Reactive oxygen species levels control NF-κB activation by low dose deferasirox in erythroid progenitors of low risk myelodysplastic syndromes

Anemia is a frequent cytopenia in myelodysplastic syndromes (MDS) and most patients require red blood cell transfusion resulting in iron overload (IO). Deferasirox (DFX) has become the standard treatment of IO in MDS and it displays positive effects on erythropoiesis. In low risk MDS samples, mechanisms improving erythropoiesis after DFX treatment remain unclear. Herein, we addressed this question by using liquid cultures with iron overload of erythroid precursors treated with low dose of DFX (3μM), which corresponds to DFX 5 mg/kg/day, an unusual dose used for iron chelation. We highlight a decreased apoptosis rate and an increased proportion of cycling cells, both leading to higher proliferation rates. The iron chelation properties of low dose DFX failed to activate the Iron Regulatory Proteins and to support iron depletion, but low dose DFX dampers intracellular reactive oxygen species. Furthermore low concentrations of DFX activate the NF-κB pathway in erythroid precursors triggering anti-apoptotic and anti-inflammatory signals. Establishing stable gene silencing of the Thioredoxin (TRX) 1 genes, a NF-κB modulator, showed that fine-tuning of reactive oxygen species (ROS) levels regulates NF-κB. These results justify a clinical trial proposing low dose DFX in MDS patients refractory to erythropoiesis stimulating agents.


SUPPLEMENTARY MATERIALS CD34 + selection from primary samples
By density gradient using Ficoll (Eurobio), the mononuclear cells were isolated from total bone marrow. HSPCs (hematopoietic stem progenitor cells) were positively selected by paramagnetic iron-dextran particles directly conjugated to anti-CD34 monoclonal antibodies (CD34 MicroBead Kit, human, Miltenyi Biotec) by using an automatized system (MACS pro, Miltenyi Biotec). The HSPCs were counted using Malassez cells (Fastread, Biosigma) and their viability was assessed by trypan blue exclusion (Sigma Aldrich). The purity of all samples was checked by flow cytometry (BD FACS Canto II, BD Biosciences) with a specific CD34 fluorochromeconjugated antibody (CD34APC, Miltenyi Biotec) and was superior to 90% for all samples used.

Functional assays
Cell proliferation was assessed by cell counting using Malassez cells. Apoptotic cells were determined by the percentage of Annexin V positive cells (FITC Annexin V, BD Biosciences) by flow cytometry, as per the manufacturer's intructions. Cell cycle was studied by DAPI staining (4',6-diamidino-2'-phylindole, dihydrochloride, 1mg/ml, Thermo SCIENTIFIC). Briefly, cells were permeabilized with commercial solution (Cytofix/cytoperm™, BD Biosciences). After washes with Permwash™ (BD Biosciences), cells were incubated with DAPI at room temperature for 1 hour. DNA content was evaluated by flow cytometry. All flow cytometry analyses were done on BD FACS Canto II.

Intracellular pathway: immunofluorescence microscopy assays and flow cytometry
For confocal microscopy, at D10, 1.10 5 cells permeabilized following the same protocol as for cell cycle. Cytospins were prepared, saturated with PBS BSA 2% during 45 min. For FOXO3a staining, incubation was done overnight at 4°C and with a dilution of 1/400 concentration for primary antibody (anti-human FOXO3A, clone 3F12, Sigma Aldrich). After a wash with PBS, the secondary antibody (Alexa-fluor 647 donkey anti-mouse, Life technologies, Eugene, USA) was incubated during one hour at room temperature at dilution of 1/200. After an ultimate wash with PBS, 5μL of a mix of Vectashield ™ (Vector laboratories) with Dapi 1/2000 (DAPI antifade ES, Cytocell) was put on cells. Preparations were read on spinning disk confocal microscope (IMIC 2.0 Till Photonics, FEI, Munich) with a camera (iXon 897, EMCCD, Andor). Then data were analyzed with Icy software (http://icy.bioimageanalysis.org). The same procedure was done for NF-κB (pNF-κB p65, clone L8F6, Cell Signaling) with a dilution of 1/400 for primary antibody and 1/200 for the secondary antibody.

Clonogenic assays
At D5, cells were harvested and were cultured in 12well plates with 500μL of MethoCultTM H4434 Classic (Stemcell Technologies) at concentration of 1x10 4 cells per well. All conditions were done in duplicate. Colony forming unit erythroid (CFU-E) and the burst forming unit erythroid (BFU-E) were counted after 7 days and 14 days of culture respectively based on morphological classical criteria. An external technician to the team read the colonies in "double blind" condition.

NF-κB pathway-specific expression array
The human NF-κB signaling targets RT 2 Profiler PCR array kit (SABiosciences) was used to assess the impact of DFX on the expression of 84 genes target of NF-κB according to the manufacturer's instructions.

Intracellular iron measurements
Three millions of K562 cells were incubated for 48h with increasing doses of DFX (from 3 to 200μM) in the same medium as described above plus 100μM of ammonium Fe 3+ citrate. Cells were lysed by heat shock in water and intracellular iron concentrations were determined by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) using a XSERIES 2 analyzer (Thermo Scientific).

IRP activity measurement
K562 cells were used to perform IRP activity measurement. Cells were cultured in the same conditions as for intracellular iron measurements with an increase dose of DFX. Total IRP RNA-binding activity was measured by electrophoretic mobility shift assays with 5 μg of total protein extracts. The minimal sequence of human ferritin H-chain Iron Responsive Element (IRE) was biotin-labeled with biotinylated cytidine (bis) phosphate using T4 RNA ligase (Thermo Scientific). The IRE-IRP reaction was carried out as previously described and the complexes were separated on non-denaturing 4% PAGE in 0.5X TBE, transferred onto Hybond™N + membrane (GE Healthcare) and the biotinylated bands were detected after interaction with the streptavidinhorseradish peroxidase conjugate by the peroxidase substrate for enhanced chemiluminescence (ECL)(GE Life Sciences). Quantitation of the signals was done with the Image J software (v1.47, Wayne Rasband, Research Services Branch, National Institute of Mental Health, Bethesda, Maryland, USA).

