Iron chelation and 2‐oxoglutarate‐dependent dioxygenase inhibition suppress mantle cell lymphoma's cyclin D1

Abstract The patients with mantle cell lymphoma (MCL) have translocation t(11;14) associated with cyclin D1 overexpression. We observed that iron (an essential cofactor of dioxygenases including prolyl hydroxylases [PHDs]) depletion by deferoxamine blocked MCL cells’ proliferation, increased expression of DNA damage marker γH2AX, induced cell cycle arrest and decreased cyclin D1 level. Treatment of MCL cell lines with dimethyloxalylglycine, which blocks dioxygenases involving PHDs by competing with their substrate 2‐oxoglutarate, leads to their decreased proliferation and the decrease of cyclin D1 level. We then postulated that loss of EGLN2/PHD1 in MCL cells may lead to down‐regulation of cyclin D1 by blocking the degradation of FOXO3A, a cyclin D1 suppressor. However, the CRISPR/Cas9‐based loss‐of‐function of EGLN2/PHD1 did not affect cyclin D1 expression and the loss of FOXO3A did not restore cyclin D1 levels after iron chelation. These data suggest that expression of cyclin D1 in MCL is not controlled by ENGL2/PHD1‐FOXO3A pathway and that chelation‐ and 2‐oxoglutarate competition‐mediated down‐regulation of cyclin D1 in MCL cells is driven by yet unknown mechanism involving iron‐ and 2‐oxoglutarate‐dependent dioxygenases other than PHD1. These data support further exploration of the use of iron chelation and 2‐oxoglutarate‐dependent dioxygenase inhibitors as a novel therapy of MCL.

