Proteomic Analysis Identifies Oxidative Stress Induction by Adaphostin

Adaphostin (NSC 680410) is a potent anticancer agent with an elusive mechanism of action. As a member of the tyrphostin family, this compound belongs to a group of chemically and mechanistically diverse inhibitors of protein tyrosine kinases (1). Adaphostin was originally identified as a more active congener of AG957, a non-ATP–competitive inhibitor of p210^Bcr/abl (2–4). The antiproliferative effects of adaphostin and AG957 are associated with degradation of p210^bcr/abl polypeptide and the rapid induction of apoptosis (5, 6). Adaphostin differs from AG957 in its 3- to 4-fold improved ability to promote Bcr/abl degradation in vitro and slightly enhanced activity with respect to 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide cytotoxicity assays (4, 5). This mechanism contrasts with ATP-dependent inhibitors of Bcr/abl, such as STI571 (Gleevec, imatinib mesylate), which prevent autophosphorylation without polypeptide degradation and induce apoptosis after longer exposure periods (18-48 h; refs. 6, 7). Further evidence pointing toward mechanistic differences came from the observation that adaphostin and STI571 synergize against T-lymphoblastic leukemia cell lines, and that adaphostin can kill STI571-resistant clones (6, 8). The specificity of adaphostin for Bcr/abl was subsequently challenged after it was shown to have activity against leukemias and glioma cell lines that do not express Bcr/abl (9, 10). This led to the suggestion that adaphostin was either a “promiscuous” kinase inhibitor or had an entirely unrelated activity.

Insights into an alternative mechanism come from reports showing adaphostin-induced cell death is accompanied by increases in reactive oxygen species (ROS; refs. 9, 11). It has been suggested that because adaphostin contains a substituted 1,4-dihydroquinone, electron transfer has the potential to generate superoxide radicals and this accounts for the increased ROS levels (9). Support for this hypothesis comes from structure-activity relationships showing that the 1,4-dihydroquinone motif is essential for activity (4). Another possible explanation for increased ROS comes from a recent cDNA array study, which linked activity to increased expression of transcripts involved in iron metabolism (12). It was subsequently shown that adaphostin promotes the release of chelatable free iron (Fe²⁺) and that this may catalyze the formation of toxic hydroxyl radicals through Fenton's reaction [H₂O₂ + Fe²⁺→ Fe³⁺ + OH•]. In spite of this wealth of experimental data, several observations remain unresolved and as a consequence the underlying molecular basis for activity has yet to be determined. For instance, adaphostin has been shown to inhibit secretion of vascular endothelial growth factor, leading to the proposal that, like other tyrphostins (e.g., SU5416, semaxanib), adaphostin has antiangiogenic activity (10). Similarly, alterations in signal transduction pathways (RAF-1/mitogen-activated protein kinase kinase/extracellular signal-regulated kinase, AKT, c-met, and p38 mitogen-activated protein kinase) and adaphostin selectivity for certain malignancies (acute myelogenous leukemia, chronic myelogenous leukemia) requires further attention (13–15).

In this study, a proteomics platform based on two-dimensional PAGE was used to identify proteins modulated by adaphostin with the intention of gaining further insight into a mechanism of action and to identify surrogate markers of activity. Total cell lysate from two adaphostin-treated myeloid leukemia cells lines, HL60 (Bcr/abl negative) and K562 (Bcr/abl positive), formed the basis of the study. Results showed a proteomic profile consistent with an oxidative stress response and induction of apoptosis. Transfection of one modulated protein, GSTP1, into cells was sufficient to confer resistance to adaphostin. The proteomic “fingerprint” of adaphostin treatment was then shown to be almost identical to that derived through exposure to a simple hydroquinone or H₂O₂, implicating the redox biology of the dihydroquinone in adaphostin as central to the mechanism of action. This prompted further analysis of the role of antioxidant enzymes, leading to the identification of Cu/Zn superoxide dismutase (SOD I) as a further marker of resistance to adaphostin. Finally, a simple hydroquinone and H₂O₂ were also able to affect degradation of Bcr/abl polypeptide in a similar manner to adaphostin treatment, suggesting that tyrosine kinase inhibitory activity is an indirect event associated with increased oxidative stress.

Materials. Adaphostin (NSC 680410) was obtained from the Drug Synthesis and Chemistry Branch of the Developmental Therapeutics Program, National Cancer Institute (Rockville, MD). All cell lines were from the Division of Cancer Treatment and Diagnosis Tumor Repository (Frederick, MD). Materials were from the following: PBS and RPMI, Quality Biologicals; BCA Protein assay, Pierce; polyvinylidene difluoride membranes and all gels, Invitrogen; complete protease inhibitor tablets and WST reagent, Roche. Primary antibodies were from the following: SOD I/II, catalase, GSTP1, and myeloperoxidase, Abcam; c-Abl (ab3) used in the detection of p210^Bcr/abl, Calbiochem, β-actin, Sigma. Secondary horseradish peroxidase–conjugated antibodies were from Jackson Immunoresearch. The Cu/Zn SOD expression construct and control base vector pCMV6-XL5 were from Origene Technologies. Unless otherwise indicated in the following methods, all other chemicals and inhibitors were from Sigma.

