Hydroxyurea inhibits proliferation and stimulates apoptosis through inducible nitric oxide synthase in erythroid cells

Abstract

Hydroxyurea (hydroxycarbamide, HU) arrests cells in the S-phase by inhibiting ribonucleotide reductase and DNA synthesis, significantly contributing to the release of nitric oxide (NO). We investigated the involvement of inducible NO synthase (NOS2) in the cytostatic effect of HU using in vitro shRNA-induced knockdown of the NOS2 transcript $\left(\text{NOS}{2}_{\text{kd}}\right)$ or a specific NOS2 inhibitor (1400W) in human erythroleukemic HEL92.1.7 cells, as well as murine erythroid progenitors (mERPs) from HU-treated wild-type (WT) and Nos2 knockout (Nos2^−/−) mice. Over the long-term, HU increased NOS2 expression in HEL92.1.7 cells (via nuclear factor kappa B [NFκB] signaling) and in mERP. In the short-term, HU increased the activity of human recombinant and erythroleukemic cell-derived NOS2, as confirmed by NO metabolite nitrite/citrulline production. In silico molecular docking predicted that HU binds to the NOS2 active site and substrate L-arginine via hydrogen bonds. Molecular dynamics simulations showed reduced rigidity of the NOS2 active site upon interaction with HU, indicating stabilization of the enzyme-substrate complex. Both 1400W and $\text{NOS}{2}_{\text{kd}}$ prevented the in vitro reduction in proliferation and induction of apoptosis in HEL92.1.7 cells by HU. $\text{NOS}{2}_{\text{kd}}$ preferentially blocked early apoptosis and HU-induced S-phase arrest in HEL92.1.7 cells. The HU-induced decrease in proliferation and stimulation of early apoptosis in mERP were prevented in Nos2^−/− mice and by 1400W in WT mice. This study demonstrated that HU induces NOS2 activity through direct interaction and increased protein expression via NFκB signaling. Moreover, NOS2 mediates the HU-induced inhibition of proliferation and stimulation of apoptosis in erythroid cells.

Graphical Abstract

Introduction

Hydroxyurea (HU, hydroxycarbamide) is a chemotherapeutic agent that inhibits ribonucleotide reductase, an enzyme that converts ribonucleotides into deoxyribonucleotides. This inhibition blocks DNA synthesis and repair, leading to proliferation arrest in the S-phase of the cell cycle and, consequently, cell death [1]. Nitric oxide (NO) also inhibits ribonucleotide reductase, as NO prodrugs and high-output NO synthase (NOS) deplete the deoxyribonucleotide pool, similar to HU [2]. HU-induced accumulation of DNA strand breaks is NOS-dependent [3]. HU is the therapy of choice for sickle cell anemia (SCA) because it increases NOS-dependent synthesis of fetal hemoglobin [4].

HU acts as an NO donor upon oxidation in the presence of heme proteins [5]. NO is produced by the reduction of nitrate and nitrite by NOS enzymes, which catalyze the oxidation of L-arginine to NO and citrulline [6]. Three NOS isoforms have been identified: the constitutive endothelial (NOS3) and neuronal (NOS1) forms, which produce low levels of NO, and the inducible (NOS2) form, which produces high levels of NO and is expressed in macrophages in response to inflammation [7]. Low levels of NO favor cell proliferation and anti-apoptotic responses, whereas higher levels of NO favor cell cycle arrest and apoptosis [8]. NOS2 is the simplest NOS, consisting of reductase, oxygenase, and tightly bound CaM-binding domains, but lacking the auto-inhibitory loop of NOS3 and the PDZ domain of NOS1 [9]. HU stimulates NOS-dependent NO levels in murine and human macrophages [10]. In addition to macrophages, NOS2 is expressed in hepatocytes, smooth muscle cells, chondrocytes, erythroid cells, and cardiac myocytes, as well as in many cancers, including colon, bladder, breast, and lung cancer [9].

The hypothesis that HU activates NOS2 is supported by previous reports showing that HU increased the expression of NOS1 and NOS3 proteins in erythroleukemic K562 cells [11], as well as NOS3 levels and activity in endothelial cells [12,13]. We examined the effects of HU treatment on protein expression and activity of the NOS2 isoform in murine erythroid progenitors (mERPs) and human erythroleukemic HEL92.1.7 cells. Molecular docking and dynamic simulations, along with in vitro enzymatic assays, were performed to determine whether NOS2 activation was due to direct interaction with HU. Using the NOS2-selective inhibitor 1400W or short hairpin RNA (shRNA)-mediated knockdown, we investigated whether enzymatic inhibition or decreased NOS2 expression was sufficient to impair the effects of HU on the proliferation and apoptosis of erythroleukemic cells. In vitro findings were supported by ex vivo data obtained from erythroid progenitors isolated from NOS2 knockout (Nos2^−/−) mice treated with HU. The results delineate the molecular mechanisms underlying HU regulation of NOS2 expression and activity, as well as the mediation of NOS2 in the HU inhibition of proliferation and induction of apoptosis in erythroid cells.

Materials and methods

Cell culture

HEL92.1.7, an erythroblast cell line isolated from the bone marrow of an erythroleukemic patient, which is commonly used to study erythroid cell differentiation, was obtained from ATCC (TIB-180, Manassas, Virginia, USA). Cells were grown in RPMI-1640 medium supplemented with 10% FBS, 1% glutamine, and 1% penicillin/streptomycin. Cells were incubated at 37°C with 5% CO₂ and maintained between 5 × 10⁵ and 1 × 10⁶ cells/mL, with medium renewal every 2–3 days. For experiments, 2 × 10⁶ cells were seeded in six-well plates and treated with HU (Abcam, ab142613, Cambridge, United Kingdom), L-NG-nitro arginine methyl ester (L-NAME, Sigma Aldrich, Louis, Missouri, USA), JSH23 (Merck, Darmstadt, Germany), or 1400W (Sigma Aldrich) for the indicated time periods. Inhibition of nuclear factor kappa B (NFκB) signaling was achieved following treatment with JSH23 (10 μM), the most effective concentration selected after dose-dependent studies. The cells were harvested and used for flow cytometry or cytospin preparation for immunocytochemistry. Whole-cell extracts were isolated for citrulline and nitrite measurements and Western blot analyses.

Nos2 knockdown with shRNA

Nos2 knockdown $\left(\text{NOS}{2}_{\text{kd}}\right)$ was generated by spinoculating 4 × 10⁵ HEL92.1.7 cells with lentiviral particles containing shRNA directed against NOS2 (MISSION^® shRNA, Sigma Aldrich) at a multiplicity of infection of 1. Cells were centrifuged at 800 × g, 37°C for 30 min, resuspended in a 24-well plate, and cultured in antibiotics-free medium for 24 h. The medium was replaced, and cells were incubated for an additional 48 h before selection with 8 mg/mL puromycin. To obtain 98% pure cultures of knockdown cells, we performed fluorescence-activated cell sorting (FACS) based on GFP expression using a BD FACS Melody instrument (Becton Dickinson, Franklin Lakes, New Jersey, USA).

