Hydroquinone Modulates the GM-CSF Signaling Pathway in TF-1 Cells

JH Zheng1, DW Pyatt1,2, SA Gross1,ATLe1, PJ Kerzic1 and RD Irons1,3,4,5

1Molecular Toxicology and Environmental Health Sciences Program, School of Pharmacy, University of Colorado Health Sciences Center, Denver, CO, USA; 2Department of Preventative Medicine, School of Medicine, University of Colorado Health Sciences Center, Denver, CO, USA; 3Department of Pathology, School of Medicine, University of Colorado Health Sciences Center, Denver, CO, USA; 4Comprehensive Cancer Center, University of Colorado Health Sciences Center, Denver, CO, USA; and 5Sino-US Joint Clinical and Molecular Laboratory, Fudan University, Shanghai, China

Human leukemogens, including alkylating chemotherapeutic expense of erythroid CFU.4–6 We have also observed clonogenic agents and benzene, enhance granulocyte–macrophage col- enhancement in human CD34 þ BM cells treated in vitro with ony-stimulating factor (GM-CSF)-dependent proliferation of the BZ metabolite, hydroquinone (HQ), and then cultured in the human CD34 þ bone marrow (BM) cells. The extracellular 7,8 signal-regulated kinase (ERK) pathway plays an important role presence of exogenous rhGM-CSF. Cell proliferation is a well- in GM-CSF-dependent proliferation and also has been impli- recognized risk factor in carcinogenesis; cell division resulting cated in the pathogenesis of acute myelogenous leukemia. in a number of chemical and mechanical stresses that increase Therefore, we investigated the effects of the benzene metabo- the likelihood of acquiring structural genomic changes. Indivi- lite, hydroquinone (HQ), on alterations in the GM-CSF signaling dual malignancies, particularly hematopoietic and lymphoid pathway in TF-1 erythroleukemia cells and human CD34 þ BM cells. HQ treatment in TF-1 cells results in a strong proliferative cancers, often exhibit unique patterns or signatures of clonal response that is dependent on ERK activation and GM-CSF lesions that are thought to play important roles in either the production. HQ also induces ERK-dependent AP-1 activation initiation or pathogenesis of these individual diseases. Never- with concomitant increased transcriptional activity of AP-1 theless, little is actually known about the orchestration of the reporter gene. However, the kinetics of ERK activation are earliest events that result in the evolution of cell- and disease- different between rhGM-CSF and HQ in TF-1 cells: rhGM-CSF type-specific clonal lesions in human leukemogenesis. We results in immediate activation of ERK, whereas HQ activation of ERK is delayed. Further, HQ and rhGM-CSF together produce hypothesize that chemicals with leukemogenic potential can an immediate increase in ERK phosphorylation, which is mimic or synergize with molecular mediators of the GM-CSF sustained for over 48 h. HQ also stimulates colony formation, receptor mitogen-signaling pathway resulting in increased AP-1 DNA binding and GM-CSF production in human CD34 þ clonal proliferation in the target cell population that constitutes BM cells. These results suggest that HQ stimulates proliferation the origin for sMDS and sAML. via activation of ERK/AP-1 and is at least partially mediated via Granulocyte–macrophage colony-stimulating factor (GM- the production of GM-CSF. Leukemia (2004) 18, 1296–1304. doi:10.1038/sj.leu.2403389 CSF) signaling is mediated via a variety of kinase cascades (eg Published online 6 May 2004 MAPK, Janus kinase (JAK)/signal transducers and activators of Keywords: GM-CSF signaling; activation of ERK; activator transcription 5 (Stat5), protein kinase C (PKC), phosphatidyl- protein-1 (AP-1); hydroquinone (HQ); cell proliferation inositide 3 (PI3) Kinase).9,10 However, cytokine-mediated stimulation of these kinase pathways ultimately results in AP-1 activation.11 In particular, the MAPK (eg ERK, Jun N-terminal kinase (JNK) and p38 kinase) have been demonstrated to play important roles in the control of both cellular growth and Introduction cytokine production.12,13 ERK is the best characterized pathway for GM-CSF-induced proliferation resulting from the activation Acute myelogenous leukemia (AML) developing secondary to of AP-1, and constitutive activation of ERK is frequently exposure to alkylating chemotherapeutic agents or chronic observed in leukemia cells.11,12,14 Therefore, we decided to exposure to benzene (BZ) frequently progresses from initial BM characterize the molecular events occurring in the ERK pathway injury or hypoplasia to myelodysplastic syndrome (sMDS) and following treatment of TF-1 cells or primary human CD34 þ sAML.1–3 A variety of studies suggest that the cell of origin in cells with HQ. sMDS/sAML is a multilineage progenitor cell capable of TF-1 cells exhibit a strong proliferative response to differentiation to myeloid and erythroid cells (ie colony-forming GM-CSF that is thought to be mediated through the Ras/Raf/ unit-granulocyte, erythroid, macrophage, megakaryocyte (CFU- MEK/ERK pathway.15–17 We used TF-1 cells to examine GEMM)).1–3 A variety of human leukemogens, including the mechanisms of HQ-induced alterations in clonogenic cyclophosphamide, melphalan and BZ, result in enhanced response associated with GM-CSF-dependent signal transduc- CFU response, following exposure both in vitro and in vivo, tion and compared these results with those obtained which is dependent upon hematopoietic cytokines, such as GM- using primary human CD34 þ BM cells. Our results show that CSF.3–5 Increases in granulocyte CFU have been observed in HQ stimulates proliferation of TF-1 cells via activation of mice within 2 h of exposure to BZ in vivo, apparently at the the ERK/AP-1 signalling pathway together with concomitant increases in GM-CSF production and clonogenic response. Correspondence: Dr RD Irons, Department of Pathology, University of In contrast, HQ enhancement of CFU response in human Colorado Health Sciences Center, 4200 E 9th Avenue, C-238, Denver, CD34 þ BM cells is dependent on the addition of CO 80262, USA; Fax: þ 1 303 315 7223; E-mail: richard.irons@ exogenous rhGM-CSF,7,8 but similarly is accompanied by AP- uchsc.edu 1 activation and an increase in GM-CSF production. Taken This project was supported by NIH Grant ES06258 to Dr Richard D Irons. collectively, these studies suggest that HQ exerts many of its Received 11 August 2003; accepted 12 March 2004; Published online hematopoietic effects via alterations in the GM-CSF signaling 6 May 2004 pathway. Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1297 Materials and methods complexes, and the samples were analyzed by electrophoresis on a prerun 6% polyacrylamide gel. The gels were dried and CD34 þ /CD19À cell purification exposed at À801C to Fuji x-ray film (Tokyo, Japan).