especially ibrutinib has impressive responses, but with almost uniform development of resistance. [1][2][3] Developing precise therapeutic strategy that will prevent the relapses and allow long-term remission without excess toxicities is still a serious challenge. We observed that MCL-derived cell lines have decreased survival and proliferation compared to non-MCL lymphoma cell lines when grown at iron-deprived conditions, 4 and Vazana-Barad et al 5 reported that MCL patients may benefit from iron-chelating agents.
Iron is essential for cell proliferation. In tumour cells, the iron metabolic alterations (an elevated entry, a decrease in its elimination and a disruption of its storage) help to facilitate the accelerating cell division and supply iron for increased DNA synthesis but, on the other hand, also sensitizes cancer cells to iron depletion. 6,7 Although iron chelators were developed for treatment of iron overload diseases, they are also potent DNA synthesis inhibitors in vitro, as they inhibit ribonucleotide reductase (RR), 8,9 an iron-dependent enzyme, essential for reduction of ribonucleotides to deoxynucleotides (dNTPs). [9][10][11] Iron chelators therefore represent promising adjuvant elements in the treatment of cancers especially in subgroup of patients dealing with resistance to the established therapy.
It has been shown that in MCL cell lines, iron chelator deferasirox down-regulates cyclin D1 which in turn leads to inhibition of Rb phosphorylation and increase of the E2F/Rb complex levels ultimately leading to G1/S arrest. 5 The mechanism by which iron may affect cyclin D1 in cancer cells has been suggested by Nurtjahja-Tjendraputra et al 12 in a study that examined the ability of iron chelators to inhibit cell proliferation and induce apoptosis. It was postulated that iron chelation caused proteasomal degradation of cyclin D1. The degradation of cyclin D1 was ubiquitin independent in iron-deplete conditions, while ubiquitination of cyclin D1 degradation takes place in iron-replete cells.
However, Zhang et al 13 showed that the mammalian cyclin D1dependent proliferation is regulated by prolyl hydroxylase 1 (PHD1, encoded by the EGLN2 gene) in a hypoxia-inducible factor independent manner by the transcriptional mechanism rather than via the proteasomal pathway. Cyclin D1 is not a direct substrate for PHD1. It was suggested that forkhead box O3A (FOXO3A) transcription factor is the link between the regulation of cyclin D1 and prolyl hydroxylase PHD1. 14 PHD1 can hydroxylate FOXO3A on two specific prolyl residues thereby blocking its interaction with the USP9x deubiquitinase and promoting its proteasomal degradation. Loss of EGLN2/ PHD1 leads to accumulation of FOXO3A, which, in turn, suppresses cyclin D1 expression.
Prolyl hydroxylases (PHDs) belong to the iron and 2-oxoglutarate (2-OG)-dependent dioxygenase family, and as principal negative regulators of HIFs, they contribute to oxygen sensing. There are three paralogues of the EGLN gene family (EGLN1/PHD2, EGLN2/ PHD1 and EGLN3/PHD3) encoding PHDs. The PHD1 is found to be exclusively present in the nucleus, and PHD2 is mainly located in the cytoplasm while PHD3 protein is homogenously distributed throughout the cytoplasm and nucleus. 15,16 While all members of the PHD protein family contribute to regulation of cellular O 2 sensing, only EGLN2/PHD1 and EGLN3/PHD3 were demonstrated to have HIF-independent functions, such as in DNA damage control 17,18 and NF-κB activity regulation. 19,20 The connection between prolyl hydroxylases and cell cycle regulation was first described in drosophila; their PHD homologue Hif-1 prolyl hydroxylase (Hph) was shown to be a regulator of cellular growth and a key mediator for the drosophila cyclin-dependent protein kinase complex cyclin D/cyclin-dependent kinase 4. 21 The mouse PHD1 homologue Falkor was identified as a DNA damage-related growth regulator in mouse embryonic fibroblasts. 17 It was shown that Falkor can also inhibit HIF-2 and a combined knockout of EGLN2 and EGLN3 leads to polycythemia/ erythrocytosis as HIF-2 is the principal regulator of erythropoietin gene. 22,23 In human breast cancer cells, EGLN2 mRNA was shown to accumulate in cells stimulated with oestrogen and participate in oestrogen-independent cancer cells' growth and their resistance to hormone therapy. 24 In the present study, we confirmed the effect of cellular iron depletion on MCL cell lines 5,12 and observed increased sensitivity to chelation treatment of MCL cell lines in comparison with the non-MCL cell lines without constitutively active cyclin D1. As the molecular mechanism inducing cyclin D1 degradation after iron chelation is not known, we postulated that it could be linked to PHD1-FOXO3A pathway. To unravel the role of prolyl hydroxylases in cyclin D1 regulation in MCL, we generated MCL cell lines harbouring the EGLN2 or FOXO3A loss-of-function (LOF) genes. In addition, MCL cells were treated with 2-OG analogue, dimethyloxalylglycine (DMOG), a competitive inhibitor of prolyl hydroxylase domain-containing proteins.
Several PHD inhibitors have been recently generated by Pharma industry, and they are already used in clinical trials of anaemia 25-28 ; further, the inhibitors of PHDs that target HIF-2α are already used in the clinical trials of HIF-dependent cancers. 29,30 These inhibitors have different selectivity against 2-OG-dependent oxygenases, 31,32 but in addition to 2-OG oxygenase inhibitory potency can exhibit also iron-chelating ability. 31 We propose that either chelating agents or broad spectrum 2-OG-dependent oxygenase inhibitors (rather than specific PHD inhibitors) can be expeditiously applied as a new avenue for MCL-targeted therapy. (1% O 2 , 5% CO 2 , 94% N 2 ), which was placed in the standard tissue culture incubator at 37°C. Cultures and assays used for analyses of mouse embryonic stem cells (mESCs) are described in Appendix S1.

| Proliferation assay
Cell number and viability were determined using CellometerAutoT4 (Nexcelom Bio-science) based on the trypan blue exclusion method or by CellTitre-Blue reagent (Promega) and Perkin-Elmer Envision analyzer.