Western blot

Whole-cell extracts were obtained from cell pellets by extraction in RIPA buffer (1 mM EDTA, 50 mM Tris-HCl pH 7.5, 0.1% SDS, 150 mM NaCl, 1% NP40, 1% sodium deoxycholate, protease inhibitor cocktail). The homogenates were separated on polyacrylamide gels and blotted onto polyvinylidene difluoride membranes (GE Healthcare, Chicago, Illinois, USA). Membranes were incubated overnight at 4°C with antibodies against phospho-NOS2 (Tyr151) (TA325739, OriGene, Rockville, Maryland, USA), NOS2 (E-AB-64331, Elabscience, Houston, Texas, USA), phosphorylated NFκB p65 (p-NFκB p65) (Ser536) (3031, Cell Signaling, Danvers, Massachusetts, USA), total NFκB p65 (E-AB-22066, Elabscience), phospho-p38 (Tyr180/Tyr182) (E-AB-21027, Elabscience), total p38 (E-AB-66279, Elabscience), phospho-p44/42 (extracellular signal-regulated kinase ½ [Erk1/2]) (Thr202/Tyr204) (9101, Cell Signaling), total p44/42 (Erk1/2) (9102, Cell Signaling), or $\text{β}$-actin (MAB8928, R&D Systems, Minneapolis, Minnesota, USA). Secondary antibodies conjugated to horseradish peroxidase (Elabscience) were detected using an enhanced chemiluminescence detection system (Bio-Rad, Hercules, California, USA). Protein bands were visualized using a ChemiDoc Imager (Bio-Rad) and quantified using ImageLab software.

Genotyping and treatment of Nos2^−/− mice

All mice were housed at the Rodent Housing Facility of the Institute for Medical Research, following institutional guidelines. Mice were maintained in a temperature- and humidity-controlled environment with a 12 h light/dark cycle, free access to water, and standard rodent chow. All experiments were performed using wild-type (WT) C57BL/6 and knockout B6.129P2-Nos2tm1Lau/J mice at 9–11 weeks of age. Nos2^−/− mice behaved similarly to WT mice, except that they were prone to obesity. Animal care and experiments were approved by the Institute for Medical Research Ethical Committee and carried out under the provisions of the Veterinary Directorate of the Ministry of Agriculture, Forestry, and Water Management (Approval Number 323–07-09964/2022–05). DNA was extracted from tail snips using phenol-chloroform and isopropanol after digestion in digestion buffer (100 mM NaCl, 10 mM Tris-HCl pH 8.0, 25 mM EDTA pH 8.0, and 0.5% SDS) with 10 μg/μL proteinase K (Sigma Aldrich). NOS2 deficiency was confirmed by genotyping and immunocytochemistry for Nos2 in the mERP, where no protein expression was detected. Nos2^−/− mice were obtained and genotyped using modified protocols from The Jackson Laboratory (Bar Harbor, ME, USA) PCR program. Bands were separated by 1.5% agarose gel electrophoresis and visualized using MidoriGreen dye (Nippon Genetics, Tokyo, Japan) on a ChemiDoc Imager (Bio-Rad). The primers used for genotyping were as follows: Nos2 common, TCA CCA CCA GCA GTA GTT GC; Nos2 WT forward, TCC GAT TTA GAG TCT TGG TGA; and Nos2 mutant forward, CCT TCT ATC GCC TTC TTG ACG.

HU was dissolved in drinking water at 200 mg/kg and renewed every 3 days. The mice were allowed to drink it for 2 weeks. At the end of the HU regimen, appropriate concentrations of NOS inhibitors in a 100 μL volume were injected subcutaneously in the left flank of the hindlimb twice daily for 3 consecutive days. The following inhibitors were dissolved: L-NAME in water at 70 mg/kg, 1400W in DMSO/ddH₂O (1:1) at 20 mg/kg, DPI in water at 1 mg/kg, and BNI in a mixture of sunflower oil and DMSO (1:10) at 25 mg/kg.

Immunomagnetic separation of mouse erythroid progenitors

We treated Nos2^−/− mice and WT mice with 200 mg/kg HU in drinking water for 2 weeks. To test the effect of Nos2 inhibition in vivo, WT mice were injected with 20 mg/kg of 1400W twice daily for 3 consecutive days. Bone marrow cells were used in the colony formation assay to assess hematopoiesis, and mERPs were isolated by immunomagnetic separation based on the expression of the surface markers CD71 and Ter119 (Supplementary Fig. 3b) to analyze the role of Nos2 in HU-induced proliferation inhibition and apoptosis. mERP cells were isolated from WT or Nos2^−/− mice treated with HU or 1400W using MACS^® cell separation instruments and reagents based on the expression of CD71 and Ter119. Bone marrow cells were flushed out of the femurs and tibias of WT or Nos2^−/− mice treated with HU or 1400W and resuspended in 100 μL of buffer (1% PBS, 0.5% BSA, 2 mM EDTA) and stained with 7 μL of anti-human CD71-PE REAfinity^™ antibody (Miltenyi Biotec, 130–120-809, Bergisch Gladbach, Germany) for 15 min at 4°C. Cells were incubated with anti-PE Multisort Microbeads (Miltenyi Biotec, 130–090-757) for 15 min at 4°C. After washing, cells were resuspended in 500 μL of buffer and run on an MS column placed on a magnetic stand previously activated with 500 μL of buffer. The column was washed with 500 μL of buffer, separated from the magnet, filled with 1 mL of buffer, and pressed with a plunger to collect the CD71+ cell population in a 10 mL tube. Next, 20 μL of Multisort Release Reagent was added, and the tube was incubated for 10 min at 4°C (Miltenyi Biotec, 130–090-757). Cells were resuspended in 60 μL of buffer and 30 μL of Multisort Stop Reagent (Miltenyi Biotec, 130–090-757) and 10 μL of anti-Ter119 antibody (Miltenyi Biotec, 130–049-901) was added. Immunophenotypic characterization of the mERP cells was performed by estimating the percentage of CD71-PE-positive and Ter119-FITC-positive cells using a BD FACSCalibur flow cytometer. The mERP cells were used for immunocytochemical and flow cytometric analyses.

Colony formation assay

Bone marrow cells obtained from mouse femurs were monodispersed in Dulbecco’s modified Eagle’s medium (Biowest, Lakewood Ranch, Florida, USA) supplemented with 5% fetal calf serum (Biowest). Afterward, 2 × 10⁵ and 3 × 10⁵ cells were diluted in methylcellulose-based medium containing 3 U/mL erythropoietin (MethoCult M3334, StemCell Technologies, Vancouver, Canada) or medium supplemented with 50 ng/mL stem cell factor, 10 ng/mL interleukin (IL)-3, and 10 ng/mL IL-6 (MethoCult GF M3434). The cells were seeded in duplicate in 35 mm tissue culture plates and placed in an incubator maintained at 37°C with 5% CO₂ and > 85% humidity. Colony-forming unit-erythroid (CFU-E) colonies were manually counted after 3 days. Culture dishes were scored for burst-forming unit-erythroid (BFU-E) and CFU-granulocyte/macrophage (CFU-GM) colonies after 14 days under an inverted phase-contrast microscope at a 4× objective magnification.

