Purpose:

In search of novel strategies to improve the outcome of advanced prostate cancer, we considered that prostate cancer cells rearrange iron homeostasis, favoring iron uptake and proliferation. We exploited this adaptation by exposing prostate cancer preclinical models to high-dose iron to induce toxicity and disrupt adaptation to androgen starvation.

Experimental Design:

We analyzed markers of cell viability and mechanisms underlying iron toxicity in androgen receptor–positive VCaP and LNCaP, castration-resistant DU-145 and PC-3, and murine TRAMP-C2 cells treated with iron and/or the antiandrogen bicalutamide. We validated the results in vivo in VCaP and PC-3 xenografts and in TRAMP-C2 injected mice treated with iron and/or bicalutamide.

Results:

Iron was toxic for all prostate cancer cells. In particular, VCaP, LNCaP, and TRAMP-C2 were highly iron sensitive. Toxicity was mediated by oxidative stress, which primarily affected lipids, promoting ferroptosis. In highly sensitive cells, iron additionally caused protein damage. High-basal iron content and oxidative status defined high iron sensitivity. Bicalutamide–iron combination exacerbated oxidative damage and cell death, triggering protein oxidation also in poorly iron-sensitive DU-145 and PC-3 cells.

In vivo, iron reduced tumor growth in TRAMP-C2 and VCaP mice. In PC-3 xenografts, bicalutamide–iron combination caused protein oxidation and successfully impaired tumor expansion while single compounds were ineffective. Macrophages influenced body iron distribution but did not limit the iron effect on tumor expansion.

Conclusions:

Our models allow us to dissect the direct iron effect on cancer cells. We demonstrate the proof of principle that iron toxicity inhibits prostate cancer cell proliferation, proposing a novel tool to strengthen antiandrogen treatment efficacy.

Translational Relevance

The negative outcome of advanced prostate cancer remains a major clinical problem because virtually all patients exposed to long-term androgen receptor modulation and androgen deprivation therapy becomes refractory and enters into the castration-resistant phase. Novel therapeutic strategies aimed at optimizing drug efficacy and overcoming resistance represent an urgent need. We propose to exploit iron toxicity to inhibit cancer cell growth. We show the molecular rationale that supports combination of iron at high concentration with androgen inhibitors, an approach that might intersect cancer cells redox homeostasis, maximizing the metal toxicity. Importantly, iron effect is observed with a commercially available iron compound, which we tested in preclinical studies at the dose used in patients. Altogether, our results pave the way to pilot clinical studies aimed at confirming the effect of high-dose iron supplementation in different types and stages of prostate cancer.

Prostate cancer is one of the most frequent cancers diagnosed in the United States and Europe and the third leading cause of cancer deaths (1). Roughly 10% of men treated with curative intent will develop metastasis over time (2). Current therapy for metastatic disease is based on androgen deprivation and/or androgen receptor (AR) modulation. However, virtually all patients exposed to long-term treatment become refractory and enter into the castration-resistant prostate cancer (CRPC) phase (3). The complete understanding of CRPC evolution and the setup of novel therapeutic approaches to overcome resistance are unmet clinical needs. In this context, iron might be a pivotal player. Iron homeostasis has been shown to play a role in prostate cancer onset and progression (4), an event observed in other tumors, like breast cancer and multiple myeloma (5, 6).

Iron is a fundamental nutrient, although a strong prooxidant when in excess. Cellular iron balance is maintained by the coordinated regulation of the storage protein ferritin and iron transporters, including the importer transferrin receptor-1 (TfR1) and the exporter ferroportin (7). Macrophages, professional iron-storing cells, detoxify iron excess, and recycle iron through ferroportin. Iron also accumulates in hepatocytes, which, according to body iron needs, modulate the expression of the hormone hepcidin that binds and degrades ferroportin, favoring macrophage and hepatocyte iron retention and blocking intestinal iron absorption (7).

In prostate cancer, local hepcidin synthesis degrades ferroportin through an autocrine mechanism, favoring iron retention and cancer growth (4). Increasing levels of TfR1 and progressive ferroportin downregulation have been reported in the transformation from primary tumors to metastatic cancer and in prostate cancer cell lines (4). However, while iron favors the proliferation of prostate cancer cells under physiologic availability, it may become an Achilles' heel in condition of iron excess, because iron balance is essential to cancer as well as to normal cells. Iron excess causes detrimental reactive oxygen species (ROS) increase that activates cell death programs, including ferroptosis, a form of iron-dependent death induced by excessive formation of lipid hydroperoxides (8, 9). We have previously explored this strategy in models of multiple myeloma, a monoclonal B-cell disorder, where malignant plasma cells rearrange iron-trafficking proteins to favor iron uptake and retention (6, 10). We reported that iron burden impairs myeloma cells proliferation, inducing lipid damage and cell death, and increases the therapeutic efficacy of proteasome inhibitor bortezomib, both in vitro and in vivo preclinical models (11–13).

Iron excess might contrast tumor growth also indirectly by affecting microenvironment. Interestingly, tumor-associated macrophages loaded with iron shift from a prosurvival to a proinflammatory phenotype that negatively affects tumor expansion in cancer models (14–16).

Rearrangement of iron trafficking in favor of uptake occurs in prostate cancer (4); however, whether iron overload–dependent modifications of the tumor and its microenvironment take place in prostate cancer has not been addressed yet. This would be key to propose iron supplementation as a potential innovative therapeutic tool. Here, we hypothesized that iron is toxic for prostate cancer cells and used cellular and murine models to explore the underlying mechanisms. Moreover, considering the relevance of AR inhibition in the setting of prostate cancer treatment, we showed in the same models the proof of principle of a novel therapeutic approach based on iron in combination with bicalutamide, a representative and widely studied androgen inhibitor.

Cell culture

VCaP (ATCC, catalog no. CRL-2876, RRID:CVCL_2235), LNCaP (ATCC, catalog no. CRL-3313, RRID:CVCL_4783), DU-145 (ATCC, catalog no. HTB-81, RRID:CVCL_0105), and PC-3 (ATCC, catalog no. CRL-1435, RRID:CVCL_0035) were cultured in RPMI1640 (Lonza, catalog no. 12-115Q); TRAMP-C2 (ATCC, catalog no. CRL-2731, RRID:CVCL_3615) in DMEM. Media were supplemented with 2 mmol/L l-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin, and 10% heat-inactivated FBS (Thermo Fisher Scientific, catalog no. 10270106). All cells were tested negative for Mycoplasma.

To test cell proliferation, cells (5 × 104/well) were seeded in 6-well plates and grown in presence or absence of 1–600 μmol/L ferric ammonium citrate (FeAC; Sigma-Aldrich, catalog no. 1185-57-5) or ferric nitrilo triacetate (FeNTA; BOC Sciences, catalog no. 16448-54-7) and/or 2,5–10 μmol/L bicalutamide (Sigma-Aldrich, catalog no. B9061). At 96 hours, adherent and floating cells were collected, resuspended in PBS, diluted 1:1 with trypan blue exclusion dye (Bio-Rad, catalog no. 1450013), and counted using TC20 Automated Cell Counter (Bio-Rad, catalog no. 145-0102). Alternatively, cells (7 × 105) were plated in 60 mm Petri dishes and treated with FeAC, bicalutamide or bicalutamide–FeAC for 24 hours. Then, cell death was measured by trypan blue assay and cell pellets processed for analysis.

For gene overexpression, prostate cancer cells were transfected with plasmids pCMV3-NFE2L2-OFPSpark (Sino Biologicals, catalog no. HG1784-ACR) or pUC19-hLPCAT3 (Sino Biologicals, catalog no. HG20457-U) or pCMV3-AKR1C2-Myc (Sino Biologicals, catalog no. HG14175-CM) or with pCMV3 or pUC19 empty vectors by using Lipofectamine 2000 reagent (Invitrogen, catalog no. 18324-111). After 24 hours, cells were collected, replated (7 × 105) in 60 mm Petri dishes, and treated with 200–600 μmol/L FeAC for 24 hours. Finally, cell death was measured by trypan blue assay. Gene overexpression was confirmed by qRT-PCR at 24 hours after transfection.