| RNA isolation and quantitative RT-PCR
RNA was isolated using TRI reagent (Sigma-Aldrich), and 500 ng of DNA-free RNA was reverse-transcribed using the First Strand cDNA Transcriptor Synthesis Kit (Roche) or 1000 ng of DNA-free RNA was reverse-transcribed using the RevertAid Reverse Transcriptase (ThermoFisher Scientific) according to the manufacturer's manual.
Gene expression experiments were performed on LightCycler 480 system (Roche) with following TaqMan probes: Hs00765553_m1

| The effect of cellular iron depletion on human mantle cell and non-mantle cell lymphoma cell lines
We extended our previous experiments 4 Figure 1D). In addition, the overexpression of cyclin D1 made MCL cell lines more susceptible to treatment with DFO (percentage of G1 cells of MCL cell lines was significantly higher than in non-MCL cell lines, Figure 1D). Despite the report 12

| Regulation of cyclin D1 in MCL cell lines is not controlled by EGLN2/PHD1 and its hydroxylation target FOXO3A
It has been previously reported that an inability of PHD1 to hydroxylate FOXO3A promotes its accumulation in cells, which in turn suppresses cyclin D1 expression by a yet unknown mechanism. 13 In order to decipher whether iron chelation down-regulates cyclin D1 by inhibiting PHD1 function and thus prevents FOXO3A proteasomal degradation, we created EGLN2 and FOXO3A CRISPR/Cas9 based LOF MCL Mino cell lines (Figure 2A,B). The loss of PHD1 did not lead to the down-regulation of cyclin D1 expression in MCL cell line (Figure 2A, upper panels). In order to validate our CRISPR/ Cas9 system, we created EGLN2 LOF in HEK293 cells, and as ex- Up-regulation of FOXO3A expression was also detected in other MCL cell lines after DFO treatment ( Figure 2C).

| Down-regulation of EGLN2 and accumulation of FOXO3A mRNA after DFO treatment in MCL cell lines is caused by induced hypoxia
As DFO is a known hypoxia-mimetic agent, we asked whether the It has been previously shown that EGLN2 promoter contains binding sites for aryl hydrocarbon nuclear translocator (ARNT/HIF-1β) 34 which mediates its down-regulation under hypoxic conditions and that FOXO3A transcript level, in response to hypoxia, accumulates in HIF1-dependent manner, resulting in enhanced FOXO3A activity. 35 Here, our data demonstrate that down-regulation of EGLN2 and accumulation of FOXO3A mRNA after DFO treatment is rather a consequence of induced hypoxia created by iron depletion, but neither hypoxia nor ENGL2/PHD1-FOXO3A pathway alone regulate cyclin D1 expression in MCL cells.

| Treatment with prolyl hydroxylase inhibitor DMOG decreases MCL cells' viability
Despite the fact that direct PHD1 hydroxylase inactivation does not seem to influence regulation of cyclin D1 in MCL, we asked whether inhibition of 2-OG-dependent hydroxylases could impact the MCL cells.
We treated MCL cell lines with prolyl hydroxylase inhibitor DMOG, a synthetic analogue of 2-OG, which catalytically inhibits hydroxylation reaction. We found decreased proliferation rate of MCL cells (Figure 4 (Figure 4, lower panels). As DMOG is predicted to inhibit a broad spectrum of dioxygenases, including hydroxylases, it is possible that these enzymes have additional substrates with the ability to regulate aberrantly expressed cyclin D1 in MCL cells.