Cytotoxicity and cell viability. Assays were conducted as follows: 10⁴ cells in 100 μL were placed into each well of a 96-well plate 24 h before treatment. Sample or buffer control (10 μL) were added to the appropriate wells and the plates were incubated at 37°C in a humidified CO₂ incubator for the times indicated in the figure legends. To determine protein synthesis, the serum-containing medium was replaced with serum- and leucine-free RPMI containing 0.1 mCi of [¹⁴C]leucine. Incubation continued for 2 to 3 h at 37°C. The cells were harvested onto glass fiber filters using a PHD cell harvester, washed with water, dried with methanol, and counted. The results are expressed as % [¹⁴C]leucine incorporation into the control-treated cells. To assay for cell viability, 10 μL WST reagent were added to each well and the plate was incubated for 2 to 4 h followed by reading absorbance of the formazan dye product at 450 nm, using a microplate reader (Bio-Tek Instruments). Experiments were done at least twice with triplicate determinations for each point. The IC₅₀ was defined as the concentration of adaphostin required to inhibit protein synthesis or cell viability by 50% relative to control-treated cells.

Apoptosis and necrosis determination. The percentage of apoptotic and necrotic cells in culture was determined using the Vybrant Apoptosis Assay kit (Molecular Probes) comprising an Annexin V–Alexa⁴⁸⁸ conjugate and propidium iodide as described by the manufacturer. Acquisition and analysis of data was done using a FACScan flow cytometer (Becton Dickinson) controlled by Cellquest Pro Software.

Cell cycle analysis. Treated cells were harvested and washed once with PBS. The samples were resuspended in 5 mL PBS and 5 mL cold 70% ethanol was added drop wise. After 5-min incubation, the cells were centrifuged, resuspended in 10 mL cold 70% ethanol, and stored at 4°C for 1 h. The cells were washed twice with 5 mL PBS and resuspended in 1 mL PBS containing 50 μg/mL propidium iodide (Molecular Probes) and 100 μg/mL RNase A (Sigma). After 1 h at 37°C, cell cycle analysis was done using the FL3-A channel on a FACScan flow cytometer.

Western blotting. Treated cells were washed twice in PBS and lysed in RIPA-CHAPS buffer [50 mmol/L Tris-HCl (pH 7.4), 1 mmol/L EDTA, 1% CHAPS, 1% deoxycholate, 1× complete protease inhibitor]. Lysates were sonicated, centrifuged to remove insoluble material, and protein concentration was determined using the BCA Protein assay according to the manufacturer's instructions. SDS-PAGE was done using 10 μg protein per well on a 10% NUPAGE Tris-glycine gel with subsequent transfer to a polyvinylidene difluoride membrane by electroblotting. Following overnight blocking in 4% milk/TBS, membranes were incubated with primary antibody for 2 h, washed several times in 4% milk/TBS, and incubated with the secondary, peroxidase-conjugated antisera for 2 h. Bands were visualized using enhanced chemiluminescence reagents ECL (Amersham, GE Healthcare) according to the manufacturer's protocol.

Protein carbonyl determination. HL60 cells (15 × 10⁶ in 30 mL) were treated over a time course with 5 μmol/L adaphostin or 100 μmol/L H₂O₂. Where applicable, cells were pretreated with 100 mmol/L N-acetyl-l-cysteine (l-NAC) for 30 min before the addition of adaphostin or H₂O₂. Following treatment, cells were washed with PBS, and the 0.5 mL of 0.2 mol/L NaPO₄ (pH 6.6), containing 1 mmol/L EDTA, and protease inhibitor (Roche) were added. The cells were sonicated twice for 10 s followed by centrifugation to remove any insoluble material. Protein concentration was determined using the BCA Protein assay according to the manufacturer's instructions. For carbonyl determination, the method of Levine et al. (16) was modified as follows: 0.2 mL 2 mg/mL protein was precipitated with 16% trichloroacetic acid (final concentration). Pellets were resuspended in 1 mL 2 mol/L HCl (control) or 1 mL 0.2% 2,4-dinitrophenylhydrazine in 2 mol/L HCl, incubated for 30 min in the dark at room temperature, and vortexed every 5 min. Samples were reprecipitated with 16% trichloroacetic acid (final concentration), incubated on ice for 5 min, and centrifuged for 5 min. Precipitates were washed thrice with ethanol/ethyl acetate (1:1, v/v) before being dissolved in 0.1 mL 6 mol/L guanidine HCl in 2 mol/L HCl. Insoluble debris was removed by centrifugation for 5 min. The difference spectrum between the 2,4-dinitrophenylhydrazine–treated sample and the 2 mol/L HCl control sample was determined at 366 nm using a molar extinction coefficient for 2,4-dinitrophenylhydrazine of 22,000 (mol/L)⁻¹ cm⁻¹. The results were expressed as nanomoles of 2,4-dinitrophenylhydrazine incorporated per milligram of protein. Polypeptide carbonylation was measured using the OxyBlot kit (Upstate). Cells were treated with varying concentrations of adaphostin or H₂O₂, harvested into RIPA-CHAPS buffer, sonicated, and the protein concentration was determined using the BCA Protein assay. Fifteen to 25 μg protein were labeled according to the manufacturer's instructions and run on a 10% NUPAGE Tris-glycine gel. Western blot analysis was carried out using α-DNP primary antibody and a horseradish peroxidase–conjugated goat α-rabbit secondary antibody.

Glutathione S-transferase activity assay. Glutathione S-transferase activity was measured by following conjugation of the thiol group of glutathione to 1-chloro-2,4-dinitrobenzene at 340 nm in a spectrophotometric assay. Assays were done according to the manufacturer's instructions (Glutathione S-Transferase Assay kit, Sigma). Cells were collected by centrifugation (suspension cells) or by scraping (adherent cells), sonicated for 10 s in cold 100 mmol/L potassium phosphate (pH 6.5) containing 2 mmol/L EDTA and protease inhibitor (Roche), and centrifuged to remove any insoluble material. Protein concentration was determined using the BCA Protein assay according to the manufacturer's instructions. Each assay contained 10 μg protein and activity was followed over a 6-min period. The linear portion of the assay (2-6 min) and an extinction coefficient for 1-chloro-2,4-dinitrobenzene at 340 nm of 9.6 (mmol/L)⁻¹ cm⁻¹ was used in the final calculations.