Mouse studies

Mice were bred in the pathogen-free animal facility of the IRCCS Ospedale San Raffaele in accordance with the European Union guidelines. The study was approved by the local Institutional Animal Care and Use Committee. TRAMP-C2 (2.5 × 106) cells were subcutaneously injected into the right flank of C57BL/6-wild-type (Charles River). VCaP (2 × 106) or PC-3 (2 × 106) cells were subcutaneously injected into the right flank of immunodeficient BALB/C-RAG2−/−γc−/− mice. Mice were randomized to treatments when tumor masses reached approximately 100 to 300 mm3 (day 0). As iron source, mice received either a single intraperitoneal dose of 250 mg/kg iron-dextran (Sigma-Aldrich, catalog no. D8517) on day 1 or 2 consecutive intravenous injections of 20 mg/kg ferric carboxymaltose (Ferinject, Vifor-Pharma) on days 1 and 8. Xenografts were additionally randomized to receive bicalutamide (15 mg/kg, p.o.) daily as single agent, starting from (day 0) or bicalutamide–iron combination, as detailed in Results and figure legends. In experiments aimed at deplating macrophages, PC-3 xenografts were treated with 200 μL intravenous of Clodrolip (Liposoma BV) or PBS liposome (Liposoma BV) at 3 days of interval, starting from (day 4) before bicalutamide–iron dextran or saline treatment.

Tumor volume was calculated using the formula: Volume = {\frac{4}{3}} \times \pi \times {\frac{{length}}{2}} \times {\frac{{width}}{2}}$⁠. At necropsy, tumors, livers, and spleens were collected for histologic analysis, iron content determination, and Western blotting. Tissue iron content was determined as described in ref. 17 and detailed in Supplementary Data.

Iron and oxidative status

Lipid peroxidation was evaluated using the Lipid Peroxidation Assay Kit (Sigma-Aldrich, catalog no. MAK085). Oxidized proteins were assayed by Western blotting using the Oxyblot Kit (Merck-Millipore, catalog no. S7150). Labile iron pool was measured using the iron-sensitive fluorescent probe calcein-acetoxymethyl ester (Invitrogen, catalog no. C1430). Ferritin H and ferritin L were quantified by ELISA calibrated on the human FTH and FTL homopolymers using the rH02 and LF03 mAbs (kindly provided by Sonia Levi, Vita-Salute San Raffaele University, Milan, Italy), as described previously (18). All procedures are detailed in Supplementary Data.

Histologic analysis

Tumor tissues were fixed in formalin and embedded in paraffin and processed following standard procedures. Tumor sections were stained with Prussian Blue Test Kit (Sigma Aldrich, catalog no. HT20) for the detection of iron storage, following manufacturer's instructions. Alternatively, sections were incubated with anti-KI67 SP6 (Biocare Medical, catalog no. CRM325C, RRID: AB_2721189) or anti-F4/80 CI:A3–1 (Bio-Rad, catalog no. MCA497, RRID: AB_2098196) antibodies. Then, slides were washed and incubated with Expose Rabbit specific HRP/DAB detection IHC Kit (Abcam, catalog no. ab80437). Sections were evaluated using an Nikon Eclipse 55i optical microscope equipped with a DS-L1 camera.

Gene expression analysis

Gene expression was measured by qRT-PCR using the TaqMan (Applied Biosystem, catalog o. 4369016) in PC-3 ex vivo or the Syber-Green Gene Expression Master Mix (Applied Biosystem, catalog no. 4309155). Gene expression was estimated relative to ACTB (PC-3) or Gapdh (macrophages) expression following the formula 2−ΔCt. RNA extraction and retrotrascription methods, TaqMan probes Ids, and oligo sequences are detailed in Supplementary Data.

Statistical analyses

Statistical differences were determined as indicated in figure legends. Student t test and Tukey post ANOVA test were performed using GraphPad Prism Version 7.0a.

Iron reduces prostate cancer cell proliferation and potentiates bicalutamide efficacy, promoting oxidative damage

We explored the effect of iron administration, alone or in combination with bicalutamide, in human AR-positive VCaP and LNCaP cells, and in AR-negative DU-145 and PC-3 cells. Treatment with iron in low concentration (FeAC 1 μmol/L) promoted cell growth whereas at elevated dose (FeAC or FeNTA 200–600 μmol/L) iron became progressively toxic for all cell lines, reducing cell proliferation in comparison with untreated cells (Fig. 1A; Supplementary Fig. S1). Among the cell lines analyzed, VCaP and LNCaP cells were more iron sensitive than DU-145 and PC-3 (Fig. 1A).

Figure 1.

Iron reduces prostate cancer cell proliferation and potentiates bicalutamide efficacy, promoting oxidative damage. A, Human VCaP, LNCaP, DU-145, and PC-3 cells were treated with a titrated dose of Ferric ammonium citrate (FeAC) and cell proliferation measured after 96-hour growth. B, Cell proliferation measured in prostate cancer cells after 96-hour growth in presence of 200 μmol/L FeAC and/or 2,5–5 μmol/L bicalutamide (Bic). C–H, VCaP and PC-3 cells were treated with FeAC and/or bicalutamide at the indicated concentration for 24 hours. C and D, Percentage of trypan blue positive cells. E and F, Malondialdehyde (MDA) levels. G and H, Polycarbonylated (Oxy-proteins) proteins levels. Left, Representative immunoblotting. Treatment with hydrogen peroxide (100 μmol/L) was used as positive control. Right, Summary of densitometric analysis. Data are shown as means ± SEs of at least five (A and B) or three (C–H) independent experiments for every cell line. Statistical differences were determined by Tukey post one-way ANOVA test. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Figure 1.

Iron reduces prostate cancer cell proliferation and potentiates bicalutamide efficacy, promoting oxidative damage. A, Human VCaP, LNCaP, DU-145, and PC-3 cells were treated with a titrated dose of Ferric ammonium citrate (FeAC) and cell proliferation measured after 96-hour growth. B, Cell proliferation measured in prostate cancer cells after 96-hour growth in presence of 200 μmol/L FeAC and/or 2,5–5 μmol/L bicalutamide (Bic). C–H, VCaP and PC-3 cells were treated with FeAC and/or bicalutamide at the indicated concentration for 24 hours. C and D, Percentage of trypan blue positive cells. E and F, Malondialdehyde (MDA) levels. G and H, Polycarbonylated (Oxy-proteins) proteins levels. Left, Representative immunoblotting. Treatment with hydrogen peroxide (100 μmol/L) was used as positive control. Right, Summary of densitometric analysis. Data are shown as means ± SEs of at least five (A and B) or three (C–H) independent experiments for every cell line. Statistical differences were determined by Tukey post one-way ANOVA test. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Close modal

Bicalutamide, used as single agent, reduced VCaP and LNCaP cells proliferation while, as expected, was ineffective on DU-145 and PC-3 cells (Fig. 1B). Bicalutamide–iron combination impaired cell growth more than each individual treatment in all cell lines analyzed (Fig. 1B). Notably, in DU-145 and PC-3 cells, this occurred using iron and bicalutamide concentrations that were ineffective in single-agent treatment (Fig. 1B). We confirmed a similar pattern of results in LNCaP and PC-3 cells by measuring cell viability (Supplementary Fig. S2).

