auranoﬁn induces lethal oxidative and endoplasmic ... · transduced into mec-1 cells. forty-eight...

Therapeutics, Targets, and Chemical Biology

Auranofin Induces Lethal Oxidative and EndoplasmicReticulum Stress and Exerts Potent Preclinical Activityagainst Chronic Lymphocytic Leukemia

Warren Fiskus1, Nakhle Saba3, Min Shen4, Mondana Ghias3, Jinyun Liu2, Soumyasri Das Gupta5,Lata Chauhan5, Rekha Rao5, Sumedha Gunewardena5, Kevin Schorno6, Christopher P. Austin4,Kami Maddocks7, John Byrd7, Ari Melnick8, Peng Huang2, Adrian Wiestner3, and Kapil N. Bhalla1

AbstractChronic lymphocytic leukemia (CLL) exhibits high remission rates after initial chemoimmunotherapy, butwith

relapses with treatment, refractory disease is the most common outcome, especially in CLL with the deletion ofchromosome 11q or 17p. In addressing the need of treatments for relapsed disease, we report the identification ofan existing U.S. Food and Drug Administration-approved small-molecule drug to repurpose for CLL treatment.Auranofin (Ridaura) is approved for use in treating rheumatoid arthritis, but it exhibited preclinical efficacy inCLL cells. By inhibiting thioredoxin reductase activity and increasing intracellular reactive oxygen species levels,auranofin induced a lethal endoplasmic reticulum stress response in cultured and primary CLL cells. In addition,auranofin displayed synergistic lethality with heme oxygenase-1 and glutamate-cysteine ligase inhibitors againstCLL cells. Auranofin overcame apoptosis resistance mediated by protective stromal cells, and it also killedprimary CLL cells with deletion of chromosome 11q or 17p. In TCL-1 transgenic mice, an in vivo model of CLL,auranofin treatment markedly reduced tumor cell burden and improved mouse survival. Our results provide arationale to reposition the approved drug auranofin for clinical evaluation in the therapy of CLL.Cancer Res; 74(9);1–13. �2014 AACR.

IntroductionAccelerated expansion of chronic lymphocytic leukemia

(CLL) cells with bulky lymphadenopathy and organomegaly,with or without compromised hematopoiesis, is treated withmyelotoxic chemoimmunotherapy (1, 2). In CLL, the unmu-tated immunoglobulin heavy chain variable region genes(IGHV), acquired chromosomal abnormalities including dele-tion 17p13 and deletion 11q22, as well as increased expressionof ZAP70 (zeta-associated protein) or CD38 are features asso-ciated with poor outcome (3). Notwithstanding high remission

rates due to initial chemoimmunotherapy, eventual relapsewith treatment-refractory disease is the typical outcome,except in a minority of patients who successfully receiveallogeneic stem cell transplantation (2, 3). Therefore, noveleffective and safe treatments need to be tested and developed.To this end, repurposing of an existing and U.S. Food and DrugAdministration (FDA)-approved small-molecule drug in thetreatment of CLL is a worthy goal (4).

Compared with normal lymphocytes, CLL cells have intrin-sically higher levels of reactive oxygen species (ROS) and areunder oxidative stress due to an imbalanced redox status (5–8).ROS-mediated oxidation of the sulfur-containing amino acidsin proteins such as phosphatases and transcription factors, forexample, NF-kB, p53, hypoxia-inducible factor-1a, and nuclearfactor erythroid 2-related factor 2 (Nrf2), regulates their func-tion and role in modifying cellular growth and survival (9).Elevated ROS levels also render CLL cells more sensitive toagents that further increase ROS and oxidative stress (6). Nrf2activates genes involved in the response to oxidative stress,including heme oxygenase-1 (HMOX-1) and glutamate cyste-ine ligase modifier (GCLM), which are involved in glutathione(GSH) synthesis (10, 11). Elevated levels of ROS may overcomeantioxidant mechanisms and induce protein oxidation, whichleads to intracellular accumulation of potentially toxic, mis-folded, and polyubiquitylated (poly-Ub) proteins (12). Thisaccumulation triggers an HDAC6-mediated, adaptive and pro-tective heat shock and proteotoxic stress response (13, 14).During this, HDAC6 binds to the poly-Ub–misfolded proteins

Authors' Affiliations: 1Houston Methodist Research Institute; 2The Uni-versity of Texas M.D. Anderson Cancer Center, Houston, Texas; 3Hema-tology Branch, National Heart Lung and Blood Institute (NHLBI); 4NationalCenter for Advancing Translational Sciences, NIH, Bethesda, Maryland;5The University of KansasMedical Center; 6Institute for AdvancingMedicalInnovation, Kansas University, Kansas City, Kansas; 7Division of Hema-tology, Department of Internal Medicine, The Comprehensive CancerCenter at the Ohio State University, Columbus, Ohio; and 8Weill CornellMedical College, New York, New York

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

W. Fiskus and N. Saba contributed equally to this work.

Corresponding Author: Kapil N. Bhalla, Houston Methodist ResearchInstitute, 6670 Bertner Avenue, R9-113, Houston, TX 77030. Phone:713-441-9113; Fax: 713-441-5349; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-13-2033

