubicin
   (60–100 nM), paclitaxel (1–3 nM), cisplatin (4–5 μM) or with
   combination of PNKPi and the above drugs. Control cells were treated
   with DMSO. Cell viability was assessed by MTT 72 hr later. For IF
   experiments, cells were treated with 60 nM doxorubicin in the presence
   or absence of PNKPi 10 μM for 24 hr and processed for IF following
   incubation with the drugs.

   For PNKPi-Stattic combination signaling experiments: cells were treated
   with PNKPi and Stattic as single drugs (PNKPi 10 μM, Stattic 2 μM) or
   drugs combination for 24 hr. Cells were then processed for lysate
   preparation for WB or RNA extraction for qPCR.

   For FIN sensitivity experiments: cells were seeded in 96-wells plates.
   After 24 hr, cells were treated with the indicated concentrations of
   RSL3 or FIN56 compared to DMSO-treated cells. 72 hr later, cell
   viability was measured by MTT assay. Results are presented as % cell
   viability of control untreated (DMSO) cells.

   For cell viability rescue experiments: cells were seeded in 96-well
   plates and 24 hr later treated with the rescue compounds. For PNKPi
   viability rescue, cells were treated 24 h post seeding with PNKPi
   20 μM. Cell viability was rescued by using either 3 μM ferrostatin-1,
   3 μM liproxstatin-1, 2.5 μM UAMC-3203, 8 μM 2,2′-bipyridyl or 0.1 μM
   C-170. For experiments with siPNKP, cells were transfected with siRNA
   according to the manufacturer's protocol and 24 h later the rescuers
   were added to the cultures. Control cells were incubated with DMSO.
   Cell viability was measured by MTT assay 72 hr later. Results are
   presented as % cell viability of control untreated (DMSO) cells.

4.5. Flow cytometry

   Cells were seeded in 12-well plates and 48–72 hr later incubated with
   the indicated fluorescent probe. The cells were then washed with PBS,
   trypsinized and centrifuged at 1200 rpm, for 4 min at 4 °C. Cells were
   washed twice with PBS and resuspended in 200 μl flow cytometry buffer
   (PBS + 0.1 % BSA). Fluorescence was acquired in a BD LSR II (BD
   Bioscience) (10 000 events/second) with the BV605 channel and the mean
   fluorescence intensity (MFI) was analyzed using FlowJo software.

4.6. ROS measurements

   Levels of intracellular ROS were measured by the cell-permeable dye
   CM-H[2]DCFDA [5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein
   diacetate, acetyl ester] (Invitrogen, LSC6827). Cells were cultured in
   96-well black plates (Greiner, 655 090) for 16 hr and then treated with
   drugs as indicated. For total ROS levels cells were treated with 2 μM
   CM-H[2]DCFDA and Hoechst (1 mM) in Dulbecco's Phosphate Buffered Saline
   with Calcium and Magnesium (PBS^++) (02-020-1A, Sartorius, Gottingen,
   Germany) for 10–15 min at 37 °C. Cells were washed twice with PBS^++
   and incubated in PBS^++. Fluorescence of CM-H[2]DCFDA was measured by
   TECAN Infinite MPlex plate reader. CM-H[2]DCFDA fluorescence values
   were normalized by HOECHST values, indicative of the number of living
   cells in each well. Data is presented as percentage of ROS relative to
   control cells.

   Mitochondrial ROS were detected by MitoSOX™ Red 396/610 nm fluorescent
   probe ([300]M36008, ThermoFisher Scientific). MitoSOX™ Red (5 mM in
   DMSO stock) was diluted in serum-free RPMI to a final concentration of
   2 μM and incubated with the cells for 15 min at 37 °C. Cells were then
   processed for flow cytometry as described above. Data is presented as
   percentage of MFI relative to control cells.

4.7. GSH measurement

   GSH levels were measured by the Glutathione Colorimetric Assay kit
   (ab239709, Abcam, Cambridge, UK) according to the manufacturer's
   instructions and as previously described [[301]21].