To dissect the mechanisms of iron toxicity, we analyzed cells treated with iron, bicalutamide, or combination for 24 hours. Iron concentrations that impaired proliferation in experiments described above (i.e., 200 μmol/L FeAC in VCaP or 600 μmol/L in PC-3) caused cell death in all cell lines (Fig. 1C and D; Supplementary Fig. S3A and S3B). All iron-treated cells showed accumulation of malondialdehyde, the product of lipid peroxidation and a recognized marker of ferroptosis (Fig. 1E and F; Supplementary Fig. S3C and S3D). In highly iron-sensitive VCaP and LNCaP cells, iron caused additional oxidative damage to proteins, as shown by accumulation of polycarbonylated peptides (Fig. 1G and H; Supplementary Fig. S3E and S3F). Bicalutamide used as single agent at 2.5 μmol/L caused evident cell death in androgen-dependent VCaP and LNCaP cells, whereas DU-145 and PC-3 became partially sensitive only at higher bicalutamide concentration (10 μmol/L; Fig. 1C and D; Supplementary Fig. S3A and S3B). Bicalutamide toxicity occurred without promoting oxidative damage in virtually all cells (Fig. 1EH; Supplementary Fig. S3C–S3F); we observed an increase of polycarbonylated proteins only in VCaP cells (Fig. 1G). Bicalutamide–iron combination was more toxic than each individual treatment, inducing cell death with enhanced efficacy (Fig. 1C and D; Supplementary Fig. S3A and S3B). The combination caused a minimal if any increase of lipid damage over the high malondialdehyde accumulation caused by iron, while exacerbated the accumulation of oxidized proteins (Fig. 1EH; Supplementary Fig. S3C–S3F). The latter event was particularly evident in DU-145 and PC-3, where iron used as single agent did not increase basal polycarbonylated proteins levels (Fig. 1H; Supplementary Fig. S3F). In VCaP and PC-3, neither iron alone nor bicalutamide–iron combination induced procaspase-3 or PARP-1 cleavage in both cell lines, excluding that the experimental conditions triggered apoptosis (Supplementary Fig. S4).

We also tested the effect of iron on TRAMP-C2 cells, a cell line originated from murine prostate cancer (19). Used as single agents, iron and bicalutamide impaired murine TRAMP-C2 cells proliferation with efficacy similar to that observed in VCaP and LNCaP cells (Supplementary Fig. S5). We confirmed that the bicalutamide–iron combination had stronger effect than individual treatments and that iron caused lipid damage and cell death in murine cells (Supplementary Fig. S5).

We concluded that iron toxicity is mediated by oxidative stress, primarily affecting lipids, and that combination iron-bicalutamide exacerbates oxidative damage mainly by triggering protein oxidation.

Basal iron and oxidative status define prostate cancer cell sensitivity to iron

We then aimed at dissecting the mechanisms of the diverse iron sensitivity of human prostate cancer cells. The comparison between highly sensitive (VCaP and LNCaP) and poorly sensitive (DU-145 and PC-3) cells revealed that, in basal growth conditions, the former had higher free-iron content, the so-called labile iron pool, which is the cell bioactive and potentially prooxidant iron fraction (Fig. 2A). Labile iron levels were opposite to levels of ferritin, the protein responsible of free-iron sequestration and safe iron storage (Fig. 2B). More in detail, VCaP and LNCaP cells showed lower levels of the catalytic active H ferritin subunit than DU-145 and PC-3 cells (Fig. 2B). Moreover, the highest iron-sensitive VCaP cells showed the lowest H ferritin levels (Fig. 2B). In parallel with higher levels of free iron, VCaP and LNCaP cells showed higher malondialdehyde levels than DU-145 and PC-3 in basal condition (Fig. 2C).

Figure 2.

Basal iron and oxidative status define cell sensitivity to iron exposure. A–C, Iron and lipid oxidation status measured in basal growth condition in VCaP, LNCaP, DU-145, and PC-3 cells. A, Labile iron pool. B, Ferritin subunits H and L protein content. Statistical difference are referred to ferritin H. C, Malondialdehyde (MDA) levels. D, LNCaP cells were transfected with plasmid encoding NFE2L2 (NRF2) or empty plasmid and treated with 200 μmol/L FeAC for 24 hours. Left, NFE2L2 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. E, PC-3 cells were transfected with plasmid encoding LPCAT3 or empty plasmid and treated with 600 μmol/L FeAC for 24 hours. Left, LPCAT3 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. F and G, VCaP, LNCaP, DU-145, and PC-3 cells were treated with FeAC at the indicated concentration for 24 hours. F, Ferritin H protein content. G, Labile iron pool levels. Data are shown as means ± SEs of at least three independent experiments. Statistical differences were determined by: Tukey post two-way ANOVA test in B; Student t test in left panels of D and E; Tukey post one-way ANOVA test in all the other panels. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Figure 2.

Basal iron and oxidative status define cell sensitivity to iron exposure. A–C, Iron and lipid oxidation status measured in basal growth condition in VCaP, LNCaP, DU-145, and PC-3 cells. A, Labile iron pool. B, Ferritin subunits H and L protein content. Statistical difference are referred to ferritin H. C, Malondialdehyde (MDA) levels. D, LNCaP cells were transfected with plasmid encoding NFE2L2 (NRF2) or empty plasmid and treated with 200 μmol/L FeAC for 24 hours. Left, NFE2L2 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. E, PC-3 cells were transfected with plasmid encoding LPCAT3 or empty plasmid and treated with 600 μmol/L FeAC for 24 hours. Left, LPCAT3 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. F and G, VCaP, LNCaP, DU-145, and PC-3 cells were treated with FeAC at the indicated concentration for 24 hours. F, Ferritin H protein content. G, Labile iron pool levels. Data are shown as means ± SEs of at least three independent experiments. Statistical differences were determined by: Tukey post two-way ANOVA test in B; Student t test in left panels of D and E; Tukey post one-way ANOVA test in all the other panels. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Close modal

We then analyzed public datasets of gene expression (Gene Expression Omnibus series) of representative LNCaP and PC-3 cells focusing on genes known to be primarily involved in ferroptosis pathways. The analysis showed that, in basal growth condition, LNCaP have higher expression of genes involved in cysteine supply and oxidized lipid detoxification than PC-3 cells, including SLC7A11, CBS, and GPX4, consistent with the requirement of strong antioxidant capacity for survival (Supplementary Fig. S6). In addition, datasets showed higher expression of prosurvival genes, as FANCD2, TP53, HMGCR, and MT1G in LNCaP than in PC3 (Supplementary Fig. S6).

Overall, our results uncover the need of VCaP and LNCaP cells to face high levels of oxidative stress in basal growth condition, because of high free-iron content and according the high rate of mitochondrial respiration, a feature of androgen-dependent prostate cancer cells (20, 21). In this condition, the ability of VCaP and LNCaP cells to buffer additional oxidative stress is likely limited, explaining their enhanced sensitivity to iron-induced damage and ferroptosis compared with DU-145 and PC-3. In accordance with this conclusion, LNCaP cells were more sensitive than PC-3 to ferroptosis stimuli generated by erastin, which impairs cysteine uptake and glutathione production (8), thus limiting the antioxidant capacity (Supplementary Fig. S7).

In addition, datasets analysis showed traits that might directly predispose LNCaP cells to high iron sensitivity. Compared with PC-3, LNCaP had higher expression of LPCAT3 that potentially favors the insertion of oxidation-sensitive polyunsaturated fatty acids in membrane phospholipids; higher expression of ALOX15 that codifies for a lipoxygenase family member specifically capable of oxygenating membrane fatty acids; lower expression of NFE2L2, the master gene of antioxidant response recently proved to be specifically activated by iron-induced oxidative stress (22) (Supplementary Fig. S6). Consistent with these observations, overexpression of NFE2L2 protected LNCaP cells from iron-induced cell death (Fig. 2D). On the opposite, overexpression of LPCAT3 in PC-3 sensitized cells to iron toxicity (Fig. 2E).