Immunocytochemistry

HEL92.1.7 or mERP cells were attached to microscope glass slides using cytospins (2 × 10⁵ HEL92.1.7 cells or 5 × 10⁵ mERP cells) and fixed using methanol at room temperature. Samples were first treated with 3% H₂O₂ solution in PBS to block endogenous peroxidase activity and then incubated with the following primary antibodies: anti-antigen Kiel 67 (anti-Ki67) antibody (Dako, M7187, Glostrup, Denmark), anti-ssDNA (Abcam, ab29585), anti-NOS2 (Santa Cruz Biotechnologies, sc-5302, Dallas, Texas, USA), anti-Caspase 3 (anti-Cas3; Novus Biologicals, NB100–56113, Centennial, Colorado, USA) antibodies in a humidity chamber overnight at 4°C. Immunostaining was performed using a streptavidin-biotin technique (UltraVision Quato Detection System HRP, TL-060-QHL; Thermo Fisher Scientific, Waltham, Massachusetts, USA). Immunoreactivity was visualized using a DAB Substrate Kit (Biotium, 30015, Fremont, California, USA) and counterstained with Harris hematoxylin (HHS32–1L; Sigma Aldrich). The number of Ki67-, ssDNA-, NOS2-, or Cas3-positive per total nuclei were counted in 10 images using ImageJ.

Citrulline and nitrite measurement

Citrulline and nitrite concentrations were measured in HEL92.1.7 cells treated with HU, 1400W, or L-NAME, and in bone marrow cells isolated from WT and Nos2^−/− mice treated with or without HU. Citrulline concentration was estimated using a colorimetric Homocitrulline/Citrulline Assay Kit (Abcam, ab242292), and absorbance was measured at 540 nm using a Rayto spectrophotometer. Nitrite (NO metabolite) was measured using a fluorometric NOS activity assay kit (Abcam, ab211084) according to the manufacturer’s instructions. Green fluorescence was measured using a VICTOR2 D fluorometer. Micromoles of citrulline and NO were calculated using the standard curve method.

Cell cycle distribution

Cell cycle distribution was assessed in HEL92.1.7 cells treated with vehicle, HU alone, or in combination with 1400W or L-NAME, as well as in mERP cells isolated from WT and Nos2^−/− mice treated with HU or 1400W. Cells were washed in PBS and fixed by adding ice-cold ethanol dropwise while vortexing and incubated for 2 h at 4°C. RNA was removed by adding 7 μL of 1 mg/mL RNAse followed by incubation for 1 h at 37°C and overnight at 4°C. Just before analysis by flow cytometry, 0.5 μL of 12.5 mg/mL PI was added, and samples were analyzed using a BD FACSCalibur flow cytometer. Data were analyzed using FlowJo v10.8.1 or Novos software.

Annexin V/PI apoptotic assay

Two million HEL92.1.7 cells were seeded in a six-well plate, treated with 100 μM HU for 48 h, and harvested. Cells were washed once in PBS, once in Annexin Binding Buffer (14 mM NaCl, 0.4 mM KCl, 75 μM MgCl₂, 1 mM N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid [HEPES]), and resuspended in 100 μL of Annexin Binding Buffer. RNA was removed by adding 10 μg/μL RNAse A and incubating the cells for 1 h at 37°C. Then, 5 μL of anti-Annexin V-FITC (BD, 556419) or Annexin V-APC antibody (BD, 550475) was added, and cells were incubated for 30 min at 4°C in the dark. Unbound antibody was washed off with PBS. Cells were resuspended in 500 μL, stained with 25 μg/μL PI (ThermoFisher Scientific, P1304MP) or a 1:50,000 dilution of ZombieGreen dye (BioLegend, 423111, San Diego, California, USA) and assayed using a BD FACSCalibur flow cytometer. Data were analyzed using FlowJo v10.8.1.

In vitro enzymatic assay

The in vitro enzymatic assay was performed using an NOS activity assay kit (Cayman Chemical, 781001, Ann Arbor, Michigan, USA). A master mix was prepared containing reaction buffer (25 mM Tris-HCl, pH 7.4, 3 μM tetrahydrobiopterin, 1 μM flavin adenine dinucleotide, and 1 μM flavin adenine mononucleotide), 2.5 mM L-arginine, 1.25 mM nicotinamide adenine dinucleotide phosphate, and 0.75 mM calcium chloride. In 40 μL of the reaction mix, 2 μL of 0.24 μg/μL human recombinant NOS2 (OriGene, TP311819) was added and complemented with ddH₂O up to 50 μL. In the reaction with HU, 10, 50, or 100 μM HU was added. The negative control was the same mixture incubated at 95°C for 5 min. The reaction was incubated for 30 min, 1 h, or 2 h at 37°C, and was subsequently stopped by adding 5 μL of stop buffer (50 mM HEPES, pH 5.5, and 5 mM EDTA). Nitrite and citrulline levels were measured as described previously.

Molecular docking simulation

The crystal structure of NOS2 (PDB ID: 1nsi) [14] was extracted from the RCSB Protein Data Bank in PDB format. The L-arginine binding sites were prepared for docking by removing co-crystallized ligands and water molecules using Discovery Studio 4.0. To examine the binding affinity of HU, AutoDock 4.2 software was used [15]. The NOS2 binding sites were determined using the AMDock program [16]. The AutoDockTools graphical user interface (GUI) was used to calculate the Kollman partial charges and add polar hydrogens. The Lamarckian Genetic Algorithm (LGA) was used for protein-ligand flexible docking [17]. Grid centers with dimensions of 16.240 × 61.960 × 13.986 ų in the -x, -y, and -z directions were used to cover the NOS2 protein-binding sites and allow ligands to move freely. The LGA method was employed for flexible protein-ligand docking, with the following parameters: the maximum number of energy evaluations set to 250,000, the maximum number of generations to 27,000, and mutation and crossover rates of 0.02 and 0.8, respectively. The interactions between the target proteins and HU were analyzed as three-dimensional results and illustrated using Discovery Studio 4.0 and AutoDockTools. The free energy of binding values were calculated using AutoDock software with the following equation:

$${\text{ΔG}}_{\text{bind}}={\text{ΔG}}_{\text{vdw}+\text{hbond}+\text{desolv}}+{\text{ΔG}}_{\text{elec}}+{\text{ΔG}}_{\text{total}}+{\text{ΔG}}_{\text{tor}}-={\text{ΔG}}_{\text{unb}}$$

where ${\text{ΔG}}_{\text{bind}}$ represents the estimated docking score. ${\text{ΔG}}_{\text{vdw}+\text{hbond}+\text{desolv}}$ denotes the sum of the energies of dispersion and repulsion $\left(\text{Δ}{\text{G}}_{\text{vdw}}\right)$, hydrogen bonds $\left({\text{ΔG}}_{\text{hbond}}\right)$, and desolvation (${\text{ΔG}}_{\text{desolv}}$). ${\text{ΔG}}_{\text{total}}$ represents the final total internal energy, ${\text{ΔG}}_{\text{tor}}$ is the torsional free energy, ${\text{ΔG}}_{\text{unb}}$ is the unbound system’s energy, and ${\text{ΔG}}_{\text{elec}}$ is the electrostatic energy.

Molecular dynamics

The best-docked complex, NOS2-HU, was used as the initial model for molecular dynamics simulations. The CHARMM36 force field was employed to parameterize HU [17]. Complex protein-ligand inputs were prepared for equilibration and production using the CHARMM-GUI web server [18]. A TIP3P solvation model was used to solve the system under investigation. Sodium chloride ions were added to neutralize the system at a salt concentration of 0.15 M in KCl. Using conjugate gradient and steepest descent algorithms, the neutralized system’s energy was minimized in 5,000 steps, with a tolerance of up to 1,000 kJ/mol·nm. Subsequently, the system was equilibrated in the constant volume (NVT) ensemble condition with a 2 ns time scale at 310.15 K using the Berendsen weak coupling method. The production MD phase was performed in the NPT ensemble for a 20 ns time scale using a modified Berendsen thermostat $(\text{τT}=1\text{ps})$ and a Parrinello–Rahman barostat $(\text{τP}=2\text{ps})$, applying the LINCS algorithm [19]. Trajectory propagation for up to 20 ns was simulated using the GROMACS 5.1.5 package [20].