Cell purification was performed as previously described.18 All protocols were approved by the University of Colorado Health Western blot of ERK kinase activity Sciences Center Internal Review Board. BM aspirates were obtained from normal adult volunteers after informed consent. Treatment was performed and the proteins were prepared as Briefly, mononuclear cells were isolated using Lymphocyte described above. Approximately 50 mg of protein was analyzed Separation Medium (Cellgro, Herndon, VA, USA) and CD34 þ / by SDS polyacrylamide gel electrophoresis and transferred to CD19À cells were purified using a high magnetic gradient Mini- Nitrocellulose membrane (Bio-Rad, CA, USA). The membranes MACS purification system (Miltenyi, Sunnyvale, CA, USA). Our were probed with specific phospho-p44/p42 antibody (Bio-Lab, technique included a combination of a depletion column CA PhosphoPlus p44/42 MAP kinase antibody Kit), which (CD19 þ ) followed by a positive selection column (CD34 þ ). detects the activated phosphorylated forms of ERK kinase (ERK1 The purity of isolated cells (495% CD34 þ /CD19À) was and ERK2). The same membranes were stripped and reprobed determined using flow cytometric analysis (Coulter Electronics, with nonphospho-p44/42 antibody, which detects total ERK Hialeah, FL, USA). kinase (phosphorylation-state independent) as a control for sample variation. Western analysis of membranes was performed using a Super Signal West Pico Trail Kit (Pierce, IL, USA) Cell culture and treatment followed by exposure of ECL-film (Boehringer Mannheim, IN, USA). TF-1 cells, a growth factor-dependent human erythroleukemia 15 cell line originally obtained from ATCC (Manassas, VA, USA), Transfection assay were maintained in complete RPMI-1640 media (Gibco BRL, Grand Island, NY, USA) supplemented with 10% fetal calf TF-1 cells were deprived of rhGM-CSF for 4–6 h and resus- serum, 100 mg/ml streptomycin, 100 U/ml penicillin, and pended in hypoosmolar electroporation buffer (Eppendorf, 2 mmol/l L-glutamine in the presence of optimal concentrations Hamburg, Germany) to a final concentration of 1 Â 106 cells/ of rhGM-CSF (5 ng/ml) (Immunex, Seattle, WA, USA) at 371C, in ml. Cells were cotransfected with an experimental reporter humidified air containing 5% CO . Experimental TF-1 cells 2 gene, 4AP-1, and/or a control reporter gene, pRL-TK, by were deprived of rhGM-CSF overnight (16 h) followed by electroporation at 400 V and a charging time of 100 ms, using chemical or growth factor treatment. HQ (Sigma) was dissolved a Multiporater (Eppendorf, Hamburg, Germany). The 4AP-1 in phosphate-buffered saline (PBS, Gibco BRL) and added to the gene was constructed by inserting four copies of AP-1 sites cells to yield the final concentrations indicated. Cells were (50- GAT CGA TCC GGC TGA CTC ATC ACT AG -30) into a cultured for various times as indicated in the results. For the pGL3 vector driving the firefly luciferase gene (a gift from Dr study of ERK activation, rhGM-CSF-deprived TF-1 cells were Vasilis Vasiliou). The pRL-TK construct consists of the Renilla preincubated with MEK 1 inhibitor, PD98059 (BioLabs, Beverly, Luciferase gene under the control of the TK promoter (Promega, MA, USA), at various concentrations for 30 min prior to HQ or Madison WI, USA). Cells were treated 24 h with HQ or rhGM- rhGM-CSF treatment. Cells were then harvested at specific time CSF, washed and lysed in 50 ml of Passive Lysis Buffer points for each experiment. Purified CD34 þ cells were (Promega). Cell lysates (10 ml) were analyzed for both Firefly resuspended in complete RPMI-1640 media followed by 16 h (F) and Renilla (R) luciferase activity using the Dual-Luciferase chemical or growth factor treatment. Assay System (Promega). Each experiment was performed in duplicate, and the average of the two expressed as relative Electrophoretic mobility shift assay (EMSA) luciferase activity (F/R).