�2014 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org OF1

Research. on June 19, 2020. © 2014 American Association for Cancercancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 5, 2014; DOI: 10.1158/0008-5472.CAN-13-2033

http://cancerres.aacrjournals.org/

and shuttles these into a protective aggresome, concomitantlycausing the dissipation of the p97/HDAC6/hsp90/HSF1 (heatshock factor 1) complex, followed by induction of transcrip-tional activity of HSF1 and HSPs (15, 16). The dissociation ofHDAC6 from this complex also causes hyperacetylation andinhibition of the chaperone function of hsp90 (17), withresulting depletion of CLL-relevant, progrowth and prosurvivalhsp90 client proteins such as ZAP70, c-RAF, AKT, as well as ofHDAC6 itself (18–21). Thus, ROS-induced oxidative stress canlead to proteotoxic and unfolded protein response (UPR),which in turn also triggers estrogen receptor (ER) stress, withactivation of the mediators of the ER stress response (22–24).Normally, ER stress is designed to be protective by mediatingthe shutdown of general protein synthesis and by increasingthe production of molecular chaperones, including the ERresident hsp70 homologue, glucose-regulated protein 78(GRP78; refs. 22, 23). However, if ER stress is protracted, lethalER stress ensues through prolonged activation of the pro-deathER stress pathways mediated by CHOP (CAAT/enhancer-binding protein homologous protein) and IRE1 (inositolrequiring protein 1; refs. 23–25). Countering this, CLL cellsreceive numerous prosurvival signals from the stroma micro-environment in the bone marrow and lymph nodes throughmultiple mechanisms that activate B-cell receptor and thechemokine receptor CXCR4 signaling (26–29). Recently, stro-mal cells were also shown to protect CLL cells againstincreased intracellular levels of ROS, by providing cysteineand bolstering the intracellular levels of GSH in CLL cells (30).

Auranofin, an oral gold-containing triethylphosphine usedin the treatment of rheumatoid arthritis, has been previouslyreported to inhibit cytosolic and mitochondrial thioredoxinreductase (TrxR) and induce ROS levels (31). On the basis of thepreliminary results of an in vitro high-throughput screen togauge the activity against primary CLL cells, and toward theultimate goal of repurposing auranofin for the treatment ofCLL, we determined the in vitro and in vivo activity of aur-anofin, and its mechanism of action, against CLL cells. Ourfindings demonstrate for the first time that, auranofin induceslethal oxidative, proteotoxic, and ER stress response in cul-tured and patient-derived primary CLL cells, including thosewith the biologic and genetic features that are associated withpoor clinical outcome in CLL.

Materials and MethodsReagents and antibodies

Auranofin (C20H35AuO9PS) was suspended in dimethyl sulf-oxide at 5 mmol/L concentration and stored in volumes of 500mL at�80�C, then aliquoted in volumes of 10 mL and stored at�20�C to be used once without refreezing. Additional reagentsand detailed antibody information are provided in the Sup-plementary Materials and Methods. Polyclonal acetylated K69hsp90 antibody was generated by Alpha Diagnostics as previ-ously described (17).

Cell lines and primary CLL cellsHuman chronic B-cell leukemia, MEC-1 cells were obtained

from the DSMZ). All experiments with cell lines were conducted

within 6 months after thawing or obtaining from DSMZ. Cellline authentication was done by DSMZ. The DSMZ uses shorttandem repeat profiling for characterization and authenticationof cell lines. Cells were cultured in Iscove's Modified Dulbecco'sMedium with 10% heat-inactivated FBS and 1% penicillin/streptomycin. Cells were passaged two to three times per weekand frozen in aliquots in liquid nitrogen. HK cells were kindlyprovided and characterized by Jianguo Tao (Moffitt CancerCenter, Tampa, FL), cultured in RPMI with 10% FBS, and main-tained as previously described (32). Exponentially growingcells were used for all described experiments. Primary CLL cells(Supplementary Table S1) were obtained with informed con-sent (in accordance with the Declaration of Helsinki) under aresearch protocol approved by the Institutional Review Board ofKansas University Medical Center (Kansas City, KS; Protocol#12392) or the NIH (7, 30). Primary CLL cells were isolated fromthe peripheral blood and CD19þ B cells were purified, utilizing apositive selection immunomagnetic separation kit (Stem CellTechnologies). Normal human CD34þ cells were obtained fromdelinked, deidentified human cord blood samples.

Assessment of mitochondrial transmembrane potential(Dcm) and reactive oxygen species

CLL cells (0.5 � 106 cells/tube) were incubated with 10nmol/L 3,3-dihexyloxacarbocyanine iodide (DiOC6), 2.5mmol/L dihydroethidium, and 10 mL of CD19-APC, in 400 mLAIM-V in a humidified atmosphere of 5% CO2 at 37�C for 30minutes, then analyzed by flow cytometry.