4.8. Generation of 3D spheroids

   Spheroids were generated as follows. U-shaped 96-well plates (3799,
   Corning, NY, USA) were precoated with 20 mg/ml poly-HEMA
   (Poly-2-hydroxyethyl methacrylate, P3932, Merck-Millipore) in ethanol
   (60 μl per well) and left to dry overnight. Single-cell suspension of
   4T1 or BT549 cells (7-8 × 10^3 cells in 100 μl RPMI medium) was loaded
   into each well and plates were centrifuged at 1,000g for 15 min to
   allow formation of spheroid structures. Plates were then incubated for
   2 days at 37 °C to form a single spheroid per well, and then treated
   with the indicated drugs for 10 days (spheroids were treated by fresh
   RPMI medium containing the indicated drugs every 3–4 days). Brightfield
   pictures of spheroids were acquired by an inverted microscope at × 10
   magnification.

4.9. Cell viability and death assays

   Cells were cultured in a 96-well plate and treated either with the
   indicated drugs, transfected with siRNA or induced with doxycycline.
   Cell viability was measured 72 hr later using the MTT (M2128, SIGMA)
   assay as described previously [[302]75]. Results are presented as %
   cell viability of control untreated (DMSO) cells. For dose response
   curves, calculation of IC50 was performed using the drc package in R.

   Cell death was measured by the CellTox-green Cytotoxicity assay (G8741,
   Promega) according to the manufacturer instructions. In brief, cells
   were seeded in 96 wells black plate (Greiner, 655 090) and treated as
   indicated for 72 hr. CellTox green reagent was diluted 1:1000 in assay
   buffer and added to the cells (100 μl/well), plates were immediately
   shaken for 1 min on an orbital shaker and then incubated in the dark at
   room temperature for 10–15 min. The fluorescence signal was measured by
   the Infinite 200 PRO Tecan fluorometric microplate reader at 490/525 nm
   (excitation/emission).

4.10. RNA extraction and Real‐Time PCR

   Total RNA was extracted and purified using TRI Reagent (Sigma‐Aldrich)
   according to the manufacturer's instructions. RNA was
   reverse‐transcribed into complementary DNA (cDNA) using the
   High‐Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Cat.
   No. 4368814) with random primers according to the manufacturer's
   instructions. Real‐time PCR analysis was performed in QuantStudio‐3
   Real‐Time PCR system (Applied Biosystems, Thermo Fisher Scientific)
   using SYBR Green Master Mix reagents (Roche) using specific primers for
   the indicated genes. GAPDH was used as a housekeeping gene for
   normalization. Relative mRNA levels were calculated using the ΔΔC [T]
   method. The primers list is provided in [303]Supplementary Table 2
   (Supporting Information).

4.11. RNA sequencing

   RNAseq was performed in independent duplicates on RNA extracted from
   control (pLKO) and PNKP KD (shPNKP) MDA-MB-468 cell line. Total RNA was
   extracted as described above. RNA quality was assessed using the
   Agilent 4200 TapeStation System (Agilent Technologies, Santa Clara,
   CA). RNA‐seq libraries were generated by applying a bulk adaptation of
   the MARS‐seq protocol, as previously described [[304]76]. Libraries
   were sequenced by the Illumina Novaseq 6000 using SP mode 100 cycles
   kit (Illumina). Mapping of sequences to the genome and generation of
   the count matrix was performed by the UTAP pipeline (Weizmann
   Institute). Libraries normalization, filtration of low count genes, and
   discovery of differentially expressed genes were performed using the
   edgeR and Limma packages in R. Gene set enrichment analysis and GSEA
   plots were performed using the FGSEA package.

4.12. Gene knockdown

   shRNA lentivirus–mediated knockdown: lentiviral vectors encoding shRNAs
   of PNKP (TRCN0000050367) and STING1 were purchased (TRCN0000161345)
   from Sigma and prepared as described previously [[305]77].

   For the Tet-ON inducible knockdown: The Tet-pLKO-puro lentiviral vector
   was purchased from Addgene (plasmid #21915;
   [306]http://n2t.net/addgene:21915; RRID: Addgene_21915). The shRNA
   oligonucleotides described below were cloned into the Tet-pLKO-puro
   vector according to the manufacturer's instruction.

   Oligonucleotide sequence for PNKP knockdown:

   Forward:5′CCGGGAAGCGTATGCGGAAGTCAAACTCGAGTTTGACTTCCGCATACGCTTCTTTTTG-3′
   , Reverse:5′AATTCAAAAAGAAGCGTATGCGGAAGTCAAACTCGAGTTTGACTTCCGCATACGCTTC
   3’.