Data analysis

All data were presented as the positive standard error of the mean (SEM). Significance was calculated by a two-tailed Student’s t-test or two-way analysis of variance using GraphPad Prism software with p-values of *p < 0.05, **p < 0.01, and ***p < 0.001.

Results

HU-induced NOS2 expression in erythroleukemic cells

Treatment of erythroleukemic HEL92.1.7 cells with HU for 48 h led to a dose-dependent increase in the number of NOS2-positive cells (40%) and NOS2 protein levels (3.5-fold) (Figs. 1a and 1b). Additionally, HU induced a significant increase in the phosphorylated NFκB (p-NFκB) p65 / total NFκB p65 ratio (Fig. 1c), indicating activation of NFκB signaling, which is known to promote NOS2 expression [21]. HU treatment of HEL92.1.7 cells also increased the phosphorylation of the mitogen-activated protein kinase (MAPK) subunit p38, another inducer of NOS2 expression, as well as a regulator of proliferation and apoptosis (Fig. 1d) [22]. The activation of MAPK signaling was further confirmed by the increased phospho/total ratio of ERK1/2 during HU treatment compared with that in untreated cells (Fig. 1e). Notably, short-term HU treatment had no effect on the total NOS2 levels (Supplementary Fig. 1a). The NFκB inhibitor JSH23 prevented the HU-induced long-term increase in NOS2 protein levels (Fig. 1e), confirming that NOS2 protein expression is regulated by NFκB signaling.

Figure 1. Hydroxyurea induces NOS2 expression in erythroid cells.

a) Immunocytochemistry for NOS2 protein in HEL92.1.7 cells treated with the indicated concentrations of hydroxyurea (HU) or vehicle (Ctrl) and quantification of NOS2-positive cells. b) Western blot analysis for NOS2 protein in HEL92.1.7 cells treated for 48 h with the indicated concentrations of HU or vehicle (Ctrl). Quantification of band intensity is shown, with $\text{β}$-actin used as a loading control and normalized to Ctrl. c) Western blot for phospho-NFκB p65 (Ser536) and total NFκB p65. d) Western blot for phospho-p38 and total p38. e) Western blot for phospho-ERK1/2 and total ERK1/2, with $\text{β}$-actin, on HEL92.1.7 cells treated with 100 μM HU for 5, 15, or 30 min, or vehicle (Ctrl). Quantification of band intensity is expressed as the phospho-to-total protein ratio and normalized to vehicle-treated cells. f) Western blot for NOS2 protein in HEL92.1.7 cells treated with the indicated concentrations of HU and/or 10 μM of JSH23 for 48 h. Quantification of band intensity is shown with $\text{β}$-actin used as a loading control and normalized to Ctrl. n = 3; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. Ctrl or as indicated.

Treatment of HEL92.1.7 cells with 10, 50, and 100 μM HU led to an increase in nitrite concentration compared to the control (Supplementary Fig. 1b). Specifically, 100 μM HU increased nitrite levels (145 ± 0.15 vs. 79.2 ± 0.54 μM), and this increase was significantly reduced by the pan-selective NOS inhibitor L-NAME, indicating NOS involvement in HU-induced NO production (61 ± 0.03 μM; Fig. 2a). Furthermore, treatment with the NOS2-specific inhibitor 1400W in combination with HU resulted in a decrease in NO levels compared to HU alone (56.2 ± 0.69 vs. 145 ± 0.15 μM), confirming that the NOS2 isoform plays a significant role in HU-induced NO production (Fig. 2a). Additionally, the parallel product of the NOS enzymatic reaction, citrulline, was increased in HEL92.1.7 cells upon 100 μM HU treatment compared to the control (209 ± 0.51 vs. 111 ± 0.54 μM; Fig. 2b and Supplementary Fig. 1c), and was decreased by the addition of L-NAME (82.2 ± 0.59 μM), further confirming HU activation of NOS (Fig. 2b).

Figure 2. Hydroxyurea directly activates NOS2 in vitro.

(a) Nitrite concentration in HEL92.1.7 cells treated with vehicle (Ctrl), HU, L-NAME, 1400W, or a combination of HU and L-NAME or 1400W. (b) Citrulline concentration in HEL92.1.7 cells treated with vehicle (Ctrl), HU, L-NAME, or a combination of HU and L-NAME. c) Western blot for phospho-NOS2 (Tyr151) and total NOS2 protein in HEL92.1.7 cells treated with the indicated concentrations of HU or vehicle (Ctrl). Quantification of band intensity is expressed as the phospho-NOS2 to total NOS2 ratio and normalized to vehicle-treated cells. d) Nitrite concentration in in vitro NOS enzymatic assays with the indicated concentrations of HU and incubation times. e) Citrulline concentration in in vitro NOS enzymatic assays with the indicated concentrations of HU and incubation times. f) In silico model of HU and NOS2 interaction showing binding at amino acids ASP382, ASP385, and ARG388, and with the substrate L-arginine (ARG700). g) Molecular dynamics analysis showing root mean square deviation (RMSD) of the $\text{C}-\text{C}\alpha -\text{N}$ backbone versus simulation time for NOS2 in complex with and without HU during 20 ns. h) Root mean square fluctuation (RMSF) values of the NOS2-HU complex plotted against residue numbers. j) Radius of gyration (Rg) plots of NOS2 receptor with and without HU in active sites during 20 ns. a-e) n = 3; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. Ctrl or 0.

HU directly activates NOS2 enzymatic activity in vitro

To test the effect of HU treatment on NOS2 activity in HEL92.1.7 cells, we measured the expression of the phosphorylated (Tyr151) active form of NOS2. Western blot analysis showed a time-dependent increase in the ratio of phosphorylated NOS2 relative to total NOS2 protein following HU treatment (Fig. 2c). Regulation of NOS3 activation and protein expression by HU has been previously reported [12,13]. To determine whether HU could directly modulate NOS2 activity, we performed an in vitro enzymatic assay using purified full-length human recombinant NOS2 protein, substrate L-arginine, and cofactors, and measured nitrite and citrulline concentrations. HU treatment led to a dose- and time-dependent increase in the concentration of the reaction products, nitrite (Fig. 2d) and citrulline (Fig. 2e), indicating that HU directly stimulated the enzymatic activity of NOS2.