Thioredoxin reductase assayCLL cells were treatedwith auranofin for 8 hours. The effects

of auranofin treatment on TrxR activity were determined witha Thioredoxin Reductase Assay Kit (Cayman Chemicals)according to the manufacturer's protocol.

Evaluation of cell viabilityPrimary CLL peripheral blood mononuclear cells (PBMC;

0.5 � 106 cells per well) were plated in triplicate in a finalvolume of 100 mL, in the presence of serial dilutions ofauranofin (0.125–4.0 mmol/L) for 24 hours. Cell viability wasdetermined using a CellTiter 96 AQueous One Solution CellProliferation Assay (MTS) Kit (Promega). The percentage ofviable cells following auranofin treatment is reported relativeto the untreated control CLL cells.

Viability analysis in nurse-like cell coculturesPrimary CLL cells were cocultured in the presence or

absence of nurse-like cells (NLC; ref. 33), with or withoutauranofin, for 24 hours. Cell viability was determined by flowcytometry using DiOC6 and CD19 stain.

Assessment of apoptosisUntreated or drug-treated MEC-1 or primary CD19þ CLL

cells were stained with Annexin-V, and the percentage ofAnnexin V–positive apoptotic cells were determined by flowcytometry, as previously described (25). To analyze synergismbetween auranofin and zinc deuterophyrin IX 2,4-bis-ethyleneglycol (ZnBG) or buthionine sulfoximine (BSO), cells were

Fiskus et al.

Cancer Res; 74(9) May 1, 2014 Cancer ResearchOF2




treated with auranofin (200–1,000 nmol/L) and ZnBG (10–20mmol/L) or BSO (1–10 mmol/L) for 48 hours. The combinationindex (CI) for each drug combination was calculated utilizingCalcusyn (Biosoft). CI values of less than 1.0 indicate a syner-gistic interaction of the two agents.

Confocal microscopyPrimary CD19þ CLL cells were incubated with indicated

doses of auranofin for 16 hours and cytospun onto glassslides. Cells were fixed and permeabilized as previouslydescribed (25). Then, cells were stained with Nrf-2 andHSF-1 antibodies, counterstained with phalloidin (whichbinds F-actin) and 40, 6-diamidino-2-phenylindole (DAPI),and coverslips were mounted. The images were visualizedusing a Carl Zeiss LSM-5 PASCAL confocal microscope witha 63�/1.2 NA oil objective. Images were processed with LSMImage browser software (Zeiss).

RNA isolation and quantitative PCR analysesTotal RNA was extracted from untreated and auranofin-

treated MEC-1 or primary CD19þ CLL cells using an RNAqu-eous 4PCR Kit from Ambion and reverse transcribed. Quan-titative real-time PCR analysis and TaqMan probes were usedto determine the expression levels of GRP78, CHOP, HMOX-1,and GCLM. The relative expression of each mRNA was nor-malized to glyceraldehyde-3-phosphate dehydrogenase.

Gene expression analysisTotal RNA from primary CLL PBMCs treated with aur-

anofin for 4, 8, or 10 hours or left untreated for 10 hours waspurified using the RNeasy Mini Kit (Qiagen), and convertedto biotin-labeled cRNA. This was hybridized to a HumanGenome U133 Plus 2.0 array (Affymetrix). Signal intensityand fold changes in gene expression were analyzed by themicroarray analysis, using Affymetrix GeneChip OperatingSoftware, as detailed in the Supplementary Materials andMethods. All microarray data used in this manuscript aredeposited in the Gene Expression Omnibus.

Short hairpin RNALentiviral short hairpin RNAs (shRNA) targeting HMOX1

and GADD153/CHOP or nontargeting shRNA (sh-NT) weretransduced into MEC-1 cells. Forty-eight hours posttransduc-tion, the cells were washed with complete media andplated with or without auranofin for 8 hours for immunoblotanalysis or 48 hours for assessing apoptosis. MEC-1 cells withstable knockdown of CHOP were obtained by culturing thetransduced cells in 0.5 to 1.0 mg/mL of puromycin. Stableknockdown cells were treated for 48 hours with auranofin todetermine the effects of CHOP knockdown on auranofin-induced apoptosis.

Immunoprecipitation of hsp90 from CLL cellsPrimary CD19þ CLL cells were treated with auranofin for 8

hours. Following treatment, cells were harvested, lysed, andhsp90 was immunoprecipitated from total cell lysates aspreviously described (25). HDAC6, Hsf-1, and p97/VCP immu-noblot analyses were performed on the hsp90 immunopreci-

pitates. Blots were stripped and reprobed for detection ofhsp90.

Immunoblot analysesImmunoblot analyses were conducted as described in the

Supplementary Materials and Methods.

In vivo animal studiesAll animal procedures were approved by the Institutional

Animal Care and Use Committee at University of Texas M.D.Anderson Cancer Center (Houston, TX). TCL-1 transgenicmice genotype: (Tcl1-tg:p53�/�) have been previouslydescribed (34, 35). TCL-1 mice (n ¼ 5; age 2 months) weretreated with auranofin by intraperitoneal injection at thedosage of 10 mg/kg daily, 5 days per week (Monday–Friday)for 2weeks, and treatmentwas stopped. The total leukemia cellburden was measured in peritoneal fluid from pretreated andauranofin-treated animals, as previously described (30). Sur-vival of TCL-1 mice treated with auranofin compared withuntreated mice is represented by a Kaplan–Meier plot.