TF-1 cells and/or purified CD34 þ /CD19À cells were treated as [3H]Thymidine uptake assay described above and cell extracts were prepared as previously 18 described. Protein concentrations were determined using a Proliferation studies were performed in 96-well flat-bottom BCA protein kit (Piece, Rockford, IL, USA) and the protein was plates with 5 Â 104 cells/well (five wells/sample) in complete frozen at À801C until future analysis. AP-1 consensus sequences RPMI-1640 media. TF-1 cells were pulsed with 1 mCi/well of 0 0 0 5 TCC GCT TGA TGA GTC AGC CGG AA 3 and 5 TCT TCC [3H]thymidine (DuPont NEN.USA) for 24 h and collected onto 0 GGC TGA CTC ATC AAG CG 3 were annealed and labeled as filtermats using an automated cell harvester (Skatron, UK) and 19 previously described. Protein (4–6 mg) was incubated on ice counted as CPM using a liquid scintillation analyzer (Packard for 10 min with 1 mg dI-dC, 4 ml binding buffer (20 mmol/l HEPES Inst. Co. Meriden, CT, USA). (pH 7.9), 40 mmol/l KCl, 10% glycerol, 0.05 mmol/l EDTA, 1.6 mmol/l MgCl2), 1 mmol/l DTT, and deionized water for a total volume of 19 ml. Next, 1 mlof32P-labeled probe Colony-forming (CFU) assays (B50 000 cpm) was added and the binding reaction was continued for 30 min at 221C. For supershift samples, appro- CFU assays were performed as previously described.8 Briefly, priate specific antibodies against Fos family (c-Fos , Fos B, Fra1, chemical- and/or growth factor-treated TF-1 cells were plated in Fra2) and Jun family (phospho-c-Jun, Jun D, Jun B, c-Jun) (Santa 35-mm culture dishes at 8 Â 103 cells/ml in Iscove’s modified Cruz, CA, USA) were added to the binding buffer before protein Dulbecco’s medium (10% fetal bovine serum, 100 mg/ml addition. After complex formation, 2 ml of loading buffer streptomycin, 100 U/ml penicillin, 2 mmol/l L-glutamine, (250 mmol/l Tris HCL (pH 7.5), 0.2% bromophenol blue, 0.2% 50 mmol/l 2-mercaptoethanol, 1.2% (wt/vol) methyl cellulose) xylene cyanol, 40% glycerol) was added to the DNA–protein (Gibco, Grand Island, NY, USA). Cultures were maintained at

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1298 371Cin5%CO2 and colonies counted on day 14. A cluster of cells numbering greater than 50 was scored as a colony. Five plates were scored for each treatment group, and signiﬁcant differences (Po0.05) in colonies/culture observed between control and treated groups were determined using the Student’s t-test, and the results expressed as the mean71 standard error of the mean (s.e.m.).

Determination of [GM-CSF] in culture supernatants

Culture supernatants were analyzed for [GM-CSF] using a Human GM-CSF Immunoassay Kit (BioErgonomics, St Paul, MN, USA) following manufacture instruction. Briefly, supernatants were removed from cell cultures at 2 h, 4 h and/or 16 h post chemical and/or growth factor treatment and mixed with GM-CSF-specific capture beads and incubated for 1 h at room temperature. Capture beads were removed via centrifugation, washed and resuspended in buffer containing a GM-CSF reporter antibody, which is conjugated with phycoerythrin. Supernatant [GM-CSF] was then determined using a Coulter Epics XL Cytometer calibrated against a standard curve for known concentrations of GM-CSF. These results were confirmed using a GM-CSF neutralizing antibody.