Next, we examined the structure of human NOS2 [9] for potential binding pockets for HU with the substrate L-arginine and cofactors by molecular docking in silico. The binding affinity of HU to the active site of NOS2 was found to be –14.5 kJ mol^–1 with an inhibition constant of 4.31 mM (Table 1). Docking analyses revealed that hydrogen bond interactions were the most significant, due to the high number of polar groups in HU. Electrostatic interactions also significantly contributed to the binding energy and the stabilization of the NOS2-HU complex, whereas the torsional energies had a weak contribution to ${\text{ΔG}}_{\text{bind}}$ due to the small size and low flexibility of HU (Table 1). The amino acids ASP382, ASP385, and ARG388 in the primary NOS2 structure played predominant roles as active sites for ligand binding (Fig. 2f). ARG700, a substrate of L-arginine, also formed a hydrogen bond with HU (Fig. 2f). The structure obtained after the molecular docking simulation was used to examine the molecular dynamics of the system, including overall stability, local residues, and general structural fluctuations. Molecular dynamic analyses included root mean square deviation (RMSD), root mean square fluctuation (RMSF), and radius of gyration (Rg) (Figs. 2g and h). Direct changes in the protein from the initial coordinates were measured using RMSD. The RMSD values of the protein NOS2 backbone $(\text{C}-\text{C}\alpha -\text{N})$ with and without ligand in the active site were computed for the initial structure as a frame of reference (0–20 ns). The RMSD value for the NOS2-HU complex steadily increased from 0 to 1.5 ns and reached equilibrium throughout the simulation period. The average RMSD values for the NOS2-HU complex (7.17 ± 0.33) were slightly higher than those of the unbound NOS2 (6.88 ± 0.59), indicating that conformational changes occurred and there was a slight decrease in L-arginine binding site rigidity after HU binding (Fig. 2g).

Table 1.

The important parameters for best docking conformations of HU with protein targets.

Complex	$\text{Δ}{G}_{\text{bind}}$ (kJ mol−1)	${\text{K}}_{\text{i}}$ (mM)	${\text{ΔG}}_{\text{vdw}+\text{hbond}+\text{desolv}}$ (kJ mol−1)	${\text{ΔG}}_{\text{elec}}$ (kJ mol−1)	${\text{ΔG}}_{\text{total}}$ (kJ mol−1)	${\text{ΔG}}_{\text{tor}}$ (kJ mol−1)	${\text{ΔG}}_{\text{unb}}$ (kJ mol−1)
NOS2–HU	−13.5	4.31	−12.0	−1.5	−1.5	0.0	−1.5

The average RMSF values of the 6,862 amino acids of NOS2 in the presence and absence of HU were calculated to explore local protein flexibility. The average RMSF value for NOS2-HU (0.14 ± 0.10 nm) was higher than the unbound L-arginine binding site of NOS2 (0.1 ± 0.06 nm). Amino acids ASN366 and ARG700 in the active site of the NOS2-HU complex showed intense oscillations compared to unbound NOS2 (Fig. 2h). The average Rg value of the NOS2-HU complex (2.31 ± 0.01 nm) was higher than that of unbound NOS2 (2.28 ± 0.01 nm), indicating larger conformational changes in the secondary structure of NOS2 when HU was bound to the active site (Fig. 2j). These molecular dynamic simulations confirmed the stability of HU at the L-arginine binding site of NOS2. The docking complex NOS2-HU was further subjected to molecular dynamics energy contribution analysis using the gmx_MM/PBSA protocol, based on van der Waals, electrostatic, polar, and nonpolar solvation energies [16]. The values of the various contributions to the total energy are listed in Table 2. Based on the obtained results, it can be concluded that the ${\text{ΔE}}_{\text{elec}}$ and ${\text{ΔG}}_{\text{nonpolar}}$ were the major contributors to the total binding free energy.

Table 2.

Important thermodynamic parameters during 20 ns MD simulation generated with gmx_MM/PBSA protocol.

Complex	${\text{ΔE}}_{\text{elec}}$ (kJ mol−1)	${\text{ΔE}}_{\text{VDW}}$ (kJ mol−1)	${\text{ΔG}}_{\text{polar}}$ (kJ mol−1)	${\text{ΔG}}_{\text{nonpolar}}$ (kJ mol−1)	${\text{ΔG}}_{\text{binding}}$ (kJ mol−1)
NOS2-HU	−41.0±2.27	−38.1±0.02	54.8±0.92	−79.1±2.36	−24.3±2.53

NOS2 inhibition or knockdown prevented HU-induced inhibition of HEL92.1.7 cell proliferation

We analyzed the effects of NOS2 inhibition or knockdown on the cytostatic properties of HU. Erythroleukemic HEL92.1.7 cells were treated with increasing concentrations of the NOS2-specific inhibitor 1400W or the non-selective NOS inhibitor L-NAME in combination with HU, and we monitored the expression of the proliferation marker Ki67 by immunocytochemistry and cell cycle distribution by flow cytometry. Treatment with 1400W or L-NAME reversed the HU-induced decrease in the number of Ki67-expressing cells (Fig. 3a and Supplementary Fig. 2a). As expected, HU increased the percentage of cells in the S-phase of the cell cycle compared to untreated controls (69.2 ± 2.76 vs. 55.2 ± 1.06%) at the expense of the G0/G1 phase (15.2 ± 1.93 vs. 28.6 ± 1.00%), due to blocked DNA synthesis (Figs. 3b and c). However, 1400W did not affect cell cycle distribution, either alone or in combination with HU (Figs. 3b and c), while 1 mM L-NAME combined with HU increased the number of cells in the S-phase (73.4 ± 2.84 vs. 83.1 ± 3.06%), to the detriment of the G2/M phase (12.9 ± 2.33 vs. 6.55 ± 1.43%, Supplementary Fig. 2b and c). The pan-NOS inhibitor L-NAME, similar to the NOS2 inhibitor 1400W, prevented HU-mediated inhibition of proliferation but enhanced S-phase arrest, demonstrating that HU-mediated inhibition of proliferation is not strictly related to the S-phase of the cell cycle.

Figure 3. NOS2 inhibition or knockdown prevents HU-induced inhibition of HEL92.1.7 cell proliferation.

HEL92.1.7 cells were treated with the indicated concentrations of NOS2-specific inhibitor 1400W alone or in combination with HU. a) Immunocytochemistry for Ki67 protein and quantification of Ki67-positive cells as a proliferation marker. b) Cell cycle analysis by flow cytometry. Debris and doublets were excluded, and the distribution in the phases of the cell cycle was determined based on the incorporation of PI. c) Percentage of cells in G0/G1, S, or G2/M phases of the cell cycle was quantified. d) Gating of $\text{NOS}{2}_{\text{kd}}$ HEL92.1.7 cells based on GFP expression. e) Quantification of NOS1-, NOS2-, or NOS3-positive cells in $\text{NOS}{2}_{\text{kd}}$ and control HEL92.1.7 cells after immunocytochemistry. f) Immunocytochemistry for Ki67 protein in $\text{NOS}{2}_{\text{kd}}$ or control HEL92.1.7 cells treated or not with HU, and quantification of Ki67-positive cells. g) Example of gating for cell cycle distribution using PI by flow cytometry. Exclusion of debris, exclusion of doublets, distribution in G0/G1, S, or G2/M phases based on PI incorporation. h) Percentage of cells in G0/G1, S, or G2/M phases of the cell cycle in $\text{NOS}{2}_{\text{kd}}$ cells and controls treated or not with HU. n = 5; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. Ctrl or as indicated.