Detection of ROS. Cells were harvested, washed once in PBS, and resuspended at 1 × 10⁶/mL in serum-containing medium. The cells were then incubated for 1 h at 37°C in the presence of varying adaphostin concentrations. CM-H₂DCFDA (Molecular Probes), which reacts with peroxides to produce green fluorescence, was added to a final concentration of 10 μmol/L. After 1 h, cells were washed twice in PBS and analyzed using the FL1 channel on a FACScan cytometer.

SOD I and GSTP1 transfection. Cell transfections were done using Amaxa Nucleofector technology (Amaxa). In brief, cells were washed in PBS, 2 × 10⁶ cells were resuspended in 100 μL of transfection solution, 2 μg of plasmid DNA were added, the suspension was mixed and placed into an Amaxa cuvette and electroporated using the appropriate program.⁴

Two-dimensional PAGE. Cells (15 × 10⁶ in 30 mL) were treated for 6, 12, and 24 h with 5 μmol/L (HL60 cells) or 30 μmol/L (K562 cells) adaphostin. In addition, HL60 cells were also treated for 24 h with 50 μmol/L hydroquinone, 100 μmol/L H₂O₂, or adaphostin with 10 mmol/L l-NAC. Following treatment, the cells were washed twice with PBS. For protein determination, cells from 7 mL cultures were lysed in 250 μL RIPA-CHAPS buffer containing 1 mmol/L sodium orthovanadate and analyzed using the BCA Protein assay. For two-dimensional gel electrophoresis, cells from 21 mL cultures were lysed in 750 μL cell lysis/rehydration buffer (8 mol/L urea, 2% w/v CHAPS, 0.5% v/v Pharmalyte 3-10, 100 mmol/L dithioerythritol, 0.002% bromophenol blue, 1 μmol/L Tris; all except dithioerythritol were from Amersham Biosciences). Cell lysates were sonicated twice for 10 s and then centrifuged to remove any insoluble material. For first-dimension electrophoresis, 200 μg protein sample were adjusted to 450 μL with lysis/rehydration buffer and applied to 24 cm Immobiline DryStrips (pH 3-10 NonLinear, Amersham Biosciences). Rehydration and isoelectric focusing were done in the Ettan IPGphor apparatus (Amersham Biosciences) at 20°C, maximum of 80 μA per strip, according to the following program: 4 h at 0 V, 7 h at 30 V (rehydration); 1 h at 200 V, 1 h at 500 V, 1 h at 1000 V, then 8 to 12 h at 8,000 V (isoelectric focusing), until reaching 60 to 100 kV-h. After isoelectrofocusing, the gel strips (pH 3-10 NonLinear, Amersham Biosciences) were cut into three equal pieces, equilibrated twice for 20 min with gentle shaking in 5 mL each of a solution containing 50 mmol/L Tris-HCl (pH 6.8), 6 mol/L urea, 30% glycerol, 2% w/v SDS, and a trace of bromophenol blue. Two percent (w/v) dithioerythritol (Sigma) was added to the first equilibration step, whereas 2.5% w/v iodoacetamide (Sigma) was added to the second equilibration step. For the second dimension, the strips were placed on top of NuPAGE 4% to 12% Bis-Tris ZOOM mini gels, sealed with 0.5% agarose in NuPAGE MES SDS running buffer and run for 40 min at 200 V. For fluorescent staining with SYPRO Ruby, compatible with subsequent tryptic digestion, the gels were washed in H₂O, fixed in 7% acetic acid, 10% methanol for 30 min, and then stained overnight in SYPRO Ruby protein gel stain (Molecular Probes). To decrease background fluorescence, the gels were destained in 7% acetic acid, 10% methanol for 30 min before imaging. The gels were imaged with a Typhoon 9200 imager (Amersham Biosciences).

Gel cutting and in-gel tryptic digestion of proteins. Protein spots were excised from the SYPRO Ruby stained gels and transferred to a 96-well plate. Gel spots were washed twice with 100 mmol/L ammonium bicarbonate. The gel spots were dehydrated with acetonitrile and dried in a SpeedVac concentrator SC110A (Savant, Fisher Scientific). The dry gel spots were rehydrated with 50 mmol/L ammonium bicarbonate buffer containing 12.5 ng/μL sequencing grade porcine trypsin (Promega) for 45 min on ice. After reswelling of the gel spots, the remaining buffer was removed and replaced with 50 mmol/L ammonium bicarbonate. Digestion was carried out at 37°C for 16 h. The supernatant was removed and the tryptic peptides were extracted first with 25 mmol/L ammonium bicarbonate and thereafter with 5% formic acid for 20 min each. The combined extracts, as well as the supernatant after digestion, were dried in a SpeedVac concentrator and dissolved in 10 μL 1% formic acid, 5% acetonitrile before mass spectrometric analysis.

Mass spectrometry. The technique used for protein identification involved microcapillary liquid chromatography–tandem mass spectrometry. Trypsin-digested protein samples were analyzed using a Finnigan LTQ ion trap mass spectrometer by Protana, Inc. (Toronto, Canada) and LPAT (Science Applications International Corporation Frederick, under contract NO1-CO-12400). The resulting data files were analyzed using Mascot (Matrix Science Ltd.). The National Center for Biotechnology Information database was searched using the following variables: Variable modifications were carbamidomethylation of cysteine residues and oxidation of methionine residues and maximum of two missed cleavages.