GM-CSF neutralizing antibody assay

Dependence of TF-1 cell proliferation on GM-CSF production was determined using a GM-CSF neutralizing antibody, which is selected for its ability to neutralize the biological activity of both rhGM-CSF and natural human GM-CSF (R&D Systems, Min- neapolis, MN, USA). TF-1 cells were pretreated with either control goat IgG (10 mg/ml) or increasing doses of GM-CSF neutralizing antibody for 30 min prior to the addition of either 10 mM HQ or 0.5 ng/ml rhGM-CSF and then the incubation was continued for an additional 12 h followed by [3H]thymidine incorporation assay performed as described above. All experiments described in this study were repeated at least three separate times. Figure 1 HQ induces proliferation and colony formation in TF-1 cells. TF-1 cells deprived of rhGM-CSF were incubated for 16 h with either rhGM-CSF (5 ng/ml) or HQ (10 mM). (a) Cell proliferation was 3 m Results measured by pulsing cells with [ H]thymidine (1 Ci/well) and quantitated by liquid scintillation counting. The values were expressed as the mean7s.e.m. for five wells. (b) CFU was determined in cells HQ stimulates proliferation and colony formation cultured for 14 days in methylcellulose. A cluster of cells numbering in TF-1 cells greater than 50 was scored as a colony and colonies expressed as the mean7s.e.m. for five cultures. n Indicates significant increase relative A variety of hematotoxic alkylating agents or their metabolites to control cultures (Po0.05). stimulate GM-CSF-dependent proliferation and colony formation in human CD34 þ BM cells. In order to determine whether TF-1 cells respond in a manner similar to primary BM cells, we examined the effects of rhGM-CSF and HQ on proliferation and HQ þ rhGM-CSF, and ERK phosphorylation was measured at CFU in TF-1 cells in culture. Overnight, rhGM-CSF-deprived various time points over a 48 h period. Since ERK kinase consists TF-1 cells were incubated for 16 h with predetermined optimal of two protein kinases (ERK44 and ERK42) that are encoded by concentrations of either HQ (10 mM) or rhGM-CSF (5 ng/ml) and Erk1/Erk2, full enzymatic activation of ERK requires phosphor- 20,21 assayed for proliferation (Figure 1a) and CFU (Figure 1b). ylation on specific tyrosine and threonine sites. Treatment Treatment of TF-1 cells with either rhGM-CSF or HQ produces with HQ or rhGM-CSF results in activation of ERK in a time- significant increases in proliferation and CFU in TF-1 cells. dependent manner (Figure 2a). ERK is activated by 4 h, maximal Increases in cytokine-dependent CFU are analogous to increases at 12 h and is visible at 48 h following treatment with HQ. in the recruitment of individual cell clones to undergo However, rhGM-CSF treatment results in immediate ERK proliferation and differentiation. activation that peaks at 10 min and progressively declines over 24 h. Further, treatment of TF-1 cells with HQ and rhGM-CSF together results in immediate activation of ERK, which is HQ activates ERK kinase in TF-1 cells sustained at high levels for 12 h and remains demonstrable at 48 h (Figure 2a). In order to confirm the dose dependence of We measured ERK activation at specific time points post- HQ-induced ERK activation, TF-1 cells were treated with treatment. TF-1 cells were treated with HQ, rhGM-CSF or increasing doses of HQ and pERK measured at 16 h (Figure 2b).

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1299 In order to determine the speciﬁcity of ERK activation, TF-1 cells were pretreated with increasing doses of the MEK1 inhibitor, PD98058, then followed by treatment with 10 mM HQ. PD98058 is a selective inhibitor of the ERK kinase

Figure 3 The MEK1 inhibitor, PD98059, abrogates HQ-induced proliferation and CFU in TF-1 cells. TF-1 cells were deprived of rhGM- CSF and preincubated with increasing doses of PD98059 for 30 min, followed by the addition of HQ (10 mM) or rhGM-CSF (5 ng/ml) and incubated for 16 h. TF-1 cells were (a) pulsed with [3H]thymidine (1 mCi/well) and incorporation quantified by liquid scintillation counting, or (b) cultured in methylcellulose for 14 days and CFU enumerated. One representative experiment of three independent experiments is shown. Values were expressed as mean7s.e.m. of four wells or cultures. n Indicates a significant decrease relative to HQ- treated cells (Po0.05); nn Indicates a significant decrease relative to rhGM-CSF-treated cells (Po0.05).

cascade.20,21 PD98058 binds tightly to inactive MEK1 and prevents its activation by upstream signals such as Raf. TF-1 cells pretreated with PD98058 exhibit a dose-dependent inhibition of HQ-induced activation of ERK kinase (Figure 2c). ERK protein

Figure 2 HQ induces a time- and concentration-dependent activation of ERK in TF-1 cells measured by Western Blot analysis using anti-phospho-ERK antibody. (a) rhGM-CSF–deprived TF-1 cells were treated with either rhGM-CSF (5 ng/ml), HQ (10 mM) or both, and whole-cell extracts were prepared at speciﬁc time points. (b) Cells were treated with increasing concentrations of HQ, and ERK was measured at 16 h. (c) Cells were pretreated with increasing concentrations of the MEK1 inhibitor, PD98059, followed by treatment with HQ or rhGM-CSF for 16 h. Total ERK protein was detected with nonphoshorylated-ERK antibody. Phosphorylated ERK (pERK) and nonphosphorylated ERK (ERK) were used as positive and negative controls, respectively. (M ¼ size marker; C ¼ control cells).