To support the data obtained from chemical NOS2 inhibition, we generated stable $\text{NOS}{2}_{\text{kd}}$ cells by transfecting HEL92.1.7 cells with lentiviral particles containing shRNA directed against the NOS2 transcript and sorting the cells based on GFP expression (Fig. 3d). This resulted in a HEL92.1.7 $\text{NOS}{2}_{\text{kd}}$ clone with a ~78 ± 2.56% reduction in NOS2 protein levels without affecting the expression of NOS1 and NOS3 (Fig. 3e). $\text{NOS}{2}_{\text{kd}}$ significantly increased the number of Ki67-positive cells, both alone and in combination with HU (Fig. 3f). Furthermore, $\text{NOS}{2}_{\text{kd}}$ significantly decreased the number of cells in the S-phase of the cell cycle compared to untreated controls (32.2 ± 0.41 vs. 55.1 ± 1.33%), while increasing the number of cells in G0/G1 (47.7 ± 0.43 vs. 26.8 ± 0.46%). The combination of $\text{NOS}{2}_{\text{kd}}$ and HU led to an increase in the S-phase population compared to $\text{NOS}{2}_{\text{kd}}$ alone (48.4 ± 1.5 vs. 32.2 ± 0.4), but this was significantly lower than with HU treatment alone (48.4 ± 1.5 vs. 72 ± 2.91%, Figs. 3g and 3h). These results showed that NOS2 inhibition rescued the HU-induced decrease in Ki67 frequency in erythroleukemic cells, but could not overcome the HU-induced S-phase blockade, as seen in $\text{NOS}{2}_{\text{kd}}$.

NOS2 inhibition or knockdown prevented HU-induced apoptosis in HEL92.1.7 cells

To test the involvement of NOS2 in HU-induced apoptosis, we performed immunocytochemical staining for the early apoptotic marker ssDNA, and an Annexin V/PI apoptosis assay on HEL92.1.7 cells and $\text{NOS}{2}_{\text{kd}}$ cells in the presence or absence of HU, treating them with 1400W or L-NAME. Both 1400W and L-NAME significantly decreased HU-induced ssDNA expression, most significantly at the highest applied concentrations (Fig. 4a and Supplementary Fig. 3a). In the Annexin V/PI apoptotic assay, HU increased the percentage of early (14.4 ± 0.44 vs. 10.2 ± 0.31%) and late apoptotic HEL92.1.7 cells (6 ± 0.37 vs. 2.85 ± 0.10%) compared to untreated controls (Figs. 4b and d). 1400W treatment decreased both early and late apoptosis compared to untreated cells, most effectively at 100 μM (4.9 ± 0.13 vs. 10.2 ± 0.31% and 0.9 ± 0.04 vs. 2.85 ± 0.10%, respectively, Figs. 4b–d). Importantly, 100 μM 1400W was able to decrease the HU-induced number of early and late apoptotic cells to levels similar to the control (1.94 ± 0.05 vs. 6.03 ± 0.37% and 10.2 ± 0.33 vs. 14.4 ± 0.44%, Figs. 4b–d). Notably, inhibition of all three NOS isoforms with 5 mM L-NAME increased early (21.7 ± 0.67 vs. 8.7 ± 0.12%) and late (7.3 ± 0.19 vs. 4.74 ± 0.12%) apoptosis compared to untreated cells. The combination of L-NAME and HU had a synergistic effect on cell death compared to HU alone (34.2 ± 3.54 vs. 23.36 ± 0.31% for early, and 9.42 ± 1.04 vs. 6.36 ± 0.23% for late apoptosis, Supplementary Figs. 3b–d). Similar to NOS2 inhibition, $\text{NOS}{2}_{\text{kd}}$ was able to decrease the HU-induced number of ssDNA-expressing HEL92.1.7 cells (Fig. 4e). The Annexin V/PI assay showed a decreased percentage of $\text{NOS}{2}_{\text{kd}}$ early apoptotic cells with or without HU compared to HU-treated HEL92.1.7 cells with endogenous NOS2 levels of 8.18 ± 0.62 or 8.72 ± 0.35 vs. 15.76 ± 1.92%, respectively. However, $\text{NOS}{2}_{\text{kd}}$ was not able to rescue the HU-induced late apoptotic population (5.98 ± 0.33 vs. 6.11 ± 0.21%, Figs. 4f and g). In contrast to pan-NOS inhibition, these results indicate that NOS2 is critical for HU-induced apoptosis in erythroleukemic cells.

Figure 4. NOS2 inhibition or knockdown prevents HU-induced apoptosis in HEL92.1.7 cells.

HEL92.1.7 cells were treated with the indicated concentrations of NOS2-specific inhibitor 1400W alone or in combination with HU. a) Immunocytochemistry for ssDNA and quantification of ssDNA-positive cells. b) Example of gating for the apoptotic assay by flow cytometry. The bottom left quadrant represents early apoptotic cells, and the upper left quadrant represents late apoptotic cells. c) Percentage of early apoptotic cells. d) Percentage of late apoptotic cells. e) Immunocytochemistry for ssDNA on $\text{NOS}{2}_{\text{kd}}$ or control HEL92.1.7 cells treated or not with HU and quantification of ssDNA-positive cells. f) Example of gating for the apoptotic assay by flow cytometry with numbers of early and late apoptotic cells. g) Percentage of early and late apoptotic cells. n = 5; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. Ctrl or as indicated.

In vivo HU treatment of Nos2^−/− mice impairs HU-induced inhibition of proliferation in erythroid progenitors

To investigate the involvement of Nos2 in the molecular mechanism of HU in vivo, we treated Nos2^−/− mice and their WT littermates with 200 mg/kg HU in drinking water for 2 weeks (Fig. 5a). NOS2 deficiency was confirmed by genotyping (Supplementary Fig. 4a) and immunocytochemistry for NOS2 in the mERP, where no protein expression was detected (Supplementary Fig. 4b). To assess the effect of Nos2 inhibition in vivo, WT mice were injected with 20 mg/kg of 1400W twice daily for 3 consecutive days (Fig. 5a). Bone marrow cells were used in the colony formation assay to evaluate hematopoiesis, and mERP were isolated by immunomagnetic separation based on the expression of surface markers CD71 and Ter119 (Supplementary Fig. 4c). Corresponding cultures were used to analyze the role of Nos2 in HU-induced inhibition of proliferation and apoptosis. An increase in NOS2 frequency was observed in mERP treated with HU compared to untreated WT controls (Fig. 5b).

Figure 5. In vivo HU treatment of Nos2^−/− mice impairs HU inhibition of proliferation in erythroid progenitors.

a) Schematic representation of experimental setup: Nos2^−/− or wild-type (WT) mice were treated orally with 200 mg/kg HU or drinking water for 2 weeks. WT mice were injected with 20 mg/kg of 1400W twice daily for 3 consecutive days. Mouse erythroid progenitors (mERP) were isolated from bone marrow by immunomagnetic cell separation using anti-CD71-PE and anti-Ter119-FITC antibodies. b) Immunocytochemistry for Nos2 protein in mERP isolated from WT mice treated or not with HU. Quantification of Nos2-positive cells. c) Citrulline concentration in the bone marrow of WT and Nos2^−/− mice treated or not with HU. d) Colony formation assay showing the number of late erythroid (CFU-E), early erythroid (BFU-E), or granulocyte/macrophage progenitors (CFU-GM) in the bone marrow of WT or Nos2^−/− mice treated or not with HU. e) Immunocytochemistry for Ki67 in mERP cells isolated from WT and Nos2^−/− mice treated or not with HU. f) Quantification of Ki67-positive cells. g) Cell cycle distribution by flow cytometry showing the percentage of cells in G0/G1, S, or G2/M phases of the cell cycle. c) n = 3, f) n = 5; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. WT.