The analysis of ferritin and labile iron pool after iron exposure revealed additional differences. As expected, iron stimulated H ferritin increase in all cells (Fig. 2F), although H ferritin levels persisted lower in VCaP and LNCaP than in DU-145 and PC-3 cells considering that the fold change increase was comparable among all cell lines (Fig. 2F). Consequently, although cells were exposed to the same iron concentration (200 μmol/L), VCaP and LNCaP cells showed higher free-iron content than DU-145 and PC-3 cells (Fig. 2G). Increasing iron concentration up to 600 μmol/L in DU-145 and PC-3 did not further induce ferritin H upregulation over that promoted by 200 μmol/L, and free-iron content increased, making cells susceptible to iron toxicity (Fig. 2F and G).

In summary, our data suggest that multiple factors contribute to iron sensitivity. High basal iron content and oxidative stress, together with low capacity to buffer additional iron identify highly iron-sensitive prostate cancer cells.

Bicalutamide exacerbates iron-induced oxidative damage by promoting aldo-keto reductases expression

We addressed the question of how bicalutamide interferes with iron and redox homeostasis focusing on LNCaP and PC-3 as representative cell lines. First, we excluded that bicalutamide alters the homeostatic cell response to iron excess because the coordinated upregulation of ferritin H and downregulation of TfR1, which typically occurs in iron-loaded cells, were maintained upon bicalutamide treatment (Supplementary Fig. S8).

Looking for a possible link between redox balance and androgens metabolism, we focused on two aldo-keto reductases, AKR-1C1 and -1C2, enzymes that play a pivotal role in modulating steroid hormones functionality in prostate tissues (23) and, as reductases, can potentially cooperate in the detoxification of aldehydes produced under oxidative stress (24–26). Bicalutamide promoted the expression of AKR-1C1 and -1C2 in both LNCaP and PC-3 cells, an antioxidant response–independent effect, because bicalutamide did not induce oxidative stress, as reported above and confirmed by unchanged HMOX1 expression (Fig. 3A and B). Treatment with FeAC upregulated both AKR-1C1 and -1C2 in LNCaP, where iron induced multioxidative injuries, but not in PC-3, where oxidative damage was limited to lipid compartment, suggesting that AKRs were upregulated under oxidative stress but did not represent a first-line response (Fig. 3A and B). Expression levels of AKR-1C1 and -1C2 in cells exposed to bicalutamide–iron were similar to those of bicalutamide-treated cells and higher than those of cells treated with iron alone (Fig. 3A). In parallel, bicalutamide–iron treatment increased HMOX1 expression over the levels caused by iron, confirming a supplementary oxidative stress promoted by the combination (Fig. 2B). On the basis of AKR1Cs expression dynamics, we hypothesized that AKR1Cs induced by bicalutamide are not balanced with the antioxidant response and might operate as oxidases, as proposed previously (27), exacerbating rather than counteracting oxidative damages upon iron exposure. To validate this idea, we overexpressed AKR1C2 in PC-3 cells, as representative of AKR1Cs enzymes, and treated cells with 600 μmol/L FeAC, confirming that AKR1C2 overexpression sensitizes cells to iron toxicity (Fig. 3C).

Figure 3.

Bicalutamide exacerbates iron-induced oxidative damage by promoting aldo-keto reductases expression. A and B, LNCaP and PC-3 cells were treated with 200–600 μmol/L FeAC and/or 2,5–10 μmol/L bicalutamide (Bic) for 24 hours. Aldo ketoreductases -1C1 and -1C2 (AKR-1C1 and -1C2) and heme oxygenase 1 (HMOX1) mRNA levels measured after 24-hour treatment. Graphs indicate fold change relative to untreated cells. C, PC-3 cells were transfected with plasmid encoding AKR1C2 or empty plasmid and treated with 600 μmol/L FeAC for 24 hours. Left, AKR1C2 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. Data are presented as means ± SEs of five in (A and B) or three in (C) independent experiments. Statistical differences were determined by Student t test. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Figure 3.

Bicalutamide exacerbates iron-induced oxidative damage by promoting aldo-keto reductases expression. A and B, LNCaP and PC-3 cells were treated with 200–600 μmol/L FeAC and/or 2,5–10 μmol/L bicalutamide (Bic) for 24 hours. Aldo ketoreductases -1C1 and -1C2 (AKR-1C1 and -1C2) and heme oxygenase 1 (HMOX1) mRNA levels measured after 24-hour treatment. Graphs indicate fold change relative to untreated cells. C, PC-3 cells were transfected with plasmid encoding AKR1C2 or empty plasmid and treated with 600 μmol/L FeAC for 24 hours. Left, AKR1C2 mRNA levels measured before iron treatment start. Right, Percentage of trypan blue positive cells after iron treatment. Data are presented as means ± SEs of five in (A and B) or three in (C) independent experiments. Statistical differences were determined by Student t test. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Close modal

Iron at high dose reduces tumor growth in vivo

To prove the effect of iron in vivo, we evaluated the result of iron administration on tumor growth in VCaP xenografts and mice bearing TRAMP-C2 cells. The latter group of mice was randomized to receive either a single dose of iron-dextran (250 mg/kg) or two consecutive injections of ferric carboxymaltose (20 mg/kg; Supplementary Fig. S9), iron formulations widely used in murine preclinical studies and clinical practice, respectively. Iron supplementation reduced tumor growth in both experimental settings, with similar efficacy (Supplementary Fig. S9A). Analysis of body iron distribution at necropsy, showed that iron preferentially accumulated in the physiologic iron storage organs, liver and spleen, but was also evident within tumor masses (Supplementary Fig. S9B). Perls' staining of tumor sections revealed a number of blue positive iron-loaded macrophages, distributed along the margins and inside the tumor masses (Supplementary Fig. S9C). In accordance with the different iron dose used, iron accumulation in storage organs, tumors, and macrophages was higher in iron-dextran than in ferric carboxymaltose–treated mice.

Then, we tested the ferric carboxymaltose bicalutamide–iron combination in VCaP xenografts. Mice were randomized to receive two consecutive injections of ferric carboxymaltose 20 mg/kg or bicalutamide 15 mg/kg daily (a human equivalent dose) or bicalutamide–iron combination. Both iron and bicalutamide used as single agents reduced tumor growth compared with saline-treated mice (Fig. 4A; Supplementary Fig. S10). Iron alone caused stronger impairment of tumor growth than bicalutamide (Fig. 4A; Supplementary Fig. S10) making difficult to appreciate any further effect of the antiandrogen addition (Fig. 4A; Supplementary Fig. S10). As expected, iron content was increased in tumors of iron-treated mice while extra iron accumulated in liver and spleen (Fig. 4B).

Figure 4.

Iron at high dose reduces tumor growth in vivo. RAG2−/−γc−/− mice were subcutaneously injected with VCaP cells. When tumor masses reached 100–300 mm3, mice were treated with saline (N = 4) or daily with 15 mg/kg bicalutamide (Bic; N = 4) or with two consecutive injections of 20 mg/kg ferric carboxymaltose (FeCM; N = 4) or with bicalutamide–iron combination (Bic+FeCM; N = 4). A, Tumor growth. B, Liver, spleen, and tumor iron content measured in ex vivo samples. Statistical differences were determined by Tukey post two-way ANOVA test in A and by Tukey post one-way ANOVA test in B. ns, nonstatistically significant. *, P < 0.05. ***, P < 0.001. C, Representative tumor sections stained with anti-Ki-67 antibody, showing brown positive nuclei of actively proliferating cells. Red arrows point to representative tumor areas characterized by cells with dark condensed Ki-67–negative nuclei and fragmented cytoplasm, resembling dead cells. Original magnification: 100×, 200×, and 400×. D, Representative Perls' stain of tumor sections. Black arrows point to representative iron positive macrophages. Original magnification: 200×.

Figure 4.