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1300 levels do not appear to be affected by treatment with rhGM-CSF, activation of ERK pathway is required for cell proliferation and HQ and/or PD98059. These results suggest that HQ-induced for colony formation by HQ in TF-1 cells. activation of ERK occurs via phosphorylation, and that induction of ERK protein is not involved in the pathway of ERK activation. HQ induces ERK-dependent AP-1 DNA binding in TF-1 cells HQ-induced proliferation and colony formation in TF-1 cells is dependent on activation of ERK kinase The ERK pathway is known to be a major activator for AP-1 in many cells.16,22 Therefore, we investigated whether HQ- In order to determine whether HQ-induced cell proliferation induced activation of ERK coincides with AP-1 DNA binding and enhanced CFU was dependent on activation of the ERK in TF-1 cells. TF-1 cells were deprived of rhGM-CSF and then kinase pathway, TF-1 cells were preincubated with increasing treated with either rhGM-CSF or increasing concentrations of doses of PD98059 prior to incubation with either HQ or rhGM- HQ for 16 h. Treatment with HQ results in a concentration- CSF. TF-1 cells, pretreated with PD98059, exhibit a dose- dependent increase in AP-1 DNA binding (Figure 4a) that is dependent inhibition of both proliferation (Figure 3a) and CFU inhibited by pretreatment with the MEK1 inhibitor, PD98059 (Figure 3b) induced by HQ. These results demonstrate that (Figure 4b). The speciﬁcity of AP-1 DNA binding was demon-

Figure 4 HQ induces ERK-dependent AP-1 DNA binding in TF-1 cells. (a) Cells were harvested after 16 h treatment with increasing doses of HQ or rhGM-CSF (5 ng/ml), and EMSA was performed on protein from whole-cell lysates. (b) Cells were preincubated with increasing doses of PD98059 for 30 min and then incubated with HQ (10 mM) or rhGM-CSF (5 ng/ml), for 16 h. (c) HQ-induced AP-1 DNA binding consisting of Fos and Jun family members, which were identiﬁed using speciﬁc antibodies against Fos and Jun family members.

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1301 strated by supershift experiments with antibodies against the Fos and Jun family. These results demonstrate that HQ-activated AP- 1 DNA binding is ERK speciﬁc, and the complexes are composed of Fos B, Fra1, Fra2, Phospho-c-Jun, Jun D and Jun B at 16 h post-treatment (Figure 4c).

HQ transactivates an AP-1-dependent reporter gene in TF-1 cells

In order to determine whether HQ-induced AP-1 DNA binding results in an increase in transcriptional activation of an AP-1- dependent gene, we transfected TF-1 cells with an AP-1- dependent reporter gene that was generated by inserting four copies of AP-1 (4AP-1) binding sites into a pGL3 vector. TF-1 cells, deprived of rhGM-CSF, were transfected, exposed to rhGM-CSF (5 ng/ml) or increasing concentrations of HQ, and the luciferase activity was determined after 24 h as described in the materials and methods. HQ induced a concentration-dependent increase in activation of the reporter gene 4AP-1 (Figure 5). Although rhGM-CSF treatment produced a strong AP-1 DNA Figure 5 HQ induces concentration-dependent transactivation of binding response, it resulted in a smaller 4AP-1 reporter gene an AP-1-dependent reporter gene. TF-1 cells were cotransfected with activation signal than HQ. Taken together, these results AP-1-dependent reporter gene (4AP-1) driving the ﬁreﬂy luciferase demonstrate a strong correlation between HQ-induced AP-1 gene (F) and control reporter gene pRL-TK that consists of the Renilla Luciferase gene (R), followed by increasing doses of HQ or rhGM-CSF DNA binding and activation of the AP-1 promoter in TF-1 cells (5 ng/l ml) treatment for 24 h. Cells were harvested and luciferase independent of cellular stimulation with rhGM-CSF. activity was measured in the cell lysate. Values are expressed as relative luciferase activity (F/R). The average of two duplicate experiments is presented with the range indicated by the bars and HQ produces an increase in GM-CSF production omitted when the value is smaller than the histogram line. in TF-1 cells