   Cells were transfected with the Tet-shPNKP construct, positive cells
   were selected with puromycin and incubated for 4–5 days with
   doxycycline (1 μg/ml) before experiments conduction.

   For siRNA knockdown: PNKP siRNA (M-006783-02-0005, Dharmacon) was used
   to knockown PNKP, while a non-targeting siRNA (D-001210-05-05,
   Dharmacon) was used as control. Transfection of siRNA was performed
   using DharmaFECT Transfection Reagents (Dharmacon), following the
   manufacturer's instructions.

4.13. Cell lysates and Western Blotting

   Cells were lysed in lysis buffer containing 0.5 % Triton X‐100, 50 mM
   Hepes (pH 7.5), 100 mM NaCl, 1 mM MgCl[2], 50 mM NaF, 0.5 mM NaVO[3],
   20 mM β‐glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml
   leupeptin, and 10 μg/ml aprotinin. For animals-derived tumors, tumor
   tissues were manually chopped with a sterile blade on dry ice, and
   70 mg tissue was homogenized in 400 μl lysis buffer using 4 mm
   stainless steel beads in a homogenizer device (TissueLyser‐LT, QIAGEN)
   for 4 min. Cell lysates were centrifuged at 15 000 rpm for 15 min at
   4 °C, and protein concentration of the supernatants was measured by
   Bradford assay (Bio‐Rad, Hercules, CA). Equal amounts of total protein
   (30–50 μg per sample) were analyzed by WB as previously described
   [[307]77].

4.14. Iron measurement

   Total (Fe^3+ and Fe^2+) and labile (Fe^2+) cellular iron were measured
   using the colorimetric Iron Assay Kit (ab83366, Abcam, Cambridge, UK)
   as described previously [[308]21].

4.15. Coimmunoprecipitation (Co-IP)

   Co-IP was performed as described previously [[309]78] with slight
   modifications. Briefly, cells were lysed in IP-Buffer (1 % NP-40, 20 mM
   Hepes, pH 7.5, 100 mM NaCl, 5 mM MgCl[2], 1 mM PMSF, 10 μg/ml
   leupeptin, and 10 μg/ml aprotinin) and incubated on ice for 30 min.
   Supernatants were prepared by centrifuging cell lysates at 10 000 g for
   15 min at 4 °C. 500 μg of protein sample was incubated with 0.5 μg of
   anti-PNKP antibody (sc-271505) and rotated over-night at 4 °C. For each
   sample, 30 μl of 50 % (v/v) protein A/G–PLUS-Agarose beads (sc-2003)
   were washed twice with PBS and then rotated for 1–2 h with IP-Buffer
   containing 1 % BSA for blocking. The pretreated beads were then mixed
   with the samples, rotated for 1 h at 4 °C and washed three times with
   IP-Washing Buffer (0.1 % Triton X-100, 20 mM Hepes, pH 7.5, 100 mM
   NaCl, 5 mM MgCl[2], 1 mM PMSF, 10 μg/ml leupeptin, and 10 μg/ml
   aprotinin). Samples were then mixed with 2X Laemmli buffer, boiled and
   separated by SDS-PAGE. WB analysis with the relevant antibodies is
   described in the figure legends.

4.16. Cell fractionation

   The nuclear and cytoplasmatic fractions were extracted from
   [MATH: <mrow><mn>1</mn><mo linebreak="goodbreak"
   linebreakstyle="after">×</mo><msup><mn>10</mn><mn>7</mn></msup></mrow>
   :MATH]
   cells/group as we previously described in Ref. [[310]21].

4.17. Fluorescence staining

   Cells grown on glass coverslips in a 24‐wells plate were treated with
   drugs as indicated, washed twice with PBS, fixed in 4 %
   paraformaldehyde (PFA), and processed for IF staining essentially as we
   described previously [[311]21].