Iron at high dose reduces tumor growth in vivo. RAG2−/−γc−/− mice were subcutaneously injected with VCaP cells. When tumor masses reached 100–300 mm3, mice were treated with saline (N = 4) or daily with 15 mg/kg bicalutamide (Bic; N = 4) or with two consecutive injections of 20 mg/kg ferric carboxymaltose (FeCM; N = 4) or with bicalutamide–iron combination (Bic+FeCM; N = 4). A, Tumor growth. B, Liver, spleen, and tumor iron content measured in ex vivo samples. Statistical differences were determined by Tukey post two-way ANOVA test in A and by Tukey post one-way ANOVA test in B. ns, nonstatistically significant. *, P < 0.05. ***, P < 0.001. C, Representative tumor sections stained with anti-Ki-67 antibody, showing brown positive nuclei of actively proliferating cells. Red arrows point to representative tumor areas characterized by cells with dark condensed Ki-67–negative nuclei and fragmented cytoplasm, resembling dead cells. Original magnification: 100×, 200×, and 400×. D, Representative Perls' stain of tumor sections. Black arrows point to representative iron positive macrophages. Original magnification: 200×.

Close modal

At histologic examination, tumor masses of iron-supplemented mice showed macro areas of aggregates of tumor cells with dark condensed nuclei and fragmented cytoplasm, negative to Ki-67 proliferation marker, likely representing dead cells (Fig. 4C). These areas were present at lower extent in tumors from mice treated with bicalutamide alone while absent in saline-treated controls (Fig. 4C). In the remaining tumor regions of iron-treated mice, Ki-67–positive nuclei were fewer than in saline-treated counterparts (Fig. 4C). Iron-loaded macrophages were present both at margins and inside the tumors (Fig. 4D).

Iron administration synergizes with bicalutamide in PC-3 xenografts

The effect of bicalutamide–iron combination was more clearly evaluated in xenografts of PC-3 cells that are poorly iron sensitive. In a first set of experiments, mice were randomized to receive a single dose of iron-dextran (250 mg/kg) or bicalutamide (15 mg/kg) daily or bicalutamide–iron combination (Fig. 5A). Tumor growth was not affected by bicalutamide or iron alone, but combined treatment reduced tumor expansion, compared with both saline and single agents (Fig. 5A). As expected, iron was stored in liver, spleen, and tumor tissues (Supplementary Fig. S11A). Perls' staining of tumor sections showed positive iron staining in macrophages (Supplementary Fig. S11B). Ex vivo samples from tumor masses of iron-treated mice showed downregulation of TFRC (TfR1) expression compared with samples from bicalutamide- or saline-treated mice, indicating that cancer cells had uptaken iron (Fig. 5B), although accumulation was lower than in macrophages and did not reach Perls' staining sensitivity. As for oxidative damage, iron as a single agent did not cause polycarbonylated proteins accumulation, whereas bicalutamide–iron combination caused strong oxidative damage to proteins, in accordance with in vitro results (Fig. 5C).

Figure 5.

Iron administration synergizes with bicalutamide in PC-3 xenografts. A–C,RAG2−/−γc−/− mice were subcutaneously injected with PC-3 cells. When tumor masses reached 100–300 mm3, mice were treated with saline (N = 5) or a single injection of 250 mg/kg iron-dextran (FeDe; N = 3) or daily with 15 mg/kg bicalutamide (Bic; N = 6) or with Bic+FeDe combination (N = 8). A, Left, Scheme of the experiment. Middle, Tumor growth. Right, Tumor volume measured at necropsy. B, Human transferrin receptor 1 (TFRC) mRNA levels measured in ex vivo tumor samples. C, Left, Representative Western blotting showing polycarbonylated proteins levels measured in ex vivo tumor samples. Right, Summary of densitometry quantification made on the entire pool of mice. D, PC-3 xenografts were treated with saline (N = 5) or two consecutive injections of 20 mg/kg ferric carboxymaltose (FeCM; N = 3) or daily with 15 mg/kg bicalutamide (N = 5) or with Bic+FeCM combination (N = 5). Left, Scheme of the experiment. Middle, Tumor growth. Right, Tumor volume measured at necropsy. Statistical differences were measured by Tukey post two-way ANOVA test in central panels of (A and D) and by Tukey post one-way ANOVA test in all the other panels. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Figure 5.

Iron administration synergizes with bicalutamide in PC-3 xenografts. A–C,RAG2−/−γc−/− mice were subcutaneously injected with PC-3 cells. When tumor masses reached 100–300 mm3, mice were treated with saline (N = 5) or a single injection of 250 mg/kg iron-dextran (FeDe; N = 3) or daily with 15 mg/kg bicalutamide (Bic; N = 6) or with Bic+FeDe combination (N = 8). A, Left, Scheme of the experiment. Middle, Tumor growth. Right, Tumor volume measured at necropsy. B, Human transferrin receptor 1 (TFRC) mRNA levels measured in ex vivo tumor samples. C, Left, Representative Western blotting showing polycarbonylated proteins levels measured in ex vivo tumor samples. Right, Summary of densitometry quantification made on the entire pool of mice. D, PC-3 xenografts were treated with saline (N = 5) or two consecutive injections of 20 mg/kg ferric carboxymaltose (FeCM; N = 3) or daily with 15 mg/kg bicalutamide (N = 5) or with Bic+FeCM combination (N = 5). Left, Scheme of the experiment. Middle, Tumor growth. Right, Tumor volume measured at necropsy. Statistical differences were measured by Tukey post two-way ANOVA test in central panels of (A and D) and by Tukey post one-way ANOVA test in all the other panels. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Close modal

In a second group of mice, we tested the effect of ferric carboxymaltose, randomizing mice to two consecutive injections of iron (20 mg/kg), bicalutamide or bicalutamide–iron combination (Fig. 5D). As expected, tumor growth was not affected by iron or bicalutamide used as single agents, but it was reduced by bicalutamide–iron combination (Fig. 5D). Although based on reduced iron dose, this schedule of iron–bicalutamide combination was sufficient to increase tumor iron content (Supplementary Fig. S11C) and to impair PC-3 cells proliferation.

Altogether, our experiments show that iron supplementation efficiently impairs in vivo growth of iron-sensitive cells and in combination with bicalutamide reduces proliferation of iron-insensitive and castration-resistant cells.

Macrophages dictate body iron distribution but do not limit the iron effect

Macrophages are professional iron-storing cells that may influence both iron distribution and tumor expansion (14, 15). Iron accumulation in macrophages is predicted to drive M1-like polarization (14–16, 28). In vitro, addition of 300 μmol/L FeAC to murine bone marrow–derived macrophages treated with IFNγ or IL4 or medium conditioned by LNCaP or PC-3 cells, promoted the expression of M1-like markers, such as Tnf, in all conditions, suggesting that iron might override the effect of cytokines that usually induce the alternative M2-like polarization, as IL4 or other mediators potentially released by prostate cancer tumor cells (Supplementary Fig. S12).

To evaluate the role of macrophages in vivo, we aimed at deleting macrophages in PC-3 xenografts, treating mice with clodronate containing liposomes (Clodrolip) at 3 days interval, starting 4 days before bicalutamide–iron dextran or saline treatment (Fig. 6A). Histologic analysis of liver, spleen, and tumor masses confirmed the reduction of F4/80+ cells as well as Perls' positive F4/80+ cells upon iron injection in Clodrolip mice compared with control mice treated with empty liposomes (Fig. 6B). In iron-injected mice, the systemic depletion of iron sequestering macrophages, induced an increased tumor iron content in Clodrolip group than in control mice, suggesting that macrophages depletion diverts iron to cancer cells. In Clodrolip mice, extra iron still accumulated in the liver, but with different cellular distribution since stored in parenchymal hepatocytes rather than in macrophages (Fig. 6B and C).

Figure 6.