One of the many known activities of GM-CSF is the stimulation of its own production.23,24 Enhanced clonogenic response to of its own production. Therefore, we hypothesized that HQ- GM-CSF is frequently observed in AML,25,26 and GM-CSF- induced activation of the GM-CSF signaling pathway and dependent proliferation and clonogenic response has been enhanced cell proliferation might be mediated via the produc- repeatedly demonstrated to be mediated via the Ras/Raf/ERK/ tion of GM-CSF. rhGM-CSF-deprived TF-1 cells were treated MAP kinase pathway.12,17 Independently, exposure to certain with increasing concentrations of HQ and [GM-CSF] measured alkylating agents, including cyclophosphamide, melphalan and in culture supernatants at the indicated time points. HQ BZ, results in an enhanced clonogenic response (ie CFU) and treatment revealed no detectable [GM-CSF] at both 2 and 4 h proliferation of human CD34 þ BM cells that is also dependent (data not shown), but results in a dose-dependent increase in on GM-CSF.7,8 Particularly for BZ, increases in granulocyte CFU [GM-CSF] at 16 h (Figure 6a). Independently, pretreatment of have been observed in mice within 2 h of exposure in vivo, TF-1 cells with neutralizing anti-GM-CSF antibody abrogated apparently at the expense of erythroid CFU.4–6 Clonogenic HQ-induced proliferation (Figure 6b). enhancement is also observed in human CD34 þ BM cells treated in vitro for 30 min with the BZ metabolite, HQ, and 7,8 HQ activates AP-1 DNA binding and GM-CSF cultured in the presence of exogenous rhGM-CSF. Activation production in primary human CD34 þ BM cells of the ERK/MAP kinase pathway is also involved in GM-CSF- induced proliferation in TF-1 cells.17 Therefore, we hypothe- Based on our results obtained using TF-1 cells, we evaluated sized that proliferation and clonal expansion of human CD34 þ AP-1 DNA binding and GM-CSF production in primary human multilineage hematopoietic progenitor cells might be mediated CD34 þ /CD19À BM cells. Puriﬁed CD34 þ /CD19À cells were via activation of the ERK/MAP kinase pathway, and that TF-1 treated with HQ and then cultured in complete RPMI-1640 cells might serve as useful surrogates for primary CD34 þ BM media supplemented with or without rhIL-3, Epo and SCF, in cells in the study of the molecular mechanisms of chemical- absence of rhGM-CSF for 16 h. Whole lysates were assayed for induced alteration in GM-CSF receptor signaling. AP-1 activation by EMSA (Figure 7a) and the supernatants were Previous studies have demonstrated that several alkylating assayed for [GM-CSF] by ImmunoFlow beads (Figure 7b). These chemotherapeutic agents share the ability to alter early events in results indicate that HQ stimulates AP-1 DNA binding and GM- the proliferation and recruitment of multilineage hematopoietic 1–3 CSF production in primary human CD34 þ BM cells that is progenitor cells that are the cells of origin in sMDS and sAML. comparable to that observed in TF-1 cells. The earliest common event appears to be enhanced clonogenic response of hematopoietic progenitor cells to GM-CSF. Pretreatment of human CD34 þ BM cells with HQ results in a Discussion 50–200% increase in the outgrowth of clones in response to rhGM-CSF.7,8 TF-1 cells exhibit a strong proliferative and GM-CSF is a pleiotropic hematopoietic cytokine that exerts clonogenic response to rhGM-CSF, which is similar to that multiple functions on hematopoietic progenitor cells (HPC) previously characterized in CD34 þ HPC.12,15 Our ﬁndings including proliferation, survival, differentiation and stimulation demonstrate that in both CD34 þ and TF-1 cells, HQ results in

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1302

Figure 6 HQ induces the production of GM-CSF in TF-1 cells. (a) The culture supernatants from cells treated with increasing doses of HQ (0.1, 1, 5, 10, 20 mM) in two independent experiments (indicated as m and ’) were measured at 16 h for [GM-CSF] using flow cytometry techniques as described in Materials and methods. Each value represents 500–1000 individual events. Concentrations were obtained by normalizing flow experiments against a standard curve obtained using the same method. (b) Confirmation of GM-CSF production in HQ-treated TF-1 cells was determined by adding a GM-CSF neutralizing antibody to the cell cultures prior to the addition of HQ. After 16 h incubation, treated cells were pulsed with [3H]thymidine (1 mCi/well) over an additional 24 h and the radioactivity incorporated by the cells was quantified by liquid scintillation counter. One representative experiment of three independent experiments is shown. The values are expressed as mean7s.e.m. of five wells. n Indicates a significant decrease compared to the HQ-treated sample (Po0.05); nn Indicates a significant decrease compared to the rhGM-CSF- treated sample (Po0.05).