   Live cell imaging of lipid peroxidation, lysosomes, mitochondria or
   intracellular iron was performed with BODIPY-C11 (#27086, Cayman
   Chemicals), LysoTracker (Red DND‐99 Invitrogen), MitoTracker
   ([312]M22425, Invitrogen) or FerroOrange (F374, DOJINDO) respectively,
   as described in 10.1002/advs.202307263. Lipid peroxidation
   quantification in live cells was performed by measuring the ratio
   between oxidized (green) and reduced (red) BODIPY-C11 normalized to
   Hoechst as previously reported [[313]21]. Quantification of labile
   Fe^2+ was performed as described in Ref. [[314]21].

4.18. Mitochondrial DNA damage measurement

   Mitochondrial DNA damage quantification was performed as previously
   described [[315]39] with slight modifications. Cells were plated in
   12-wells plate and 24 hr later cells were trypsinized, washed with PBS
   and cell pellets were stored at −80° for later processing. Total DNA
   was extracted using the DNeasy® Blood & Tissue Kit (Qiagen, Hilden,
   Germany). Real‐time PCR was performed with mitochondrial specific
   primers ([316]Supplementary Table 2) using 6 ng of total DNA. qPCR was
   performed as described above. The relative levels of mtDNA were
   calculated using the ΔΔC [T] method by using nuclear actin as
   housekeeping gene, as described in Ref. [[317]79]. DNA leakage in the
   cytoplasm was detected by IF using anti-dsDNA antibody. Cells were
   seeded in 12-well plates on coverslips and 24 hr later transfected with
   siCTRL or siPNKP. After 72 hr, cells were incubated with 70 nM
   MitoTracker for 30 min and processed for IF as described above.
   Localization of dsDNA within or outside the mitochondria was analyzed
   by using an anti-dsDNA antibody (1:25000) and Hoechst.

4.19. TNBC patients derived organoids (PDO)

   Fresh tissues were obtained from patients with TNBC, treated in
   IRCCS-IRE Regina Elena National Cancer Institute. Collected samples
   were maintained in MACS Tissue Storage Solution (130-100-008, Miltenyi
   Biotec) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin
   and 100 μg/ml antimycotic for a maximum of 24 hr at 4 °C. Subsequently,
   the samples were washed twice in PBS, mechanically minced in a Petri
   dish, and the small pieces were transferred to a T75 flask or storage
   at −80 °C tissue into the cryotubes with 1 ml of MACS Freezing solution
   for each cryovial (130-129-552, Miltenyi Biotec). The single-cell
   suspensions were obtained using Tumor Dissociation Kit (130-095-929,
   Miltenyi Biotec), according to the manufacturer's instructions. The
   tissues were digested at 37 °C for 1 hr with occasional pipetting until
   the visible pieces disappeared. The obtained cellular suspension was
   then filtered with a 70 μm strainer (130-098, Miltenyi Biotec).
   Dissociated cell clusters were spun down at 1200 rpm for 5 min, washed
   once with PBS, and spun down again at 1200 rpm for 5 min. Erythrolysis
   was performed with Ammonium Chloride Solution (07850, STEMCELL
   Thecnologies) before the washing step. Dissociated cell clusters were
   resuspended in cold Matrigel (356231, Corning) and seeded in a
   prewarmed 24-well plate at density of 6 × 10^5 cells per 30 μl drops.
   The drops were solidified in a 37 °C and 5 % CO2 incubator for 30 min,
   and then 500 μl organoid culture medium: Advanced DMEM/F-12 (12634010,
   Gibco), 200 mM GlutaMAX (35050038, Gibco), 1 mM HEPES (15630080,
   Gibco), 1X B27 supplement (17504001, Invitrogen), 1X N2 supplement
   (17502001, Invitrogen), Spondin 1 (HZ-1328, HumanKine), 100 ng/mL
   Noggin (HZ-1118, HumanKine), 5 nM Neuregulin (Qk045, HumanKine), 5 μM
   A83-01 (S7692, HumanKine), 50 ng/mL EGF (HZ-1326, HumanKine), 20 ng/mL
   FGF10 (Qk003, HumanKine), 5 ng/mL FGF7 (HZ-1100, HumanKine), 500 nM
   SB202190 (S7067, Sigma Aldrich), 5 μM Y-27632 (HY-10583, HumanKine) was
   added to each well and refreshed every 2–3 days. The organoids were
   passaged every 5–9 days, depending on proliferation rate.