Establishing conditions for proteomic analysis. Adaphostin treatment (structure in Fig. 1A) is associated with the inhibition of protein synthesis and the rapid induction of apoptosis (5, 6). Several assays were used to monitor the effects of adaphostin on HL60 (Bcr/abl negative) and K562 (Bcr/abl positive) cells, with the aim of elucidating the optimal conditions for proteome analysis. The first set of assays measured the effect of adaphostin on protein synthesis and cell viability (Table 1) over 72 h. In both assays, the maximal effect of adaphostin on HL60 and K562 cells was achieved within the first 24 h. The concentration of adaphostin required to inhibit protein synthesis for days 1, 2, and 3 by 50% was 0.35, 0.3, and 0.4 μmol/L for HL60 cells, and 3.5, 2.5, and 3.0 μmol/L for K562 cells, respectively. The IC₅₀ values determined for cell viability were 2.0, 1.5, and 1.5 μmol/L for HL60 cells and 10, 7, and 7 μmol/L for K562 cells for days 1, 2, and 3, respectively. In both assays, the Bcr/abl–negative cell line, HL60, was ∼5- to 10-fold more sensitive to adaphostin than the Bcr/abl–positive cell line, K562 (protein synthesis: IC₅₀ 0.35 and 3.5 μmol/L; cell viability: IC₅₀ 2.0 and 10 μmol/L, HL60 and K562, respectively). In the second set of assays, flow cytometry was used to monitor the induction of apoptosis using Annexin V/propidium iodide (Fig. 1B). A time course of the response of HL60 and K562 cells to 1.0 and 10 μmol/L adaphostin showed detectable increases in Annexin V–positive cells already after 4 h. By 24 h, the majority of cells were Annexin V/propidium iodide positive. In addition, cell cycle analysis of HL60 and K562 cells at 24 h showed an increase in the percentage of cells with subdiploid DNA content without phase-specific inhibition (14-23% and 15-30%, HL60 and K562 cells, respectively) after adaphostin treatment (Fig. 1C). These data support previous observations that Bcr/abl expression is not an absolute requirement for adaphostin activity (10, 11).

Proteomic analysis of adaphostin-treated cells. To characterize changes in the cellular proteome induced by adaphostin, two-dimensional PAGE was done on whole-cell lysates of control and treated HL60 and K562 cells. A mini two-dimensional gel format, in which the 24 cm isoelectric focusing nonlinear strip (pH 3-10) was divided into three 8-cm pieces and applied to three 4% to 12% Bis-Tris Zoom gels, was chosen for its high-throughput, simplicity in handling, excellent reproducibility, and the ability to perform subsequent Western blot analysis. Initial experiments using 0.1, 1, and 5 μmol/L adaphostin with exposure times of 6, 12, and 24 h showed that optimal responses, in terms of the number of modulated spots observed, were seen using 5 μmol/L adaphostin for HL60 cells. No changes in the proteomic profiles were observed for K562 cells at these concentrations. Increasing the adaphostin concentration to 30 μmol/L with exposure times of 12 and 24 h proved optimal for K562 cells. Several hundred spots could be observed in every composite of three mini gels after staining with SYPRO Ruby. A representative two-dimensional gel composite from an HL60 experiment is shown in Fig. 2. SYPRO Ruby–stained gels were inspected manually for modulated spots, which were only considered legitimate if they were seen in three separate experiments using separate lysate preparations. Only those spots that were either present upon treatment (i.e., up-regulated and absent in control gels) or absent upon treatment (i.e., down-regulated) were chosen for further analysis. Analysis of gel sets showed adaphostin-induced differential expression of 60 spots in HL60 cells and 59 spots in K562. The time course analysis of HL60 cells (data not shown) showed that by 6 h, the majority of changes had already occurred: 69% of down-regulated proteins and 86% of up-regulated proteins were already modulated, pointing to a rapid response to adaphostin treatment. By 12 h, these numbers increased to 92% and 100% down- and up-regulated spots, respectively. One hundred different protein spots exhibiting expression trends (mainly present versus absent) and several reference spots were then excised from SYPRO Ruby–stained gels, digested with trypsin, and subjected to nanospray microcapillary liquid chromatography–tandem mass spectrometry for protein identification. This work culminated in the identification of 79 differentially regulated protein spots. Condensing the list to remove multiple isoforms of the same protein generated a refined list of 49 unique proteins that were all modulated by adaphostin treatment. A comparison of differentially expressed proteins identified from HL60 and K562, subdivided into molecular function and showing the type of spot modulation, are shown in Table 2A. Analysis of spots (some reference and some differentially regulated) occupying the same position on the gels from HL60 and K562 cell lysates revealed identical proteins. A magnified view of several modulated spots with protein identification is shown in Fig. 3. Of the protein spots present in both cell lines, 78% of spots (39 of 50) were modulated in the same manner (23 up-regulated and 16 down-regulated proteins).

At the protein level, there were 19 differences between HL60 and K562, eight proteins (proliferating cell nuclear antigen, TBB2, PLEC1, VINC, T complex protein β subunit, ERP29, GSTP1, and PRS6A) were up-regulated and one protein (SNP29) was down-regulated in K562 cells only (and either not detected or not changed in HL60 cells), whereas 10 proteins (transitional endoplasmic reticulum ATPase, S10A7, GELS, LAM2, G3P2, KPYM, HNRPC, PCBP1, RLA1, and VDAC2) were down-regulated in HL60 cells (and either not detected or not changed in K562 cells). These results show that adaphostin treatment induces profound changes in the cellular proteome. Combined manual and PANTHER database⁵