Oxidative stress metabolite measurements
Thiobarbituric acid reactive species include products of the oxidative degradation of polyunsaturated fatty acids, in particular MDA. We used the modified method of Ohkawa et al (Ohkawa, Anal Biochem, 1979), based on the reaction of aldehyde functions of MDA released by acid hydrolysis at 95°C with thiobarbituric acid forming a pink-colored complex quantified by fluorimetry. Carbonyl assay is based on the reaction of carbonyl groups in protein with 2,4-dinitrophenylhydrazine to form 2,4-dinitrophenylhydrazone, which was estimated spectrophotometrically at 380 nm after trichloroacetic acid precipitation of proteins. Glutathione peroxidase activity was determined by the modified method of Gunzler using tertbutyl hydroperoxide as substrate (Gunzler, Chem Klin Biochem, 1974).
Cu-Zn superoxide dismutase activity (Cu-Zn SOD) was determined by monitoring the auto-oxidation of pyrogallol using the Marklund method (Marklund, Eur J Biochem, 1974).
Total SOD activity Corresponding to the activity of SOD1 (Cu-Zn SOD) and SOD2 (Mn SOD) was measured using an assay based on the competition between pyrogallol oxidation by ·O2 and superoxide dismutation by SOD. SOD2 activity was determined by assaying for SOD activity in the presence of sodium cyanide, which selectively inhibits SOD1 but not SOD2. SOD1 activity was calculated by subtracting SOD2 activity from total SOD activity. The rate of auto-oxidation is taken from the increase in the absorbance at 420 nm.

Ferroptosis investigation
RT-qPCR: RNA was extracted using the mir Vana™ miRNA Isolation Kit (Thermo Fisher Scientific) according to the manufacturer's protocol. 500 ng total RNA for each sample was used as input for each reverse transcription reaction, performed using the TaqMan RT kit (Applied Biosystems). Primer for IREB2, CS, ATP5G3, EMC2, ASCF2, RPL8 and GAPDH come from commercial solutions (Thermo Fisher Scientific). Quantitative PCR reactions were performed using TaqMan ® Universal Master Mix II, no UNG (life technologies). Triplicate samples per condition were analyzed on a Stratagene Mx3000p qPCR instrument (Agilent Technologies). Differences in mRNA levels compared to GAPDH internal reference control were computed between control and experimental conditions using the ΔΔCt method.

Thioredoxin 1 and 2 knock down cell lines production
K562 cells were transfected by electroporation (Amaxa ® Cell Line Nucleofector ® Kit V K562, Lonza, Cologne) with Nucleofector II (Amaxa Biosystem, Lonza) with three different pEBV-based plasmids expressing shRNA sequences: one with an inefficient shRNA sequence widely documented in the literature and used as a control (CTL), one targeting the TRX1 gene (NM_003329; nucleotides 208 to 226) and one against TRX2 (NM_012473, nucleotides 195 to 213). These replicative pEBV-based plasmids imposed a stable and efficient knock down, as reported in the literature for numerous genes (Biard, Nucleic Acids Res, 2007). konck down cells were called TRX1 KD and TRX2 KD cells. All plasmids contain a resistance gene against hygromycin. Knock down phenotype was checked by Western blotting analyses. Twenty to forty μg of total proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 8% gels, and proteins were transferred to polyvinylidene difluoride membranes. The blots were saturated with 2% bovine serum albumin in TBS-Tween 0.2% and probed overnight at 4°C with antibodies against Trx1 (1:200, Santa Cruz), Trx2 (1:1000, Cell Signaling) or actin (1:250, Sigma Aldrich). Following three washes with TBS-Tween 0.2%, the blots were incubated with peroxidase-coupled goat anti-rabbit IgG (Bethyl) at a dilution of 1:5000 for 1 h at room temperature, followed by detection with the ECL reagent. Both plasmids contain a different luciferase activity. The first plasmid displays the firefly luciferase activity under the dependence of a NF-κB responsive element, whereas the second plasmid constitutively produces the renilla luciferase activity and was used as control of transfection. After nuclear electroporation, the cells were cultured in the same iron overloaded medium as described before with DFX 3μM or N-acetyl cysteine 1mM for 48 hours. Then the cells were harvested, washed and the number of cells and their viability was determined with a Luna™ automated cells counter (Logos Biosystems, Korea). Then luciferase activity was measured using the Dualluciferase ® reporter assay system (Promega, Madison) on a CLARIOstar plate reader (BMG Labtech, Ortenberg, Germany) following the manufacturer's protocol. Each condition was read in duplicate. For each condition, the firefly luciferase activity was reported to the Renilla luciferase activity and the data were normalized with the number of cells.

NFKB reporter assays
Life Technologies ® ). A high depth of coverage (>2000X) was obtained for all genes. Data were processed by Torrent Browser (Life Technologies ® ) and SeqNext (JSI Medical System ® ). Frameshift and nonsense variants were always considered as relevant mutations. Single nucleotide variants were retained in the absence of description into public databases of human polymorphisms and effects on protein function were predicted with established prediction tools (SIFT, PolyPhen-1, PolyPhen-2, MAPP, PhD-SNP and SNAP). Notably, because of technical limitations, the mutation c.1934dupG in ASXL1 cannot be detected with PGM sequencing justifying its systematic validation by