Statistical analysisSignificant differences between values obtained in a pop-

ulation of MEC-1 or primary CD19þ CLL cells treated withdifferent experimental conditions were determined using atwo-tailed, paired t test or a one way ANOVA analysis withinMicrosoft Excel 2010 software or using GraphPad Prism 5.0(GraphPad Software, Inc.). P values < 0.05 were assignedsignificance. For the in vivo studies, differences in the meanleukemia burden between pretreatment and post-auranofintreatment were determined by a one-way ANOVA test.Pvalues of <0.05 were assigned significance. Differences inthe survival of TCL-1 mice treated with auranofin werecalculated by a Mantel–Cox log-rank test. P values < 0.05were assigned significance.

ResultsAuranofin causes loss of viability of CD19þ B-CLL cells

Our findings in Fig. 1A demonstrate that exposure to aurano-fin for 24 hours dose dependently induced loss of viability ofprimary CLL cells. The clinical and prognostic molecular-cyto-genetic features of each of the 50 primary CLL samples aredescribed in Supplementary Table S1. The table also providesthe IC50 values of auranofin against each of the primary CLLsamples. Treatment with auranofin also significantly inducedapoptosis of primary CLL, but not of normal CD34þ hemato-poietic progenitor cells (Fig. 1B). Auranofin treatment increasedthe percentage of Annexin V–positive, CD19þ primary CLL cells,with relative sparing of the normal CD3þ T lymphocytes (Fig.1C–E). As shown in Fig. 2A, treatment with auranofin wasequally active against CLL with unmutated as compared withthe mutated IGHV genes. In contrast, treatment with auranofininduced significantly greater loss of cell viability of ZAP70þ

versus ZAP70� CLL cells (Fig. 2B). This was especially sofollowing exposure to the lower versus thehigher concentrationsof auranofin (Fig. 2C). Exposure to auranofin dose dependentlyinduced apoptosis in the cultured CLLMEC-1 cells, which carry

Auranofin Induces Oxidative/ER Stress in CLL Cells

www.aacrjournals.org Cancer Res; 74(9) May 1, 2014 OF3




the del17p chromosome. Thiswas associatedwith an increase inthe percentage of Sub-G1 cells and cleavage of caspase-3 andPARP (Fig. 2D and E and Supplementary Fig. S2). At lowerconcentrations, there was a significant difference in auranofin-induced apoptosis in the primary CLL cells containing del17pcompared with those containing del13q or del11q (P < 0.05; Fig.2D). On the other hand, at higher concentrations, there was nosignificant difference in auranofin-induced apoptosis in theprimary CLL cells containing del17p, del13q, or del11q (Fig.2D). Overall, auranofin treatment was as effective in inducingapoptosis of the favorable as compared with CLL cells with theunfavorable cytogenetic predictors of clinical outcome (Fig. 2F).

Auranofin overcomes protection due to stroma-associated cells and exerts lethal in vivo activity againstCLL cells

Wenext determined the in vitro effect of stroma, representedin our studies by coculturewith theHKcells or theNLCs, on the

lethal effects of auranofin onCLL cells (32, 33). As shown in Fig.2G, auranofin treatment dose dependently induced loss of cellviability of MEC-1 cells, which was significantly attenuated bycoculture with HK cells. Coculture with NLCs also improvedthe viability of primary CLL samples in vitro (Fig. 2H). However,our findings also show that auranofin treatment induced lossof viability of primary CLL cells, regardless of the coculture invitrowithNLCs (Fig. 2H). In contrast, treatmentwith auranofindid not reduce the viability of NLCs (Supplementary Fig. S3A–S3C). We next tested the in vivo activity of auranofin againstCLL cells in the TCL-1 transgenic mice (34, 35). Because theTCL-1 gene is among the top genes upregulated in CLL cells bycoculture with the stromal cells, this mouse model is alsoparticularly relevant to ascertain the in vivo protective role ofthe stromal microenvironment for CLL cells (36). As shownin Fig. 2I and Supplementary Table S2, treatment with aur-anofin caused a reduction in the leukemia cell burden in eachof the 5 TCL-1 mice tested. The number of CLL cells in the