Expression of GSTP1 confers resistance to adaphostin. Up-regulation of the antioxidant enzyme GSTP1 in K562 cells prompted further investigation into a potential role in resistance to adaphostin (Fig. 4). K562 cells were transfected with GSTP1 to investigate whether an increase in expression would modulate adaphostin activity (Fig. 4A). As expected from proteomic results, Western blotting of K562 cells revealed endogenous expression of GSTP1. After transfection, prevailing levels of GSTP1 increased considerably. This elevated expression translated into a significant increase in adaphostin resistance (IC₅₀, 4.5 to 10.2 μmol/L and 17 to >30 μmol/L) in the context of [¹⁴C]leucine (Fig. 4A) and WST viability assays (data not shown), respectively. Attempts at transfecting the same construct into HL60 cells resulted in unacceptable levels of cell death (>50%). However, transfection of IGROV1 cells with the GSTP1 construct confirmed increased adaphostin resistance (data not shown) using both [¹⁴C]leucine (IC₅₀ 5.0 to 15 μmol/L) and WST assays (IC₅₀ 15 to >30 μmol/L). Western blotting of a panel of cell lines using an anti-GSTP1 antibody was then done to determine whether a correlation exists between GSTP1 expression and adaphostin activity (Fig. 4B). Results failed to show any link between GSTP1 expression and activity. However, measurement of total glutathione S-transferase activity in the same panel of lines using an assay based around the glutathione S-transferase substrate chloro-2-4-dinitrobenene revealed a positive correlation between total activity and adaphostin resistance (compare Fig. 4C with D).

Adaphostin, 1,4-dihydroxybenzene, and H₂O₂ increase intracellular ROS and inhibit protein synthesis. Recent studies implicating ROS as central to the mechanism of adaphostin led to an investigation into the ability of the hydroquinone, 1,4-dihydroxybenzene, or H₂O₂ to replicate certain aspects of biological activity observed with adaphostin (9, 11). We investigated this hypothesis using two different methods. In the first set of experiments, HL60 and K562 cells were treated with varying concentrations of adaphostin, 1,4-dihydroxybenzene, or H₂O₂ for 24 h, and protein synthesis was measured (Fig. 5A). Results show marked similarity between treatments in that K562 cells were more resistant to each of the treatments than HL60 cells (IC₅₀ 0.8 versus 3, 20 versus 40, and 45 versus 80 μmol/L for adaphostin, 1,4-dihydroxybenzene, and H₂O₂ in HL60, and K562 cells, respectively). To further evaluate the role of ROS, the effect of adaphostin on protein synthesis in the presence of the antioxidant l-NAC was determined. l-NAC attenuated the effect of adaphostin on protein synthesis by 5-fold for HL60 and 2.2-fold for K562 cells (IC₅₀ 0.6 versus 3.0 μmol/L for HL60 cells, and 2.5 versus 5.5 μmol/L for K562 cells in the absence and presence of 12 mmol/L l-NAC, respectively). In the second set of experiments, HL60 and K562 cells were treated with adaphostin (10 μmol/L), hydroquinone (20 μmol/L), or H₂O₂ (200 μmol/L) for 1 h, followed by the addition of CM-H₂DCFDA (Fig. 5B). Reaction of CM-H₂DCFDA with peroxides generates the green fluorescent molecule 5-chloromethyl-2′,7′-dichlorofluorescein. Analysis by flow cytometry showed that in both cell lines, fluorescence increased significantly after incubation with all three reagents. These assays provide the first evidence of potential similarity between adaphostin, a simple hydroquinone, and a direct ROS-generating agent (H₂O₂).

Adaphostin increases protein carbonylation by a process prevented by l-NAC. Carbonylation, the addition of an oxygen atom to a polypeptide carbon backbone, is the most frequent protein modification arising through increased ROS. To examine the effects of adaphostin treatment on protein carbonyls, lysates derived from HL60 cells treated with adaphostin (5 μmol/L) for varying lengths of time were assayed by a standard 2,4-dinitrophenylhydrazine–coupled spectrophotometric method (Fig. 6A). Results showed time-dependent increases in protein carbonyl groups for adaphostin and the positive control (H₂O₂), with maximum carbonylation peaking at 8 h. Similarly, reacting adaphostin- or H₂O₂-exposed lysates with 2,4-dinitrophenylhydrazine and imaging carbonyl groups by Western blotting using an anti-DNP antibody revealed concentration-dependent increases in carbonyls for both treatments (Fig. 6B). In addition, patterns of protein carbonylation were identical in cell lysates prepared from adaphostin and H₂O₂-treated cells. Last, the ROS dependence of increased protein carbonylation was confirmed by demonstrating that preincubation with the antioxidant l-NAC reduced the amount of protein carbonyls detected (Fig. 6C).

The proteomic profile of adaphostin-treated cells is similar to that derived by treatment with hydroquinone or H₂O₂ and can be reversed by l-NAC. To further elaborate on the role of hydroquinone-derived ROS in adaphostin activity, proteomic profiles of HL60 cells treated with 1,4,dihyroxybenzene or the ROS-generating reagent, H₂O₂, were compared. In addition, the effects of the antioxidant l-NAC on the proteomic profile of adaphostin-treated cells were investigated. HL60 cells were incubated with either adaphostin (5 μmol/L), 1,4,dihyroxybenzene (50 μmol/L), or H₂O₂ (100 μmol/L) for 24 h and subjected to two-dimensional gel electrophoresis. Table 2B compares the protein changes induced by adaphostin, 1,4,dihyroxybenzene, or H₂O₂. For hydroquinone-treated cells 87% (69 of 79) of proteins showed identical changes to adaphostin treatment. Similarly, for H₂O₂-treated cells, 91% (72 of 79) of proteins showed identical changes between the sets. Table 2C shows the effect of the antioxidant l-NAC on adaphostin-induced changes. Here, changes observed in 68% (54 of 79) spots were prevented by 50% or more due to the addition of l-NAC. Only 14% of spots (11 of 79) were prevented <20% by l-NAC. Thus, similarities between proteomic profiles derived from adaphostin, 1,4,dihyroxybenzene, or H₂O₂ exposure, and the ability of l-NAC to prevent adaphostin-induced changes, supports a role for hydroquinone-derived ROS in adaphostin activity.