Macrophages dictate body iron distribution but do not limit the iron effect. RAG2−/−γc−/− mice were subcutaneously injected with PC-3 cells. When tumor masses reached 100–300 mm3, mice were treated with clodronate containing liposomes (Clodrolip) or empty liposomes at 3 days interval, starting 4 days before bicalutamide–iron dextran (Bic+FeDe) or saline treatment (day 0; 3 mice/group). A, Scheme of the experiment. B, Representative liver, spleen, and tumor sections of Bic+FeDe ± Clodrolip treated mice stained by Perls'stain or with anti-F4/80 antibody, showing brown iron positive macrophages. C, Liver, spleen, and tumor iron content. D, Tumor growth. E, Representative tumor sections stained with anti-Ki-67 antibody, showing brown positive nuclei of actively proliferating cells. Statistical differences were measured by Tukey post one-way ANOVA test in B and Tukey post two-way ANOVA test in C. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Figure 6.

Macrophages dictate body iron distribution but do not limit the iron effect. RAG2−/−γc−/− mice were subcutaneously injected with PC-3 cells. When tumor masses reached 100–300 mm3, mice were treated with clodronate containing liposomes (Clodrolip) or empty liposomes at 3 days interval, starting 4 days before bicalutamide–iron dextran (Bic+FeDe) or saline treatment (day 0; 3 mice/group). A, Scheme of the experiment. B, Representative liver, spleen, and tumor sections of Bic+FeDe ± Clodrolip treated mice stained by Perls'stain or with anti-F4/80 antibody, showing brown iron positive macrophages. C, Liver, spleen, and tumor iron content. D, Tumor growth. E, Representative tumor sections stained with anti-Ki-67 antibody, showing brown positive nuclei of actively proliferating cells. Statistical differences were measured by Tukey post one-way ANOVA test in B and Tukey post two-way ANOVA test in C. ns, nonstatistically significant. *, P < 0.05. **, P < 0.01. ***, P < 0.001.

Close modal

Treatment with Clodrolip did not affect tumor growth in saline-treated mice and did not interfere with the effect of bicalutamide–iron combination, being tumor growth similarly reduced in Clodrolip and empty liposome groups (Fig. 6D). In accordance, tumor sections of bicalutamide–iron-treated mice showed lower Ki-67–positive cells than saline-treated mice, without differences between mice treated with Clodrolip or empty liposomes (Fig. 6E).

Altogether, our results indicate that tumor macrophages have a marginal influence in controlling the expansion of subcutaneous tumor masses once established in RAG2−/−γc−/− mice.

In this article, we tested the effect of iron at high doses and the combination of iron with antiandrogen treatment in prostate cancer cells both in vitro and in murine models. Our results illustrate the proof of principle that iron toxicity interferes with prostate cancer cell growth and that the positive iron effect is enhanced by combination with an antiandrogen compound, such as bicalutamide.

Iron sensitivity parallels AR activity

Our study was performed on multiple human prostate cancer cell lines. The VCaP and LNCaP express the AR and are sensitive to androgen inhibitors. The DU-145 and PC-3 cells almost entirely lack AR and are models of severe castration resistance. Exposure to iron caused lipid damage and ferroptosis in all cell lines analyzed, revealing that VCaP and LNCaP are highly iron sensitive and that in all cells, combination of iron with bicalutamide increases oxidative stress and cell death compared with iron alone. By comparing basal iron and oxidative status of all cell lines and analyzing gene expression data available for LNCaP and PC-3 cells we concluded that high basal free iron content and oxidative status are the main determinants of the high iron sensitivity in VCaP and LNCaP cells. This finding is consistent with the high-mitochondrial metabolic activity dependent on the active androgens signaling (20, 21). On the basis of these results, we propose that active androgen metabolism parallels iron sensitivity. Accordingly, TRAMP-C2 cells, a murine model of AR-positive prostate cancer cells, showed high iron sensitivity both in vitro and in vivo experiments.

Our results suggest that tumors with active androgen signaling might be intrinsically iron sensitive. We speculate that this would also be relevant for castration-resistant tumors where AR signaling is active, such as tumors characterized by abnormal AR overexpression or constitutively active mutated AR. Importantly, our results also show that castration-resistant tumors with inactive or absent androgen signaling pathway, like those lacking AR and/or characterized by neuroendocrine differentiation might be converted into iron sensitive by combining iron with antiandrogens.

Bicalutamide enhances the oxidative circuit initiated by iron

All prostate cancer cells proliferate less when exposed to bicalutamide–iron combination than when exposed to each single treatment. This effect is particularly evident in PC-3 cells, whose proliferation is barely impaired in vitro and unaffected in vivo by iron or bicalutamide alone. Although AR signaling is nonfunctional in PC-3 cells, high-dose bicalutamide mildly reduces PC-3 viability, suggesting that the androgen inhibitor has off-target effects that might intersect oxidative iron homeostasis, indirectly increasing iron toxicity.

Bicalutamide promotes the expression of AKR1C1 and AKR1C2, members of the AKR1C superfamily of steroid hormone–transforming enzymes that under physiologic conditions act as reductases since the usual cell concentration of NAD(P)H prevents their oxidase activity (23, 29). In prostate cells, both enzymes contribute to inactivate 5α-dihydrotestosterone by catalyzing its reduction to 3α-diol, a steroid hormone with low AR affinity, or to 3β-diol, a high affinity ligand of estrogen receptor β, a recognized antiproliferative factor for prostate cancer cells (30–33). The latter effect might explain why bicalutamide, by promoting AKR1Cs upregulation, slightly impairs PC-3 cell viability, at least in vitro.

As reductases, AKR1Cs are also components of the cellular defense machinery against oxidative insults (17–19), and may be induced by ROS (34). Our data revealed that their increase promoted by oxidative stress is not a primary response, as it does not occur in iron-treated PC-3 cells, irrespective of lipid damage induced by iron. We speculated that AKR1Cs, if not regulated by the antioxidant response, might act as oxidases in bicalutamide–iron treated cells, causing supplementary oxidative stress and cell death over levels induced by iron alone. A similar event has been previously described in lung cancer cells, during the metabolic activation of polycyclic aromatic hydrocarbons (27). Our results obtained by AKR1C2 genetic overexpression are in line with this possibility. In our model, iron is the prooxidant that trigger the consumption of GSH and antioxidant capacity to buffer toxicity. In this condition, it is plausible that excessive AKR1Cs levels induced by the antiandrogen might overcome the low affinity for NADP+, inducing a prevalent oxidases activity that contributes to generate rather than prevent oxidative damage.

The role of macrophages

Macrophages are main players of systemic iron homeostasis. They safely store iron and release it accordingly to body iron needs. As expected in our experimental settings, extra iron accumulated in spleen and liver macrophages, but also in those associated with tumor masses. Macrophage tumors might influence the iron effect in opposite directions. On one side, iron sequestration by macrophages might potentially limit iron uptake of cancer cells, buffering iron toxicity; on the opposite side, it might contribute to negatively affect cancer cells proliferation by influencing macrophages polarization toward the proinflammatory phenotype (14, 28) that has been shown to reduce tumor size in cancer models (14–16). We confirmed that iron loading in vitro promotes Tnf expression in macrophages, and represses markers associated with antiinflammatory protumoral phenotype, even in presence of cytokines potentially released by prostate cancer tumor cells.

However, our in vivo results point to a marginal influence of iron-loaded macrophages in controlling tumor expansion in our experimental models. This was clear in mice injected with TRAMP-C2 cells where different iron doses provided similar reduction of tumor growth, irrespective of the great difference in the number of Perls' positive macrophages generated in tumor masses. Also the results of our experiments of macrophage depletion by Clodrolip showed that the contribution of macrophages on tumor growth reduction is limited, if not negligible. It seems that also their ability to sequester iron does not impact on the effect of iron on cancer cells. Indeed, iron content in tumors, following macrophage depletion, does not translate into a further reduction of growth, suggesting that the amount of iron uptaken by cancer cells in control mice was sufficient to induce the oxidative circuit promoted by iron.