| DNA damage is induced in MCL cells treated with DFO but not with DMOG
Iron chelation inhibits multiple enzymes functioning in DNA replication, DNA repair and cell cycle progression. 8,9 One of these enzymes is RR, inhibition of which leads to dNTP deficiency. 10 Decreased dNTP pools are known to induce DNA damage and replication stress in oncogene expressing proliferating cells. 11,[36][37][38] We first tested whether dNTP deficiency resulting from DFO-mediated RR inhibition causes DNA damage and apoptosis in fast proliferating cells. For this purpose, we used mouse embryonic stem cells (mESCs), suffering from intrinsic deficiency of dNTP pools, exhibiting intrinsically high phosphorylation of histone H2AX (γH2AX), a DNA damage response (DDR) marker, 39 and displaying high sensitivity to further dNTP depletion 40 (see Appendix S1). These experiments revealed that DFO-mediated inhibition of RR activity causes DNA damage, DDR and apoptosis through depletion of dNTP pools, as it can be rescued by addition of deoxynucleosides to the media (Figures S1 and   S2). As these processes were shown to be p53-activation/caspase Expression levels of EGLNs, selected HIF target genes (VEGF and SLC2A) and FOXO3A were determined by quantitative PCR after incubation with DMOG for 24 h 3 cleavage dependent, 41 we did not expect them to be induced by DMOG treatment, which is known to inhibit PHD3-mediated hydroxylation of p53, preventing its accumulation and apoptotic activity. 18

| D ISCUSS I ON
In our experiments, we tested DFO, a potent iron chelator that induces G1/S arrest and/or apoptosis in many somatic cell types, including cancer cell lines, 8 To unravel the specific molecular targets that trigger cell cycle arrest and apoptosis of hyperproliferative cells exposed to DFO, we used mESCs that are characterized by intrinsic deficiency of dNTP pools and high intrinsic replication stress as a model. 39 As the role of iron in regulation of cyclin D1 expression is not completely understood, we investigated the molecular mechanism underlying decreased cyclin D1 mRNA and protein levels in MCL cell lines after DFO-induced iron deficiency. PHDs are dependent on iron 45 to catalyse its hydroxylation activity; thus, iron chelation decreases their enzyme activity. Nevertheless, EGLN2/PHD1 LOF in Mino cells did not affect cyclin D1 expression and FOXO3A LOF did not restore cyclin D1 levels after chelation treatment. Therefore, the cyclin D1 in MCL cells escapes this regulation circuit and its downregulation by iron depletion is mediated by another, yet unknown mechanism(s). We hypothesize that in MCL cells production of cyclin D1, which is aberrantly localized in the proximity of a nucleolus and influenced by specific transcription enhancers (eg, nucleolin) 46 because of t (11;14) translocation, escapes the PHD1-FOXO3A regulation. It is known that iron chelators enhance HIFs-α accumulation 47 and thus induce hypoxia response. We measured expression of cyclin D1, EGLN2 and known HIF target genes VEGF and SLC2A after 24 hours in 1% O 2 hypoxia and found that cyclin D1 level was not altered, but EGLN2 expression was down-regulated, suggesting that its down-regulation after DFO treatment is caused by hypoxia.
FOXO3A is a transcription factor known to be involved in many cellular processes such as apoptosis, 48-50 autophagy, 51 oxidative stress 51 F I G U R E 5 The effect of cellular iron depletion and 2-OGdependent enzymes inhibition on DNA damage in mantle cell lymphoma (MCL) cell lines. Treatment with deferoxamine induces H2AX phosphorylation on S139, a DNA damage response marker indicative of DNA damage, which is reversed by concomitant administration of FAC. No increase in γH2AX signal is detected in cells treated with DMOG. CtBP was used as a loading control, and as a functional control, cellular cyclin D1 level in MCL cell lines is monitored (during the time course of the project, both batches of CD1 antibody (#2922S, Cell Signaling, lot:3) have started to detect unspecific band with higher molecular weight than cyclin D1, and cyclin D1 is indicated with an arrow) and DNA repair. 52 We ruled out the role of FOXO3A in cyclin D1 repression because of iron depletion, but we observed accumulation of FOXO3A after chelation treatment as a result of induced hypoxia.
In many MCL tumour tissues, FOXO3A is constitutively inactivated and it was reported that its reactivation by nuclear export inhibitors had profound impact on cell viability. 53 We can only speculate that FOXO3A induction by chelation treatment would be also beneficial for MCL therapy.
In conclusion, iron chelation and treatment with non-selective hydroxylase inhibitor DMOG, 54

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interest.

AUTH O R ' S CO NTR I B UTI O N S
O. Babosova performed the research and drafted the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon request.