A B

1.0

µm

ol/L

AF

, 24

hC

on

tro

l 1.0 µmol/L AF, 24 h

C

0

10

20

30

40

50

60

70

80

0 1.0

% A

po

pto

sis

µmol/L, AF, 48 h

CD34+ cord blood, n = 3Primary CLL, n = 5

D

CLL 26

CLL 31

CLL 32

CLL 33

CLL 34

CLL 35

CLL 36

CLL 370

20

40

60

80

100 CD3CD19

% R

edu

ctio

n in

via

bili

ty

CD3+ CD19+

1.0 µmol/L AF, 24 h

P = 0.0002E

0

20

40

60

80

100

% R

edu

ctio

n in

via

bili

ty

n = 50

µmol/L, AF, 24 h

0

50

100

150

�

�

��

Mea

n v

iab

ility

(%

) �, P < 0.0001Figure 1. Treatment with auranofin(AF) induces apoptosis of primaryCLL CD19þ but not CD3þ cells ornormal CD19þ or CD34þ cells. A,cell viability of primary CLL (n¼ 50)following treatment with auranofinfor 24 hours. Columns, mean lossof viability; bars, SEM. �, cellviability values significantly less inauranofin-treated cells thanuntreated cells (P < 0.05). B,percentage of apoptotic normalCD34þ cord blood cells (n ¼ 3)and primary CLL cells (n ¼ 5)followingexposure to 1.0mmol/Lofauranofin for 48 hours. Columns,mean apoptosis from the CD34þ

cord blood cells or primary CLLcells; bars, SEM. C, schematic ofthe flow cytometry gating methodused to determine the percentageof apoptotic CD3þ or CD19þ

cells following treatment with1.0 mmol/L of auranofin for24 hours. D, individual reduction inviability (relative to untreatedcontrol cells) of CD3þ and CD19þ

cells from patients with CLLtreated with 1.0 mmol/L ofauranofin for 24 hours. E, the %reduction in viability induced by24-hour treatment with 1 mmol/L ofauranofin inCD3þ andCD19þ cellsfrom patients with CLL (n¼ 8). Thehorizontal black bar indicates themedian reduction in cell viability �SE. �, significantly greaterreduction in viability of auranofin-treated CD19þ cells comparedwith CD3þ cells (P < 0.05).

Fiskus et al.





peritoneal cavity was reduced by >90% during the 2 weeks ofauranofin treatment (P ¼ 0.00019). These findings demon-strate that auranofin is able to exert in vitro and in vivo lethaleffects against CLL cells despite the protective effects ofstroma. Treatment with auranofin also significantly improvedthe survival of the TCL-1 mice compared with mice thatreceived no treatment (P ¼ 0.001; Fig. 2J).

Auranofin inhibits TrxR and induces oxidative stress andapoptosis in CLL cells

We next determined whether exposure to auranofin inhibitsTrxR activity, as well as induces ROS levels and apoptosis of theMEC-1 and primary CLL cells. Treatment with auranofininhibited the total intracellular TrxR activity determined bya TrxR activity assay (Cayman Chemical Company), and

B

D

0

20

40

60

80

100

120

0 0.25 0.5 1.0

% A

po

pto

sis

µmol/L, AF, 48 h

MEC-111q deleted (n=4)13q deleted (n=10)17p deleted (n=10)

***

ZAP70–

IC50

µm

ol/L

ZAP70+0.0

0.5

1.0

1.5

2.0

2.5 P = 0.048

µmol/L, AF, 24 h

C

0

20

40

60

80

100 ZAP70+

ZAP70–

� �

Viab

ility

(%)

� P < 0.05

0102030405060708090

100

0 100 250 500

% A

po

pto

sis

nmol/L, AF, 48 h

MEC-1

MEC-1/HK cells

*

*

G

I

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Control Auranofin

Leu

kem

ia b

urd

en(×

106

cells

)

P = 0.00019

A

0.0

0.5

1.0

1.5

2.0

2.5

IGVH-U

IC50

µm

ol/L

P = 0.2751

IGVH-M

F

CytogeneticsADV FAV

IC50

µm

ol/L

0.0

0.5

1.0

1.5

2.0

2.5 P = 0.43

H

AF – + – +0

20

40

60

80

100 6308642664516464

No NLC NLC

Via

bili

ty (

%)

P = 0.049

E

β-Actin

Cl PARP

CleavedCaspase-3

0

10

20

30

40

50

60

Control 1.0 µmol/L AF

% o

f C

ells

(24

h)

MEC1Annexin V+

Sub-G1

J

P = 0.001

2-week auranofindaily treatment

100

TCL1-Tg:p53–/–

Per

cen

tag

e su

rviv

al

50

00 2 4 6 8

Months10

Figure 2. Treatment with auranofin (AF) selectively induces apoptosis of cultured and primary CLL CD19þ cells independent of cytogenetics or IGHVmutation status, despite coculture with NLCs and reduced in vivo leukemia burden. A, IC50 values for the primary CLL patient cells treated with auranofinfor 24 hours subgrouped by IGHV mutational (U, unmutated; M, mutated) status. Black line, median IC50 value. B, IC50 values for primary CLL patientcells treated with auranofin for 24 hours subgrouped by ZAP70 status. Black line, median IC50 value. C, cell viability of primary CLL cells from patientssubgrouped by ZAP70 status (ZAP70þ, n¼ 12; ZAP70�, n¼ 15) following treatment with the indicated concentrations of auranofin for 24 hours. Each point onthe line represents the mean viability of the CLL cells at each concentration of auranofin; bars, SEM. �, cell viability values significantly less in ZAP70þ