Cu/Zn SOD I confers resistance to adaphostin. Proteomic profiling of the response to drug treatment was followed by an investigation into a possible link between expression of antioxidant enzymes and adaphostin toxicity (Fig. 7). A panel of nine cell lines (two myelogenous leukemia, HL60 and K562; one T-cell leukemia, MOLT4; two ovarian, IGROV1 and SKOV3; two colon, COLO205 and HCC2998; and two renal, A498 and CAKI) were chosen based on their high or low susceptibility to adaphostin as defined in the panel of 60 human tumor cell lines of the Developmental Therapeutics Program of the National Cancer Institute. Cell lysates were prepared and probed with antibodies specific for catalase, myeloperoxidase, and SOD I or II (Fig. 7A). The relative expression of each protein was then compared with the IC₅₀ values determined from protein synthesis inhibition assays. Catalase expression was constant throughout the panel, whereas myeloperoxidase was only detected in HL60 cells. Expression of the SODs, however, had a close correlation with IC₅₀ values. For SOD II, high expression correlated with resistance to adaphostin in all cases except for the renal cancer cell pair, where CAKI cells expressed high levels of SOD II but were relatively sensitive to adaphostin. The most consistent correlation was found for SOD I, where high expression levels of SOD I correlated with resistance to adaphostin.

The association between SOD I expression and adaphostin activity was further explored using three different assays. In the first assay, diethyldithiocarbamate, an inhibitor of SOD I, was used to determine whether pretreatment could modulate adaphostin activity. In its capacity as a Cu²⁺ chelator, diethyldithiocarbamate is regarded as specific for SOD I (Cu/Zn) rather than manganese-containing SOD II (Fig. 7B). Cells were preincubated in diethyldithiocarbamate for 30 min followed by the addition of adaphostin at different concentrations. After 24 h, treated cells were analyzed for the change in protein synthesis (Fig. 7B). Results showed that pretreatment with the SOD I inhibitor diethyldithiocarbamate increased the sensitivity of cells toward adaphostin by 3-fold for HL60 and 30-fold for K562 cells (IC₅₀ 0.3 and 0.1 μmol/L for HL60 cells and 3 and 0.1 μmol/L for K562 cells in the absence and presence of diethyldithiocarbamate, respectively). The large change in IC₅₀ for K562 cells relative to HL60 may reflect the higher levels of endogenous SOD I in K562 cells compared with HL60 cells. This is further shown by the higher concentration of diethyldithiocarbamate required to achieve the maximal effect in K562 (18 μmol/L) versus HL60 (3 μmol/L) cells. Interestingly, the IC₅₀ values for both cell lines in the presence of the SOD I inhibitor were similar (0.1 μmol/L each). The second set of assays examined the effect of exogenous catalase and SOD I on the activity of adaphostin. As shown in Fig. 7C, the addition of either catalase or SOD I attenuated the effect of adaphostin. In the presence of 2 μmol/L adaphostin, protein synthesis was reduced to 13.9 ± 2.8% of control. Addition of exogenous catalase or SOD I decreased protein synthesis inhibition to 39.8 ± 4.6% and 22.5 ± 1.2%, respectively. Similar results were noted at 1 μmol/L adaphostin (51.7 ± 3.2% in the absence of any additions and 90 ± 2.9% and 82.8 ± 4.4% in the presence of added catalase or SOD I, respectively). In the third assay, a construct encoding SOD I was used to establish a causal relationship between SOD I levels and resistance to adaphostin (Fig. 7D). The ovarian carcinoma cell line IGROV1 was selected for transfection on the basis of low intrinsic SOD I expression. As shown in the inset to Fig. 7D, cells transiently transfected with the vector encoding SOD I had an increased expression of SOD I, relative to the vector control. To determine what effect elevated SOD I had on adaphostin activity, protein synthesis assays were done on transfected cells. Results showed that there was a 4.7-fold increase in IC₅₀ for cells transfected with SOD I, compared with the vector control (IC₅₀ 7.0 versus 1.5 μmol/L, respectively), suggesting that increasing levels of SOD I confer resistance to adaphostin. Similar results were noted for cell viability (IC₅₀ 15 versus 100 μmol/L, control vector and SOD1 vector, respectively, data not shown). These results highlight the role of antioxidant enzymes, in particular SOD I, in the cellular sensitivity to adaphostin.

Adaphostin, hydroquinone, and H₂O₂ affect polypeptide degradation of the tyrosine kinase p210^Bcr/abl. To determine whether induction of oxidative stress alone could explain Bcr/abl polypeptide degradation, K562 cells were treated with increasing doses of adaphostin, 1,4,dihyroxybenzene, or H₂O₂ for 24 h (Fig. 8A). Results showed that Bcr/abl was preferentially degraded relative to β-actin in K562 cells after exposure to adaphostin (>3 μmol/L), 1,4,dihyroxybenzene (>30 μmol/L), or H₂O₂ (>50 μmol/L). In addition, all treatments resulted in the appearance of c-Abl antibody–reactive high molecular weight aggregates. Second, the effect of antioxidants (l-NAC, TIRON, and ascorbate) on Bcr/Abl degradation induced by adaphostin, 1,4,dihyroxybenzene, and H₂O₂ was studied (Fig. 8B). Results showed that antioxidant pretreatment was incapable of restoring Bcr/abl expression.