an increase in AP-1-DNA binding and GM-CSF production. Our One explanation for these findings is that HQ alone findings also show that rhGM-CSF and HQ activate AP-1 in TF-1 modulates the GM-CSF signaling pathway, specifically invol- cells and that the HQ-induced activation of AP-1 is ERK- ving the activation of ERK and AP-1, which ultimately results in specific. However, the kinetics of ERK activation differ GM-CSF production. TF-1 cells transfected with activated Raf significantly between rhGM-CSF and HQ: rhGM-CSF-induced exhibit activation of ERK and the secretion of an autocrine phosphorylation occurs within 10 min, and subsides over 24 h; growth factor, which results in the abrogation of TF-1 cell HQ-induced phosphorylation is delayed, appears by 4 h and cytokine dependency.16 The cytokine, TNFa has also been subsides over 48 h; phosphorylation induced by concomitant demonstrated to stimulate proliferation in HU-3, MO7e and TF- treatment with HQ and GM-CSF is immediate, sustained at high 1 cell lines by inducing the production of GM-CSF in these levels for 12 h and is clearly demonstrable at 48 h. cells.27 Independently, autocrine production of fibroblast

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1303 Acknowledgements

We gratefully acknowledge Dr Vasilis Vasiliou for the generous gift of the 4AP-1 reporter gene, Wayne Stillman, Yanzhu Yang, Karen Helm for technical assistance and Ann Louden for the manuscript preparation. We also thank Dr Yanli Ouyang and Dr Ronda Baker for their constructive comments.

References

1 Sandoval C, Pui C-H, Bowman LC, Heaton D, Hurwitz CA, Raimondi SC et al. Secondary acute myeloid leukemia in children previously treated with alkylating agents, intercalating topoisome- rase II inhibitors, and irradiation. J Clin Oncol 1993; 11: 1039–1045. 2 Casciato DA, Scott JL. Acute leukemia following prolonged cytotoxic agent therapy. Medicine 1979; 58: 32–47. 3 Foucar K, McKenna RW, Bloomfield CD, Bowers TK, Brunning RD. Therapy-related leukemia. Cancer 1979; 43: 1285–1296. 4 Irons RD, Stillman WS. Cell proliferation and differentiation in chemical leukemogenesis. Stem Cells 1993; 11: 235–242. 5 Irons RD, Stillman WS, Colagiovanni DB, Henry VA. Synergistic action of the benzene metabolite hydroquinone on myelopoietic stimulating activity of granulocyte/macrophage colony-stimulating factor in vitro. Proc Natl Acad Sci USA 1992; 89: 3691–3695. 6 Irons RD. Effects of the benzene metabolite, hydroquinone, on regulation of differentiation and proliferation of human hematopoietic stem and progenitor cells in vitro. Toxicologist 1997; 36: 105–106. 7 Irons RD, Stillman WS. Impact of benzene metabolites on differentiation of bone marrow progenitor cells. Environ Health Perspect 1996; 104 (Suppl 6): 1247–1250. 8 Irons RD, Stillman WS. The effects of benzene and other leukaemogenic agents on haematopoietic stem and progenitor cell differentiation. Eur J Haematol 1996; 57 (Suppl 60): 119–124. 9 De Groot RP, Coffer PJ, Koenderman L. Regulation of proliferation, Figure 7 HQ activates AP-1 DNA binding and GM-CSF produc- differentiation and survival by the IL-3/IL-5/GM-CSF receptor tion in primary human CD34 þ cells. (a) Purified CD34 þ /CD19À family. Cell Signal 1998; 10: 619–628. cells were treated with HQ, cultured for 16 h and then whole-cell 10 Wheadon H, Roberts PJ, Watts MJ, Linch DC. Changes in signal lysates were assayed for AP-1 activation by EMSA. One representative transduction downstream from the granulocyte–macrophage col- experiment is presented out of three independent experiments ony-stimulating factor receptor during differentiation of primary conducted. (b) Purified CD34 þ /CD19À cells were treated with hemopoietic cells. Exp Hematol 1999; 27: 1077–1086. increasing doses of HQ and cultured for 16 h in complete RPM-1640 11 Wisdom R. AP-1: one switch for many signals. Exp Cell 1999; 253: media supplemented with IL-3, EPO and SCF, in absence of rhGM- 180–185. CSF, and supernatants were assayed for [GM-CSF] by flow cytometry 12 Platanias LC. Map kinase signaling pathways and hematologic using ImmunoFlow beads (described in Figure 6). One representative malignancies. Blood 2003; 101: 4667–4679. experiment out of three independent experiments is shown. 13 Esnault S, Malter JS. Extracellular signal-regulated kinase mediates granulocyte–macrophage colony-stimulating factor mesenger RNA stabilization in tumor necrosis factor-a plus fibronectin-activated peripheral blood eosinophils. Blood 2002; 99: 4048–4052. growth factor has been demonstrated to induce delayed 14 Kim SC, Hahn JS, Min YH, Yoo NC, Ko YW, Lee WJ. Constitutive activation of ERK and AP-1 in human lung fibroblasts in activation of extracellular signal-regulated kinase in human acute response to TGFb1.28 leukemias: combined role of activation of MEK, hyperexpression of The response of TF-1 cells to HQ differs slightly from primary extracellular signal-regulated kinase, and downregulation of a human CD34 þ BM cells: specifically HQ alone stimulates phosphatase, PAC1. Blood 1999; 93: 3893–3899. 15 Kitamura T, Tange T, Terasawa T, Chiba S, Kuwaki T, Miyagawa K proliferation and CFU in TF-1 cells. However, HQ is unable to et al. Establishment and characterization of a unique human cell elicit a proliferative or clonogenic response in primary CD34 þ line that proliferates dependently on GM-CSF, IL-3, or erythro- BM cells in the absence of concomitant stimulation with rhGM- poietin. J Cell Physiol 1989; 140: 323–334. CSF.7,8 Similar to TF-1 cells, HQ-induced increased clonogenic 16 McCubrey JA, Steelman LS, Hoyle PE, Blalock WL, Weinstein- response in human CD34 þ BM cells is preceded by increased Oppenheimer C, Franklin RA et al. Differential abilities of AP-1 DNA binding and appears to be dependent on the activated Raf oncoproteins to abrogate cytokine dependency, prevent apoptosis and induce autocrine growth factor synthesis in production of GM-CSF. One possible explanation is that, due human hematopoietic cells. Leukemia 1998; 12: 1903–1929. to constitutive low-level production of GM-CSF, TF-1 cells may 17 Kolonics A, Apati A, Janossy J, Brozik A, Gati R, Schaefer A et al. have a lower ERK activation threshold compared to primary Activation of Raf/ERK1/2 MAP kinase pathway is involved in GM- CD34 þ BM cells. Alternatively, these results could be CSF-induced proliferation and survival but not in erythropoietin- explained by HQ induction of another secondary mediator. induced differentiation of TF-1 cells. Cell Signal 2001; 13: Future studies will be needed to ascertain in greater detail the 743–754. 18 Kerzic PJ, Pyatt DW, Zheng JH, Gross SA, Le A, Irons RD. molecular mechanisms responsible for HQ-induced AP-1 Inhibition of NF-kappaB by hydroquinone sensitizes human bone activation, production of GM-CSF and their respective roles in marrow progenitor cells to TNF-alpha-induced apoptosis. Toxicol- proliferation and survival in CD34 þ BM cells. ogy 2003; 187: 127–137.