Our experimental mouse models proved ideal to dissect the direct effect of iron loading on cancer cells. It is possible that the iron capacity to address tumor-associated macrophage polarization may be more relevant in models characterized by slow tumor progression, where microenvironment exerts long-term effects on primary tumor development and metastases. Moreover, we speculate that iron-directed macrophages polarization toward proinflammatory phenotype might be of clinical value for patients because the density of antiinflammatory macrophages has been associated with primary prostate cancer development and metastatic dissemination (35, 36). In addition, widely used antiandrogens, are known to promote antiinflammatory macrophages infiltration in metastatic lesions (37, 38).

In conclusion, our results reveal that iron-mediated toxicity significantly inhibits prostate cancer cell growth and pave the way to pilot clinical studies aimed at confirming the effect of iron supplementation at high dose in patients with prostate cancer. This strategy is promising also because the iron-dextran dosages we used in mice in this and previous works did not cause liver or kidney toxicity (13). Moreover, once converted to humans (Conversion of Animal Doses to Human Equivalent Doses, www.fda.gov/media/72309/download), iron-dextran doses approximate the equivalent human highest iron dose infused in single injection (20 mg/kg) to treat both absolute and functional iron deficiency anemia of patients with cancer (39–41). More importantly, we achieved a comparable effect by using the human dose (20 mg/kg) of ferric carboxymaltose, a compound widely used in clinics. This suggests the feasibility of the approach in selected patients with prostate cancer. Overall, our data suggest that iron addition to therapeutic strategies could be valuable in different stages of prostate cancer, both androgen sensitive and insensitive, to maximize the efficacy and possibly overcome resistance of available therapies.

M. Bellone reports personal fees from Roche (honoraria for lectures) outside the submitted work. C. Camaschella reports personal fees from Vifor Pharma during the conduct of the study and personal fees from Novartis, Celgene outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

J. Bordini: Conceptualization, resources, data curation, formal analysis, validation, investigation, visualization, methodology, writing-review and editing. F. Morisi: Data curation, investigation, visualization, methodology. A.R. Elia: Data curation, validation, investigation, visualization, methodology. P. Santambrogio: Resources, data curation, investigation, visualization, methodology. A. Pagani: Data curation, investigation, visualization, methodology. V. Cucchiara: Data curation, investigation, visualization, methodology. P. Ghia: Resources, data curation, formal analysis, validation, visualization, writing-review and editing. M. Bellone: Resources, data curation, formal analysis, funding acquisition, validation, visualization, writing-review and editing. A. Briganti: Resources, data curation, formal analysis, validation, visualization, writing-review and editing. C. Camaschella: Resources, data curation, formal analysis, validation, visualization, writing-review and editing. A. Campanella: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft.