In this study, proteomic analysis of myeloid leukemia cell lines was undertaken to provide insights into the biological activity of the putative anticancer agent, adaphostin (NSC 680410). Preliminary assays supported prior observations that adaphostin exposure was associated with inhibition of protein synthesis, oxidative stress, and induction of apoptosis (5, 6, 9, 11).

Analysis of HL60 and K562 cells using two-dimensional PAGE revealed altered expression of several proteins involved in several biological pathways, predominantly apoptosis and the response to oxidative stress. With respect to apoptosis, cleavage fragments from known caspase substrates (e.g., nucleoprotein TPR, filamin α, rho GDP-dissociation inhibitor 2, vimentin, and lamins A/C and B1) were identified in both cell lines (17–21). Likewise, down-regulation of the native form of far upstream element–binding protein 2 probably represents one facet of caspase cleavage (22). Further confirmation that apoptotic pathways are active in both lines came from detection of increased levels of programmed cell death protein 5 (PDCD5) and decreased levels of dUTPase. Expression of programmed cell death protein 5 is significantly elevated during initiation of apoptosis whereas dUTPase has been shown to be necessary for cell survival (23, 24). Identification of additional caspase cleavage products (prothymosin α, gelsolin, and lamin B2) only in HL60 cells supports the inference from the cell-based assays and time course proteomics that the response was far more advanced in these cells than in K562 cells (18, 25–27). This is further confirmed by the down-regulation of the transitional endoplasmic reticulum ATPase in HL60 cells. Transitional endoplasmic reticulum ATPase has been shown to be associated with an antiapoptotic function through activation of the nuclear factor-κB signaling pathway (28).

Several proteins implicated in the oxidative stress response were also modulated in both HL60 and K562 cells. One such example was the calcium-binding protein calmodulin. An extensive and complex interaction between ROS and Ca²⁺ has been identified (29). Under conditions of oxidative stress, calmodulin is selectively oxidized at critical methionine residues, allowing it to regulate the activity of several protein kinases and phosphatases (30). For example, oxidized calmodulin stabilizes target proteins such as plasma membrane Ca²⁺-ATPase in an inhibited state reducing ATP utilization, thereby regulating cellular metabolism (31). Up-regulation of calmodulin has also been observed in Jurkat cells undergoing FAS-induced apoptosis, emphasizing the intrinsic link between oxidative stress and apoptosis (32). Protein disulfide isomerase (PDIA1) is a chaperone found mainly in the endoplasmic reticulum that catalyzes thiol disulfide exchange (33). Increased oxidative stress is thought to modify PDIA1 at a critical cysteine residue involved in client protein binding, thereby impairing activity (34). In a proteomic project investigating the effects of oxidative stress (H₂O₂), PDIA1 was found to be a major target of proteasome-dependent oxidative degradation (35). In this project, a similar down-regulation of native PDIA1 was observed in both lines.

An important pattern of change restricted to HL60 cells concerned down-regulation of the glycolytic enzymes α-enolase, pyruvate kinase, and glyceraldehyde-3-phosphate dehydrogenase. Posttranslational modification of glycolytic enzymes is recognized as an intracellular sensor of oxidative stress (36). In a recent proteomic study, oxidizing agents, including H₂O₂, were shown to carbonylate and induce polypeptide degradation in glycolytic enzymes including α-enolase, pyruvate kinase, and glyceraldehyde-3-phosphate dehydrogenase (37). In a study of Escherichia coli, α-enolase was one of the primary targets of H₂O₂– or O₂⁻-mediated carbonylation (38). Therefore, if “sensors” of oxidative stress are being modulated, this suggests that higher prevailing levels of ROS are present in HL60 cells.

A number of proteomic adaptations exclusive to K562 provide a possible molecular basis for increased adaphostin resistance relative to HL60 cells. For example, increased expression of the T complex protein β subunit was observed in only K562 cells. This type II chaperone is indispensable for cell survival because folding of an essential set of cytosolic proteins (α/β actin, α/β tubulin, myosin heavy chain) requires T complex protein β subunit and this function cannot be done by other chaperones (39). Likewise, up-regulation of proliferating cell nuclear antigen in K562 cells may represent a protective mechanism. In addition to a role in DNA replication, PCNA is involved in nucleotide mismatch and base excision repair (40). Increased levels of PCNA may therefore enhance the ability of K562 cells to withstand cellular stresses that result in DNA damage. Similarly, the stress-inducible protein ERp29 was also up-regulated in K562 cells. Although the biological function of this chaperone remains unclear, studies using an ERp29-overexpressing rat thyroid cell line suggested that ERp29 assists in protein folding and secretion in association with other endoplasmic reticulum chaperones (GRP94, BiP, ERp72; ref. 41). The most compelling candidate for an adaphostin response modifier involved the detoxifying enzyme GSTP1, which showed elevated expression in K562 cells. This phase II metabolizing enzyme plays an important role in protection from oxidative stress. Transcription of GSTP1 is under the control of an antioxidant response element (42). Overexpression of GSTP1 has been shown to confer resistance to doxorubicin and chlorambucil, whereas clinically, increased levels of GSTP1 are a negative prognostic indicator (43). Here, the marked decrease in adaphostin activity observed for GSTP1-transfected cells provides strong evidence for a role in drug resistance. Likewise, the correlation observed between overall glutathione S-transferase activity and adaphostin activity hints at the potential of this family of enzymes in resistance to adaphostin.