4.20. Organoids treatment, viability assay and morphological parameters
assessment

   In all, 1 × 10^3 cells were seeded into 96-well plates. PDOs were
   cultured in 3 μl of Matrigel and seeded into 96-well Pheno Plate
   (6055300, Revvity), using the Corning Matribot bioprinter. Allowed
   cells and organoids growth, they were treated for 72 h with
   0.12–0.24 μM Doxorubicin (E2516); 20–40 μM A12B4C3 (A8736, Merck),
   drugs combination or DMSO. Cells and PDOs viabilities were assessed
   using ATPlite luminescence assay (Revvity, Massachusetts, USA)
   according to the manufacturer's instructions.

4.21. Animal studies

   For PNKP mouse xenograft combination model, MDA‐MB‐468 cells were
   infected with Tet‐pLKO‐shPNKP lentivirus, selected with puromycin
   (1 μg/ml) and then implanted bilaterally into the mammary fat pads
   (2 × 10^6 per gland) of the fourth inguinal gland of 5–6 weeks old
   female nude mice. When tumor size reached an average of ∼100 mm^3, mice
   were randomized into four groups (n = 5 mice per group). Two naïve
   groups, n = 5 each; naïve ± Doxorubicin. Two Tet-shPNKP groups, n = 5
   each; doxycycline ± Doxorubicin. Naïve mice received regular water,
   while the doxycycline experimental group (Tet-shPNKP) were treated with
   2 mg/mL of doxycycline hyclate (D9891, SIGMA) in drinking water.
   Doxycycline was replaced every 3–4 days. At the second doxycycline
   administration, mice began receiving also doxorubicin intraperitoneally
   (4 mg/kg in HBSS buffer, 100 μl). Mouse body weights and tumor size
   were measured every 4 days. Tumor volume was measured by a digital
   Vernier caliper and calculated according to the (width^2 × length)/2
   formula. At the end of the experiment, mice were euthanized, tumors
   were excised, weighted and documented using a digital camera and
   processed for WB analysis.

4.22. Statistical analysis

   Data are presented as means ± SD. To compare between experimental
   groups, we used Student's t-test (two-sided), unless otherwise stated
   in the figure legends.

CRediT authorship contribution statement

   Avi Maimon: Methodology, Investigation, Formal analysis, Data curation,
   Conceptualization. Pier Giorgio Puzzovio: Methodology, Investigation,
   Formal analysis, Data curation. Yaron Vinik: Formal analysis, Data
   curation. Gavriel-David Hannuna: Data curation, Formal analysis. Sara
   Donzelli: Data curation. Daniela Rutigliano: Data curation. Giovanni
   Blandino: Investigation. Sima Lev: Writing – original draft,
   Supervision, Project administration, Funding acquisition,
   Conceptualization.

Declaration of competing interest

   None.

Ethics approval and consent to participate

   All animal studies were performed according to protocols approved by
   the Weizmann Institutional Animal Care and Use Committee (IACUC
   approval number 05560723-2).

   The Ethical Research Committee of IRCCS-IRE Regina Elena National
   Cancer Institute (Rome, Italy) approved this study. All patients gave
   written informed consent.

Data availability statement

   The RNAseq data performed for this study was deposited in the GEO
   repository (accession number [318]GSE291269).

Funding

   This research was supported by the Ministry of Innovation, Science &
   Technology, Israel and by the Italian Ministry of Foreign Affairs and
   International Cooperation (Italy-Israel), the Israel Science Foundation
   (grant 1564/23), by “the Minerva foundation with funding from the
   Federal German Ministry for Education and Research”, by the
   Weizmann‐Pasteur joint research grant, by research grants from the
   Estate of Thomas Gruen and the Estate of Gertrude Buchler and by Roland
   N. Karlen foundation.

Declaration of competing interest

   The authors declare the following financial interests/personal
   relationships which may be considered as potential competing interests:
   Sima Lev reports financial support was provided by Weizmann Institute
   of Science. Giovanni Blandino reports financial support was provided by
   IRCCS Regina Elena National Cancer Institute. Sima Lev reports a
   relationship with Weizmann Institute of Science that includes:
   employment. Sima Lev has patent No Patent pending to No License. No If
   there are other authors, they declare that they have no known competing
   financial interests or personal relationships that could have appeared
   to influence the work reported in this paper.

Acknowledgements