Leukemia Hydroquinone modulates the GM-CSF signaling pathway JH Zheng et al 1304 19 Pyatt DW, Zheng JH, Stillman WS, Irons RD. Inorganic lead 24 Metcalf D. The granulocyte–macrophage colony-stimulating fac- activates NF-kappaB in primary human CD4+ T lymphocytes. tors. Science 1985; 229: 16–22. Biochem Biophys Res Commun 1996; 227: 380–385. 25 Willman CL, Whittaker MH. The molecular biology of acute myeloid 20 Payne DM, Rossomando AJ, Martino P, Erickson AK, Her JH, leukemia. Proto-oncogene expression and function in normal and Shabanowitz J et al. Identification of the regulatory phosphoryla- neoplastic myeloid cells. Clin Lab Med 1990; 10: 769–796. tion sites in pp42/mitogen-activated protein kinase (MAP kinase). 26 Haase D, Fonatsch C. Karyotype and in vitro-response to GM-CSF. EMBO J 1991; 10: 885–892. Blut 1990; 60: 192–197. 21 Alessi DR, Cuenda A, Cohen P, Dudley DT, Saltiel AR. PD 098059 27 Quentmeier H, Dirks WG, Fleckenstein D, Zaborski M, Drexler is a specific inhibitor of the activation of mitogen-activated HG. Tumor necrosis factor-alpha-induced proliferation requires protein kinase kinase in vitro and in vivo. JBiolChem1995; 270: synthesis of granulocyte-macrophage colony-stimulating factor. 27489–27494. Exp Hematol 2000; 28: 1008–1015. 22 Karin M. The regulation of AP-1 activity by mitogen-activated 28 Finlay GA, Thannickal VJ, Fanburg BL, Paulson KE. Transforming protein kinases. J Biol Chem 1995; 270: 16483–16486. growth factor-beta 1-induced activation of the ERK pathway/ 23 Kinoshita T, Yokota T, Arai K, Miyajima A. Suppression of activator protein-1 in human lung fibroblasts requires the autocrine apoptotic death in hematopoietic cells by signalling through the induction of basic fibroblast growth factor. J Biol Chem 2000; 275 IL-3/GM-CSF receptors. EMBO J 1995; 14: 266–275. (36): 27650–27656.

Leukemia