Bone marrow cells from WT mice treated with HU showed increased citrulline production compared to untreated mice, whereas Nos2 ablation abolished this increase (Fig. 5c). To evaluate the effect of in vivo HU treatment on proliferation and hematopoiesis, we performed a colony formation assay using bone marrow cells isolated from Nos2^−/− mice and WT controls treated with HU. HU treatment led to a decrease in early erythroid (BFU-E, p < 0.05) and late erythroid (CFU-E, p < 0.01) progenitors, without affecting the number of CFU-GM colonies (Fig. 5d). HU treatment in Nos2^−/− mice resulted in an increase in both late and early erythroid progenitors, favoring erythropoiesis (Fig. 5d). In addition to CFU-GM, Nos2 deficiency was sufficient to overcome the HU-induced decrease in CFU-E and BFU-E colonies, compared to HU-treated WT and untreated mice (Fig. 5d).

Next, we isolated mERP from WT, Nos2^−/−, or 1400W-injected mice treated with HU or vehicle and analyzed the expression of Ki67 by immunocytochemistry and cell cycle distribution by flow cytometry. HU treatment decreased the percentage of Ki67-expressing mERP compared to controls, with a smaller decrease in Nos2^−/− mice and an increase in 1400W-treated mice (Figs. 5e and f). In vivo treatment of WT mice with HU led to an increase in S-phase (31.1 ± 8.8% vs. 12.7 ± 2.43%) and a decrease in the G0/G1 phase compared to untreated mERP (60 ± 8.7% vs. 80.7 ± 1.88%, Fig. 5g and Supplementary Fig. 3d). HU also increased the S-phase population in mERP from Nos2^−/− mice compared to untreated Nos2^−/− mice (28.7 ± 4% vs. 17.6 ± 1.9%) and in 1400W-treated compared to vehicle-treated mice (25.5 ± 2.2% vs. 11.2 ± 0.47%; Fig. 5g and Supplementary Fig. 3d). The S-phase distribution of mERP isolated from HU-treated Nos2^−/− mice or HU and 1400W-treated WT mice (25.5 ± 2.37% or 28.7 ± 4%, respectively, vs. 31.1 ± 8.8%) did not significantly change compared to HU-treated WT cells (Fig. 5g and Supplementary Fig. 4d). The results in mice were consistent with those obtained in HEL92.1.7 cells, confirming the partial involvement of Nos2 in HU inhibition of proliferation. Nos2^−/− mice were able to partially rescue the HU-induced decrease in the percentage of Ki67-positive cells but could not overcome S-phase blockade. In contrast to Nos2-deficient mice, $\text{NOS}{2}_{\text{kd}}$ in human cells reduced HU-induced S-phase cell accumulation, suggesting the possibility of in vivo compensation by constitutive NOS isoforms. Moreover, it has been reported that NO production by activation of human NOS2 is different from that produced by mouse Nos2 [23].

In vivo HU treatment of Nos2^−/− mice impairs HU stimulation of apoptosis in erythroid progenitors

Apoptosis was assessed ex vivo in mERP from WT, Nos2^−/−, or 1400W-injected mice treated with HU or vehicle by analyzing the expression of Cas3 via immunocytochemistry and performing the Annexin V/PI assay using a flow cytometer. HU treatment of WT mice increased the frequency of Cas3-positive mERP compared to that in untreated mice (Fig. 6a and b). Nos2 ablation or Nos2 inhibition decreased the number of Cas3-expressing cells when combined with HU (Fig. 6a and b). In the Annexin V/PI assay, mERP from HU-treated WT mice showed an increased percentage of early apoptotic cells compared to untreated littermates (27.2 ± 3.63% vs. 15.4 ± 2.48%), while the percentage of late apoptotic cells remained unchanged (Figs. 6c and f). mERP from HU-treated Nos2^−/− mice exhibited a dramatic reduction in the number of early apoptotic cells (9.19 ± 0.22% vs. 27.2 ± 3.63%), while the number of late apoptotic cells increased (9.66 ± 1.03% vs. 2.6 ± 0.27%; Figs. 6c and f). However, the pooled percentage of all apoptotic mERP was significantly reduced in HU-treated Nos2^−/− mice compared to HU-treated WT mice (Fig. 6f). Nos2 inhibition combined with HU was sufficient to decrease early (16.7 ± 0.27% vs. 15.4 ± 2.48%) and late apoptosis (3.04 ± 0.1% vs. 2.43 ± 0.38%) in mERP to levels similar to those of WT mice without HU treatment (Figs. 6c and f). Our results show that Nos2 inhibition in vivo decreased both early and late apoptosis of mERP induced by oral HU treatment, whereas NOS2 ablation rescued only cells in early apoptosis. The data obtained from the mERP analysis strongly support those from HEL92.1.7 erythroleukemic cells (Figs. 4b and d) and underline the involvement of the NOS2 protein in HU-induced apoptosis of erythroid cells.

Figure 6. In vivo HU treatment of Nos2^−/− mice impairs HU stimulation of apoptosis in erythroid progenitors.

a) Immunocytochemistry for Cas3 in mouse erythroid progenitors (mERP) after treatment of WT or Nos2^−/− mice with 1400W and/or HU. b) Quantification of Cas3-positive mERP. c) Example of gating for Annexin V/PI apoptotic assay by flow cytometry with numbers of early and late apoptotic mERP cells. d) Percentage of early apoptotic mERP cells. e) Percentage of late apoptotic mERP cells. f) Percentage of total apoptotic mERP cells. n = 5; mean + SEM, *p < 0.05, **p < 0.01, ***p < 0.001 vs. WT.

Discussion

Our results demonstrate that HU can produce NO in erythroid cells through the increased expression and activity of NOS2. We previously showed that HU increases NO production by elevating NOS3 expression and activity in endothelial and erythroleukemic cells [11–13]. Short-term NOS2 activation and long-term augmentation of NOS2 protein expression simultaneously affect NO production. Consistent with our study, HU has been shown to induce NfκB activation in K562 cells and p38 in mice [24,25]. In addition, we demonstrated NfκB dependence in the prolonged augmentation of NOS2 protein levels by HU.

Importantly, we demonstrated that HU directly increases NOS2 activity in vitro. In silico molecular docking suggested that HU interacts via hydrogen bonds with NOS2 and its substrate, L-arginine. Furthermore, molecular dynamics simulations showed a significant decrease in rigidity after the binding of HU to the NOS2 active site, corroborating the docking results. Our model suggests that HU binding might enhance the catalytic activity of NOS2 by supporting enzyme-substrate binding; however, further structural biological analyses are necessary to confirm the NOS2-HU interaction. Further exploration of these interactions could shed light on the mechanisms underlying the pharmacological effects of HU and its analogs, leading to the development of novel therapies targeting NOS2-related pathways.

NOS2 inhibition and knockout demonstrated that NOS2 contributes to HU-induced inhibition of erythroid cell proliferation. Consistent with our findings, NOS2 overexpression reduced in vitro cancer cell proliferation and in vivo tumor progression in xenograft models [26]. Inhibition of NOS2/Nos2 proteins and the absence of the Nos2 gene prevented HU inhibition of proliferation.

However, only transcriptional regulation of NOS2 was sufficient to further prevent HU-induced accumulation of cells in the S-phase. Nos2 deficiency rescued the reduction in BFU-E and CFU-E colonies upon in vivo HU treatment. This is in accordance with our previous study showing that the NO scavenger 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide can reverse HU inhibition of CFU-E formation [27].