Structurally, adaphostin consists of a dihydroquinone and an inert adamantyl group. Under alkaline conditions, hydroquinones are oxidized by molecular oxygen to form superoxide radicals and the respective quinone (44, 45). Several reports have suggested that hydroquinone redox activity may account for some of the biological observations attributed to adaphostin (9). Interestingly, in a study of the effects of benzene metabolites on HL60 cells, simple hydroquinones were confirmed as potent ROS generators (46). Hydroquinone treatment has also been shown to induce apoptosis in HL60 cells and to inhibit protein synthesis and the secretion of cytokines (e.g., interleukin-1β) in monocytes (47–49). Here, we confirm rapid increases in ROS after treatment with the hydroquinone 1,4-dihydroxybenzene, and show that the proteomic profile is almost identical to cells treated with adaphostin. A logical extension was therefore to compare adaphostin activity with a simple ROS generator, such as hydrogen peroxide (H₂O₂). HL60 cells were ∼2-fold more sensitive to H₂O₂ than K562 cells, mirroring the relative susceptibility of these cell lines to adaphostin. Proteomic analysis of H₂O₂-treated HL60 cells revealed a profile almost synonymous with that found after adaphostin exposure. These results support the concept that direct generation of ROS by the dihydroquinone in adaphostin is the predominant effector mechanism. However, the importance of the adamantyl group in adaphostin should not be overlooked. This inert nonpolar group has been shown to enhance the membrane penetration of several biological molecules, and this may account for the increased activity of adaphostin relative to 1,4-dihydroxybenzene (50).

Given that hydroquinone oxidation is associated with generation of superoxide ions, we speculated that antioxidant enzymes and in particular, SODs, may act as modifiers of adaphostin activity. In this regard, a limited survey of antioxidant enzymes in a diverse panel of cell lines was conducted with the result that expression of Cu/Zn SOD I correlated with resistance to adaphostin. This observation was strengthened by the finding that cells transfected with SOD I showed resistance, whereas preincubation with a SOD I inhibitor potentiated toxicity. It is noteworthy that anthracyclins such as doxorubicin undergo redox conversions to produce superoxide ions and alkylating semiquinone radicals (51). In a study of gastric cancer cell lines, resistance to doxorubicin was also conferred by SOD (52).

Degradation of Bcr/abl polypeptide in adaphostin-treated chronic myelogenous leukemia cells is a seminal experiment used to imply tyrosine kinase inhibitor activity (5, 6). Here, we show that a hydroquinone or direct oxidative stress (H₂O₂) affects degradation of Bcr/abl polypeptide in an identical manner to adaphostin treatment. It has been suggested that because Bcr/abl degradation cannot be reversed with antioxidants, the process is distinct from oxidative stress induction (9). Interestingly, we report an identical result but show that Bcr/abl degradation promulgated by hydroquinone or H₂O₂ was also unaffected by treatment with antioxidants. These results imply that adaphostin-induced Bcr/abl degradation may yet be related to increased levels of oxidative stress but the effect is independent of antioxidants.

Therefore, a wealth of evidence supports the hypothesis that the biological activity of adaphostin is reliant on the redox properties of the dihydroquinone group. Further evidence can be found in recent adaphostin publications. Most importantly, structure activity studies showed that the hydroquinone moiety in adaphostin was essential for activity (4). Also, in a study of glioblastoma cell lines, a positive correlation was found between sensitivities to both adaphostin and ROS (53). This study also noted that catalase expression was not altered with adaphostin treatment. Several other reports have elaborated on the role of ROS in adaphostin activity, while speculating on the existence of a molecular target (9, 11, 13, 54). The data presented here, showing hydroquinone- and H₂O₂-mediated Bcr/abl degradation, identical proteomic profiles, and the identification of GSTP1/SODI as resistance markers, undermine any suggestion that adaphostin operates by direct interaction with Bcr/abl. However, a primary molecular target for adaphostin-derived ROS may yet exist and one plausible candidate is the ubiquitous sarcoplasmic reticulum Ca²⁺ ATPase. Evidence to support a role for sarcoplasmic reticulum Ca²⁺ ATPase comes from studies of 2-5-di-(t-butyl)-1,4-hydroquinone, a compound with a very similar chemical structure to adaphostin. This simple hydroquinone has been shown to mobilize Ca²⁺ specifically from inositol-sensitive stores by inhibiting sarcoplasmic reticulum Ca²⁺ ATPase in a process mediated by superoxide ions (55).

This hydroquinone-centric mechanism provides possible explanations for several reported observations. For example, there is direct evidence to show that hydroquinone and the semiqiunone intermediates of several anticancer agents promote the release of iron from ferritin in reactions inhibited by SOD (56, 57). An identical mechanism may account for the reported increases in free ferrous iron and ferritin mRNA after adaphostin treatment (12). Likewise, the increases in protein carbonylation observed after adaphostin treatment would place increasing strain on protein catabolism, providing a basis for previous reports of synergy with proteasome inhibitors (54). Also, if adaphostin activity were divorced from kinase inhibition, it would provide a basis for the ability to “overcome” resistance to ATP-dependent kinase inhibitors (15, 58). From a molecular perspective, adaphostin-associated peturbations in signal transduction pathways (p38 mitogen-activated protein kinase, RAF-1/mitogen-activated protein kinase kinase/extracellular signal-regulated kinase, and AKT) cannot be extricated from changes that would also occur with increased oxidative stress (13, 14). Similarly, the selectivity of adaphostin for hemopoietic malignancies such as acute myelogenous leukemia, chronic myelogenous leukemia, and chronic lymphocytic leukemia can be explained given that these diseases are exquisitely sensitive to other ROS-generating reagents, such as arsenic compounds or sesquiterpene lactones (9, 15, 59, 60). In conclusion, evidence from proteomic and biochemical studies supports classification of adaphostin as a redox-active substituted hydroquinone.