auranofin-treated cells than ZAP70� auranofin-treated cells (P < 0.05). D, percent apoptosis of MEC-1 cells and CD19þ primary CLL cells (11qdeleted, 13q deleted, and 17p deleted) exposed to the indicated doses of auranofin for 48 hours. �, apoptosis values significantly greater in 17p-deleted CLLthan in 11q- and 13q-deleted CLL cells at 0.25 mmol/L of auranofin (P < 0.05). ��, apoptosis values significantly greater in 17p-deleted CLL than in 11q-and 13q-deleted CLL cells at 0.5 mmol/L of auranofin (P < 0.05). E, the percentage of sub-G1 cells and percentage of Annexin V–positive, apoptoticMEC-1 cells following 24-hour exposure to auranofin. Immunoblot analyses were also conducted as indicated. F, IC50 values for the primary CLL patient cellstreated with auranofin for 24 hours, subgrouped by adverse (ADV) or favorable (FAV) cytogenetics. Black line, median IC50 value. G, percent apoptosis ofMEC-1 cells with and without coculture on HK cells following treatment with auranofin for 48 hours. �, values significantly less in MEC-1 cellscocultured with HK cells compared with MEC-1 without coculture (P < 0.05). H, primary CLL cells (n ¼ 4) were cultured with or without NLCs and exposedto two times the IC50 of auranofin for 24 hours. At the end of treatment, the percentage of nonviable cells was determined by flow cytometry. Theabsolute viability of the CLL cells in each condition is shown. I, treatment of TCL-1 mice (n ¼ 5) with auranofin (10 mg/kg, 5 days per week for 2 weeks)significantly reduced the leukemia cell burden. Columns represent the mean � SEM of the leukemia cell burden in all 5 mice tested. J, Kaplan–Meiersurvival plot of TCL-1 mice treated with auranofin compared with untreated mice. (P ¼ 0.001, log-rank (Mantel–Cox) test.






concomitantly increased ROS levels inMEC-1 and primary CLLcells (Fig. 3A). Cotreatment with the antioxidant N-acetylcysteine (NAC) inhibited auranofin-mediated ROS inductionin MEC-1 and primary CLL cells (Fig. 3B and SupplementaryFig. S1). Because they exhibit higher levels of ROS and activityof Nrf2, CLL cells have concomitant overexpression of thedownstream Nrf2 targets such as HMOX-1 (10, 11). Therefore,here, we determined the effect of auranofin-induced ROS onthe levels of Nrf2 and its target genes in CLL cells. The confocalimmunofluorescence analyses in Fig. 3C demonstrate thatauranofin treatment not only increased the levels of Nrf2protein, but also augmented its nuclear localization in primaryCLL cells. Moreover, immunoblot analyses of the cell lysates of

MEC-1 and primary CLL cells confirmed that auranofin treat-ment increasedNrf2 levels aswell as the levels ofNrf2-activatedgenes, including HMOX-1, TxNIP, and GCLM (Fig. 3D). Inaddition, cotreatment with NAC attenuated auranofin-medi-ated increase in the levels of Nrf2 and its targeted geneexpressions, especially of HMOX-1 and GCLM, confirmingtheir linkage to auranofin-induced oxidative stress (Fig. 3E).In primary CLL cells, auranofin-induced ROS levels wereassociated with increases in the mitochondrial permeabilitytransition (Dcm) and percentage of Annexin V–positive cells(Fig. 3F). Cotreatment with NAC inhibited auranofin-inducedapoptosis of primary CLL cells (Fig. 3G). In cultured MEC-1cells, cotreatment with NAC and auranofin inhibited

D

TxNIP

0 0.1 0.25 0 0.25 1.0 0 0.25 1.0 µmol/L, AF

Primary CLL

MEC-1 CLL#2

HMOX1

Nrf2

ββ-Actin

GCLM

CLL#1

0

5

10

15

20

25

30

35

40

Control 0.25 1.0

% In

crea

se in

RO

Sµmol/L, AF, 1 hour

AF treatmentAF + 1 mmol/L NACtreatment

MEC-1

DAPIPhalloidin egreM2frN

Primary CLL cells

Control

1.0µmol/L, AF

MEC-1Primary CLL

E Primary CLL

0 1.0 1.0 µmol/L, AF0 0 2.5 mmol/L, NAC

HMOX1

Nrf2

GCLM

β-Actin

1.0 8.52 5.26

1.0 267.5 32.0

1.0 6.49 2.39

G

0

20

40

60

80

% A

po

pto

sis

Primary CLL (n = 5)

0 0 1.0 1.0 µmol/L, AF0 1.0 0 1.0 mmol/L, NAC

2.5 mmol/L NAC

1 µmol/L AF

NAC + AFControl

Primary CLL cells (n = 8)1.0 µmol/L AF, 24 h

% (

corr

ecte

d t

o c

on

tro

l)