NOS2 is implicated in apoptosis during various physiological and pathological processes. NOS2 is indispensable for neutrophil turnover and homeostasis through the induction of Cas8 and Cas3 [28]. Additionally, NOS2 induces apoptosis in cardiomyocytes and neurons [29,30]. In pancreatic cancer, NOS2 expression is positively correlated with the apoptotic index [31]. Our finding that NOS2 is involved in HU-mediated apoptosis aligns with the literature and contributes to a better understanding of the molecular mechanisms of action of HU. However, results obtained with the pan-selective NOS inhibitor L-NAME showed that L-NAME treatment in combination with HU increased both early and late apoptosis in HEL92.1.7 cells compared to HU treatment alone. Although this may seem contradictory to the data showing decreased apoptosis with the NOS2-specific inhibitor 1400W, it is not surprising given the dual effect of NOS enzymes on proliferation and apoptosis. Low levels of NO produced by constitutive NOS promote tumor progression and metastasis, whereas high levels of NO produced by NOS2 induce cell cycle arrest, senescence, and apoptosis [8]. In line with this, NOS1 and NOS3 have been shown to promote proliferation and suppress apoptosis in many cancer types, whereas NOS2 has been found to have pro-apoptotic effects [32–34]. Therefore, NOS enzymes can balance cell survival and death by regulating NO levels. It is possible that the pan-NOS inhibitor L-NAME enhances HU-induced apoptosis by abolishing the protective effect of low, continuous NO production by NOS1/3, which itself promotes proliferation. Moreover, inhibiting all three NOS isoforms may hinder cell survival due to the critical roles these enzymes play in key cellular processes, as evidenced by the low survival of NOS1–3 triple knockout mice [7,35]. Therefore, it is of great scientific interest to precisely define the role of NOS isoforms in the molecular mechanisms of HU to fully understand the differences in apoptosis rates between pan-NOS and NOS2 inhibition.

Similar to NO, the prevailing mechanism of HU inhibition is that transient nitroxide-like radicals from HU quench tyrosyl free radicals at the active site of ribonucleotide reductase [36]. Is NO production mediated by HU chemical degradation or enzymatic stimulation of NOS? HU increases intravascular NO levels in patients with SCA [1], as supported by the NOS substrate L-arginine [37,38]. HU decreases arginase activity with a concomitant induction of NOS activity in SCA [39,40]. Moreover, HU augmented plasma nitrite levels and attenuated plasma arginase levels in humanized sickle cell mice [41]. Therefore, HU increases NO levels by inhibiting arginase, resulting in greater availability of arginine for NOS, which further supports the major role of NOS in HU stimulation of NO production, corroborated by the direct interaction of HU and NOS2 in the present study. Chemical and biological oxidants, including iron-containing proteins, also in vitro converted HU into NO [42]. In vivo studies have provided strong evidence that NO from nitrosyl hemoglobin (HbNO) is produced by the metabolism of the NOH group of HU, whereas in vitro incubation of blood with HU did not produce detectable HbNO, demonstrating the relevance of the vasculature [43]. HbNO formation in venous erythrocytes is mainly of vascular NO origin, regardless of erythrocyte NOS activity [44]. Moreover, HbNO formation during the incubation of HU with blood is too slow to justify in vivo HbNO formation or NO production [45]. These reports highlight the importance of NOS enzymes in vascular endothelial and immune cells for HbNO formation during HU treatment.

This study had two major limitations that should be addressed in future research. First, regarding cell cycle distribution analysis, the quantification of total DNA content with PI could be supported by 5-bromo-2-deoxyuridine (BrdU) labeling of active DNA synthesis during the S-phase of the cell cycle. This would be particularly useful for HU- and L-NAME-treated HEL92.1.7 cells, where we detected an increase in the S-phase population, allowing the clear separation of BrdU-positive cells in G1 from the early and late S-phase from the G2/M phase. Another minor limitation is the inability to measure the exact concentration of HU administered to mice per os; however, this is justified by mimicking the route of administration used in patients and is addressed by using relatively high concentrations of HU.

Conclusion

This study demonstrated that HU increases NOS2 protein expression in erythroid cells by activating the NfκB signaling pathway. HU also potentiated NOS2 enzymatic activity in vitro, possibly through direct interaction with the enzyme and by increasing its binding affinity to the substrate L-arginine. NOS2 significantly mediates the inhibition of proliferation by HU. S-phase arrest was rescued only in vitro after NOS2 transcript depletion in human erythroleukemic cells. Furthermore, NOS2 was essential for the induction of apoptosis by HU. NOS2 enzyme inhibition and Nos2 gene/NOS2 transcript depletion effectively prevented early apoptosis in vitro and ex vivo, while late apoptosis was efficiently prevented only in vitro following NOS2 enzyme inhibition. Future studies should explore the specific roles of NOS1 and NOS3 isoforms, as the observed differences in the effects of NOS2-specific versus pan-NOS inhibition highlight the potentially diverse functions of NOS isoforms in the molecular mechanisms of HU action. Understanding the different aspects of HU’s mechanism of action may help enhance the response to therapy and reduce its negative effects.

Supplementary Material

This manuscript includes supplementary files.

Highlights.

• Hydroxyurea increases NOS2 protein expression via Nfkb signaling

• Hydroxyurea increases NOS2 enzymatic activity via direct interaction and by increasing substrate binding

• NOS2 is partially involved in HU-induced erythroid cell proliferation inhibition by regulating Ki67 expression

• NOS2 inhibition or knock-down strongly prevents HU-induced apoptosis of human erytroleukemic and mouse erythroid cells

Abbreviations

BFU-E

Burst-forming unit-erythroid

Cas

Caspase

CD71

Cluster of differentiation 71

CFU-E

Colony-forming unit-erythroid

CFU-GM

Granulocyte/macrophage colonies

ERK1/2

Extracellular signal-regulated kinase 1/2

FACS

Fluorescence-activated cell sorting

GFP

Green fluorescent protein

HbNO

Nitrosyl hemoglobin

HU

Hydroxyurea

Ki67

Antigen Kiel 67

LGA

Lamarckian genetic algorithm

MAPK

Mitogen-activated protein kinase

mERP

Murine erythroid progenitor

NFκB

Nuclear factor kappa B

NO

Nitric oxide

NOS1

Neuronal nitric oxide synthase

NOS2

Inducible nitric oxide synthase

Nos2^−/−

Nos2 knockout

NOS2kd

NOS2 knockdown

NOS3

Endothelial nitric oxide synthase

PBS

Phosphate-buffered saline

PI

Propidium iodide

Rg

Radius of gyration

RMSD

Root mean square deviation

RMSF

Root mean square fluctuation

SCA

Sickle cell anemia

shRNA

Short hairpin RNA

ssDNA

Single-stranded DNA

Ter119

Surface protein associated with glycophorin-A

WT

Wild-type

$\text{ΔGbind}$

Estimated docking score

$\text{ΔGdesolv}$

Desolvation

$\text{ΔGelec}$

Electrostatic energy

$\text{ΔGhbond}$

Hydrogen bond

$\text{ΔGtor}$

Torsional free energy

$\text{ΔGtotal}$

Final total internal energy

$\text{ΔGunb}$

Unbound system energy

$\text{ΔGvdw}$

Energies of dispersion and repulsion

Data availability

The immunoblotting images are available at https://zenodo.org/records/14018919. Other data are available upon request.