This work was partially supported by the Italian Association for Cancer Research (AIRC) Program Grant in Molecular Oncology-Extension no. 9965 (to A. Campanella) and AIRC Investigator Grant IG-16807 (to M. Bellone). We thank Francesco Montorsi and Giovanni Tonon for precious advices. We are indebted to Sonia Levi for the kind supply of rH02 and LF03 mAbs.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Torre
LA
,
Bray
F
,
Siegel
RL
,
Ferlay
J
,
Lortet-Tieulent
J
,
Jemal
A
. 
Global cancer statistics, 2012
.
CA Cancer J Clin
2015
;
65
:
87
108
.
2.
Boorjian
SA
,
Thompson
RH
,
Tollefson
MK
,
Rangel
LJ
,
Bergstralh
EJ
,
Blute
ML
, et al
Long-term risk of clinical progression after biochemical recurrence following radical prostatectomy: the impact of time from surgery to recurrence
.
Eur Urol
2011
;
59
:
893
9
.
3.
Cornford
P
,
Bellmunt
J
,
Bolla
M
,
Briers
E
,
De Santis
M
,
Gross
T
, et al
EAU-ESTRO-SIOG guidelines on prostate cancer. Part II: treatment of relapsing, metastatic, and castration-resistant prostate cancer
.
Eur Urol
2017
;
71
:
630
42
.
4.
Tesfay
L
,
Clausen
KA
,
Kim
JW
,
Hegde
P
,
Wang
X
,
Miller
LD
, et al
Hepcidin regulation in prostate and its disruption in prostate cancer
.
Cancer Res
2015
;
75
:
2254
63
.
5.
Miller
LD
,
Coffman
LG
,
Chou
JW
,
Black
MA
,
Bergh
J
,
D'Agostino
R
, et al
An iron regulatory gene signature predicts outcome in breast cancer
.
Cancer Res
2011
;
71
:
6728
37
.
6.
Gu
Z
,
Wang
H
,
Xia
J
,
Yang
Y
,
Jin
Z
,
Xu
H
, et al
Decreased ferroportin promotes myeloma cell growth and osteoclast differentiation
.
Cancer Res
2015
;
75
:
2211
21
.
7.
Hentze
MW
,
Muckenthaler
MU
,
Galy
B
,
Camaschella
C
. 
Two to tango: regulation of mammalian iron metabolism
.
Cell
2010
;
142
:
24
38
.
8.
Dixon
SJ
,
Lemberg
KM
,
Lamprecht
MR
,
Skouta
R
,
Zaitsev
EM
,
Gleason
CE
, et al
Ferroptosis: an iron-dependent form of nonapoptotic cell death
.
Cell
2012
;
149
:
1060
72
.
9.
Stockwell
BR
,
Friedmann Angeli
JP
,
Bayir
H
,
Bush
AI
,
Conrad
M
,
Dixon
SJ
, et al
Ferroptosis: a regulated cell death nexus linking metabolism, redox biology, and disease
.
Cell
2017
;
171
:
273
85
.
10.
Bordini
J
,
Bertilaccio
MTS
,
Ponzoni
M
,
Fermo
I
,
Chesi
M
,
Bergsagel
PL
, et al
Erythroblast apoptosis and microenvironmental iron restriction trigger anemia in the VK*MYC model of multiple myeloma
.
Haematologica
2015
;
100
:
534
841
.
11.
Campanella
A
,
Santambrogio
P
,
Fontana
F
,
Frenquelli
M
,
Cenci
S
,
Marcatti
M
, et al
Iron increases the susceptibility of multiple myeloma cells to bortezomib
.
Haematologica
2013
;
98
:
971
9
.
12.
Bordini
J
,
Galvan
S
,
Ponzoni
M
,
Bertilaccio
MTS
,
Chesi
M
,
Bergsagel
PL
, et al
Induction of iron excess restricts malignant plasma cells expansion and potentiates bortezomib effect in models of multiple myeloma
.
Leukemia
2017
;
31
:
967
70
.
13.
Bordini
J
,
Morisi
F
,
Cerruti
F
,
Cascio
P
,
Camaschella
C
,
Ghia
P
, et al
Iron causes lipid oxidation and inhibits proteasome function in multiple myeloma cells: a proof of concept for novel combination therapies
.
Cancers
2020
;
12
:
970
.
14.
Zanganeh
S
,
Hutter
G
,
Spitler
R
,
Lenkov
O
,
Mahmoudi
M
,
Shaw
A
, et al
Iron oxide nanoparticles inhibit tumour growth by inducing pro-inflammatory macrophage polarization in tumour tissues
.
Nat Nanotechnol
2016
;
11
:
986
94
.
15.
Costa da Silva
M
,
Breckwoldt
MO
,
Vinchi
F
,
Correia
MP
,
Stojanovic
A
,
Thielmann
CM
, et al
Iron induces anti-tumor activity in tumor-associated macrophages
.
Front Immunol
2017
;
8
:
1479
.
16.
Trujillo-Alonso
V
,
Pratt
EC
,
Zong
H
,
Lara-Martinez
A
,
Kaittanis
C
,
Rabie
MO
, et al
FDA-approved ferumoxytol displays anti-leukaemia efficacy against cells with low ferroportin levels
.
Nat Nanotechnol
2019
;
14
:
616
22
.
17.
Rausa
M
,
Pagani
A
,
Nai
A
,
Campanella
A
,
Gilberti
ME
,
Apostoli
P
, et al
Bmp6 expression in murine liver non parenchymal cells: a mechanism to control their high iron exporter activity and protect hepatocytes from iron overload?
PLoS One
2015
;
10
:
e0122696
.
18.
Cozzi
A
,
Levi
S
,
Bazzigaluppi
E
,
Ruggeri
G
,
Arosio
P
. 
Development of an immunoassay for all human isoferritins, and its application to serum ferritin evaluation
.
Clin Chim Acta
1989
;
184
:
197
206
.
19.
Foster
BA
,
Gingrich
JR
,
Kwon
ED
,
Madias
C
,
Greenberg
NM
. 
Characterization of prostatic epithelial cell lines derived from transgenic adenocarcinoma of the mouse prostate (TRAMP) model
.
Cancer Res
1997
;
57
:
3325
30
.
20.
Pinthus
JH
,
Bryskin
I
,
Trachtenberg
J
,
Luz
J-P
,
Singh
G
,
Fridman
E
, et al
Androgen induces adaptation to oxidative stress in prostate cancer: implications for treatment with radiation therapy
.
Neoplasia
2007
;
9
:
68
80
.
21.
Higgins
LH
,
Withers
HG
,
Garbens
A
,
Love
HD
,
Magnoni
L
,
Hayward
SW
, et al
Hypoxia and the metabolic phenotype of prostate cancer cells
.
Biochim Biophys Acta
2009
;
1787
:
1433
43
.
22.
Lim
PJ
,
Duarte
TL
,
Arezes
J
,
Garcia-Santos
D
,
Hamdi
A
,
Pasricha
S-R
, et al
Nrf2 controls iron homoeostasis in haemochromatosis and thalassaemia via Bmp6 and hepcidin
.
Nat Metab
2019
;
1
:
519
31
.
23.
Rižner
TL
,
Penning
TM
. 
Role of aldo–keto reductase family 1 (AKR1) enzymes in human steroid metabolism
.
Steroids
2014
;
79
:
49
63
.
24.
Burczynski
ME
,
Sridhar
GR
,
Palackal
NT
,
Penning
TM
. 
The reactive oxygen species- and michael acceptor-inducible human aldo-keto reductase AKR1C1 reduces the α,β-unsaturated aldehyde 4-hydroxy-2-nonenal to 1,4-dihydroxy-2-nonene
.
J Biol Chem
2001
;
276
:
2890
7
.
25.
MacLeod
AK
,
McMahon
M
,
Plummer
SM
,
Higgins
LG
,
Penning
TM
,
Igarashi
K
, et al
Characterization of the cancer chemopreventive NRF2-dependent gene battery in human keratinocytes: demonstration that the KEAP1–NRF2 pathway, and not the BACH1–NRF2 pathway, controls cytoprotection against electrophiles as well as redox-cycling compounds
.
Carcinogenesis
2009
;
30
:
1571
80
.
26.
Dixon
SJ
,
Patel
D
,
Welsch
M
,
Skouta
R
,
Lee
E
,
Hayano
M
, et al
Pharmacological inhibition of cystine-glutamate exchange induces endoplasmic reticulum stress and ferroptosis
.
Elife
2014
;
3
:
e02523
.
27.
Penning
TM
. 
Human aldo-keto reductases and the metabolic activation of polycyclic aromatic hydrocarbons
.
Chem Res Toxicol
2014
;
27
:
1901
17
.
28.
Handa
P
,
Thomas
S
,
Morgan‐Stevenson
V
,
Maliken
BD
,
Gochanour
E
,
Boukhar
S
, et al
Iron alters macrophage polarization status and leads to steatohepatitis and fibrogenesis
.
J Leukoc Biol
2019
;
105
:
1015
26
.
29.
Rižner
TL
,
Lin
HK
,
Peehl
DM
,
Steckelbroeck
S
,
Bauman
DR
,
Penning
TM
. 
Human type 3 3α-hydroxysteroid dehydrogenase (aldo-keto reductase 1C2) and androgen metabolism in prostate cells
.
Endocrinology
2003
;
144
:
2922
32
.
30.
Krege
JH
,
Hodgin
JB
,
Couse
JF
,
Enmark
E
,
Warner
M
,
Mahler
JF
, et al
Generation and reproductive phenotypes of mice lacking estrogen receptor beta
.
Proc Natl Acad Sci U S A
1998
;
95
:
15677
82
.
31.
Kim
IY
,
Kim
B-C
,
Seong
DH
,
Lee
DK
,
Seo
J
,
Hong
YJ
, et al
Raloxifene, a mixed estrogen agonist/antagonist, induces apoptosis in androgen-independent human prostate cancer cell lines
.
Cancer Res
2002
;
62
:
5365
9
.
32.
Yun
H
,
Xie
J
,
Olumi
AF
,
Ghosh
R
,
Kumar
AP
. 
Activation of AKR1C1/ERβ induces apoptosis by downregulation of c-FLIP in prostate cancer cells: a prospective therapeutic opportunity
.
Oncotarget
2015
;
6
:
11600
13
.
33.
Zhou
C
,
Yu
C
,
Guo
L
,
Wang
X
,
Li
H
,
Cao
Q
, et al
In vivo study of the effects of ER β on apoptosis and proliferation of hormone-independent prostate cancer cell lines PC-3M
.
Biomed Res Int
2018
;
2018
:
1439712
.
34.
Burczynski
ME
,
Lin
HK
,
Penning
TM
. 
Isoform-specific induction of a human aldo-keto reductase by polycyclic aromatic hydrocarbons (PAHs), electrophiles, and oxidative stress: implications for the alternative pathway of PAH activation catalyzed by human dihydrodiol dehydrogenase
.
Cancer Res
1999
;
59
:
607
14
.
35.
Pollard
JW
. 
Tumour-educated macrophages promote tumour progression and metastasis
.
Nat Rev Cancer
2004
;
4
:
71
8
.
36.
Lanciotti
M
,
Masieri
L
,
Raspollini
MR
,
Minervini
A
,
Mari
A
,
Comito
G
, et al
The role of M1 and M2 macrophages in prostate cancer in relation to extracapsular tumor extension and biochemical recurrence after radical prostatectomy
.
Biomed Res Int
2014
;
2014
:
486798
.
37.
Lin
T-H
,
Izumi
K
,
Lee
SO
,
Lin
W
,
Yeh
S
,
Chang
C
. 
Anti-androgen receptor ASC-J9 versus anti-androgens MDV3100 (Enzalutamide) or Casodex (Bicalutamide) leads to opposite effects on prostate cancer metastasis via differential modulation of macrophage infiltration and STAT3-CCL2 signaling
.
Cell Death Dis
2013
;
4
:
e764
.
38.
Lo
CH
,
Lynch
CC
. 
Multifaceted roles for macrophages in prostate cancer skeletal metastasis
.
Front Endocrinol
2018
;
9
:
247
.
39.
Steinmetz
T
,
Tschechne
B
,
Harlin
O
,
Klement
B
,
Franzem
M
,
Wamhoff
J
, et al
Clinical experience with ferric carboxymaltose in the treatment of cancer- and chemotherapy-associated anaemia
.
Ann Oncol
2013
;
24
:
475
82
.
40.
Hedenus
M
,
Karlsson
T
,
Ludwig
H
,
Rzychon
B
,
Felder
M
,
Roubert
B
, et al
Intravenous iron alone resolves anemia in patients with functional iron deficiency and lymphoid malignancies undergoing chemotherapy
.
Med Oncol
2014
;
31
:
302
.
41.
Toledano
A
,
Luporsi
E
,
Morere
JF
,
Scotté
F
,
Laribi
K
,
Barrière
J
, et al
Clinical use of ferric carboxymaltose in patients with solid tumours or haematological malignancies in France
.
Support Care Cancer
2016
;
24
:
67
75
.