F

0

20

40

60

80

100

H

β-Actin

Cl PARP

MEC-1

––

+–

–+

++

2.5 mmol/L NAC, 24 h1.0 µmol/L AF, 24 h

0

10

20

30

40

50

% o

f C

ells

(24

h) MEC-1

Annexin V+

Sub-G1

% In

crea

se

in R

OS 40

20

0

–20

–40

–60

–80

–100

A B C

% D

ecre

ase

in

TR

R a

ctiv

ity

Figure 3. Treatment with auranofin (AF) induces an oxidative stress response in cultured and primary CLL cells. A, the percent induction of ROS and thepercent inhibition of TRR activity in MEC-1 and CD19þ primary CLL cells exposed to 1.0 mmol/L of auranofin for 1 or 8 hours, respectively. B, percentageincrease of ROS in MEC-1 cells exposed to auranofin with or without 1 mmol/L of NAC for one hour. Columns, mean of three independent experiments;bars, SEM. C, localization of Nrf2 in primary CLL cells treated with auranofin for 6 hours. Cells were also stained with phalloidin and DAPI. D, immunoblotanalyses of MEC-1 cells and CD19þ primary CLL cells treated with auranofin for 16 hours. E, immunoblot analyses of CD19þ primary CLL cellstreatedwith auranofin andNAC for 24 hours. F, induction of ROS, loss ofmitochondrialmembrane potential, and apoptosis of primaryCLL cells (n¼ 8) treatedwith auranofin as indicated. Values represent the percent induction corrected for the untreated, control cells. G, percent apoptosis of CD19þ primary CLLcells following treatment with auranofin and/or NAC for 48 hours. �, apoptosis values significantly less in cells treated with auranofin and NAC thanthose treated with auranofin alone (P ¼ 0.00062). H, percentage of sub-G1 and Annexin V–positive MEC-1 cells following treatment with auranofin and/orNAC for 24 hours. �, values significantly less in cells treated with auranofin and NAC than those treated with auranofin alone (P < 0.001). Immunoblot analyseswere also conducted as indicated.

Fiskus et al.





auranofin-induced apoptosis as demonstrated by significantreduction in Annexin V–positive cells and decreased sub-G1

fraction, as well as inhibition of PARP cleavage (Fig. 3H).

Auranofin perturbs biologic networks of genes in CLLcellsWe also determined the effects of auranofin treatment on

gene expression microarray profile in primary CLL cells.Figure 4A shows a heatmap of gene expression changes.Auranofin treatment for 4 or 10 hours significantly upregu-lated or downregulated the mRNA expression of a largenumber of genes, and the fold changes in the 39 most alteredmRNA gene expressions are shown in Fig. 4B and Supple-mentary Table S3. The figure shows that mRNA expressionlevels of the Nrf2 target genes HMOX-1 and GCLM, as well asof CHOP and the key HSPs were markedly induced byauranofin treatment in the primary CLL cells. To confirmthe induction of CHOP and Nrf2 target genes, total RNA fromuntreated and auranofin-treated cells was also reverse tran-

scribed and the resulting cDNA was used for quantitativePCR with the specific TaqMan Real-time PCR probes. Thisconfirmed that auranofin treatment markedly increased themRNA expression of the HMOX-1, GCLM1, and CHOP genesin primary CLL cells with or without deletion of 17p (Fig. 4Cand D). Datasets of genes with altered expression profilederived from microarray analyses were imported into theIngenuity Pathway Analysis (IPA) Tool (Ingenuity H Sys-tems). Within the gene list, IPA identified the top five mostperturbed networks in primary CLL cells following treat-ment with auranofin and assigned a score for these associ-ated network functions (Supplementary Table S4). The score(i.e., a score of 41) assigned by IPA indicates the probability(1 in 1041) that the focus genes in the dataset are groupedtogether in a perturbed network due to random chancealone. The differentially expressed genes and the five mostperturbed biologic networks identified by IPA revealedsignificant connectivity between genes (nodes) within dif-ferent biologic networks (Supplementary Fig. S4). Among

0

20

40

60

80

100

120

140

160

180

HMOX-1 GCLM CHOP

Control

1.0 µmol/L AF, 8 h

A B

Primary CLL

–10

0

10

20

30

40

50

60

HM

OX

1G

CL

MD

NA

JA4

HS

PB

1T

XN

MA

FF

HS

P90

AA

1D

NA

JB4

E2F

3S

QS

TM

1G

AD

D45

BU

BB

HS

P90

AB

1N

QO

1M

AF

GG

SR

PR

DX

1C

DK

N1A

KE

AP

1N

FE

2L2

SO

D1

TL

R1

TG

FB

R1

SM

AD

3B

AC

H1

TN

IP1

FL

T3

SY

KT

AP

2T

IFA

TN

FR

SF

1BR

GS

14IL

2RA

MA

P2K

6P

DE

4DIP

NC

F1C

AK

R7A

2P

SM

B9

PS

MB

8

Fo

ld e

xpre

ssio

n c

han

ge

Primary CLL

Rel

ativ

e m

RN

A e

xpre

ssio

n

C

Primary CLL

0

10

20

30

40

50

60

HMOX-1 GCLM CHOP

Rel

ativ

e m

RN

A e

xpre

ssio

n 17p-deleted primary CLLControl

1 µmol/L AF, 8 h

D

1 2 1 2 1 2

HMOX1

TXNRD1

TXN

HSPA5

HSP90AA1

0 4 h 10 h, 1 µmol/L AF

P = 0.0025

P = 0.0018

P = 7.89 x10–5

P = 0.0012P = 4.18 x10–6

P = 0.0027

–2.1–1.7–1.3–0.8–0.4

00.420.841.261.672.09

Figure 4. mRNA expression profiling of auranofin (AF)-treated primary CLL cells. A, heatmap representing the relative expression of 98 probe sets (81 genes) inCLL cells from two patients treated with 1.0 mmol/L of auranofin for 4 and 10 hours in vitro compared with untreated cells (>2-fold change; P < 0.01 byANOVA). Patient samples are arranged in columns. Gene symbols highlight select genes. Gene expression is median centered and scaled as indicated.B, graphical representation of the top thirty-nine up- and downregulated genes in primary CLL cells following treatment with 1.0 mmol/L of auranofin for8 hours compared with untreated control cells. C, relative mRNA expression of HMOX-1, GCLM, and CHOP in primary CLL cells treated with 1.0 mmol/L ofauranofin for 8 hours compared with untreated control cells. D, relative mRNA expression of HMOX-1, GCLM, and CHOP in 17p-deleted primary CLL cellstreated with 1.0 mmol/L of auranofin for 8 hours compared with untreated control cells.






these, free radical scavenging, cellular compromise, andprotein degradation networks were the highest rated net-works with high significance scores.

Inhibition of HMOX-1 and GCLM augments auranofin-induced apoptosis of CLL cells

We next determined the mechanistic relevance of aura-nofin-induced, oxidative stress response, which is repre-sented by Nrf2-induced levels of HMOX-1 and GCLM. Forthis, we knocked down HMOX-1 gene expression by shRNAor inhibited HMOX-1 activity by treatment with its selectiveinhibitor ZnBG (37). The effect of this was determined onauranofin-induced apoptosis. As shown in Fig. 5A, treatmentwith shRNA to HMOX-1 depleted HMOX-1 protein levels butdid not affect the levels of Nrf2, GCLM1, or induce the levelsof cleaved caspase-3. In contrast, cotreatment with shRNA toHMOX-1, but not with the nontargeted shRNA, significantlyincreased auranofin-induced apoptosis (Fig. 5B). In addition,cotreatment with ZnBG and auranofin synergisticallyinduced apoptosis of MEC-1 and primary CLL cells. Thiswas assessed by the median dose effect isobologram anal-yses where the CIs were noted to be <1.0 (Fig. 5C and

Supplementary Fig. S5; ref. 38). GCLM is the modifier sub-unit that heterodimerizes with the catalytic subunit (GCLC)of glutamate cysteine ligase (GCL), an enzyme catalyzing theinitial rate-limiting step in the GSH synthesis (39). Byinhibiting GCL activity, treatment with BSO inhibits GSHsynthesis (40). We also determined the apoptotic effects ofcotreatment with BSO on auranofin-induced apoptosis ofMEC-1 and primary CLL cells. Cotreatment with BSO syn-ergistically enhanced auranofin-induced apoptosis (Fig. 5D).Again, the CI values were <1.0. Collectively, these findingsdemonstrate that abrogation of Nrf2-induced target geneexpressions, which represent the adaptive oxidative stressresponse, sensitizes CLL cells to auranofin.

Auranofin-induced ROS increases ER stress, whichcontributes to auranofin lethality in CLL cells

Increased ROS levels and perturbation in the intracellularredox status increase the levels of unfolded proteins in the ERand induce ER stress response (UPR; refs. 12, 22, 23). UPRinduces PERK-mediated phosphorylation of eukaryotic initi-ation factor-2a, which blocks cap-dependent protein transla-tion but allows preferential translation of ATF4 (22). While

0 0.2 0.4 0.6 0.8 1.00

0.5

1.0

1.5

2.0

2.5

Fractional effect

CI

AF + BSOPrimary CLLMEC-1

BA

Primary CLLAF + ZnBG

0 0.2 0.4 0.6 0.8 1.00

0.5

1.0

1.5

Fractional effect

CI

MEC-1

C D

HMOX1

Nrf2

GCLM

Cleavedcaspase-3

ββ-Actin

MEC-1

0 0.5 0 0.5 µmol/L, AF

sh-NT sh-HMOX1

1.0 1.17 1.0 1.21

1.0 7.8 0.44 3.7

1.0 5.43 1.19 5.26

1.0 2.38 1.1 5.220

10

20

30

40

50

60

0 100 250

% A

po

pto

sis

nmol/L, AF, 48 h

MEC-1sh-NT

sh-HMOX1

*

*

Figure 5. Inhibition of HMOX-1 by shRNA or cotreatment with auranofin (AF) and ZnBG or BSO synergistically induces apoptosis of cultured andprimaryCLL cells. A, immunoblot analyses ofMEC-1 cells transducedwith nontargeted shRNA (sh-NT) or sh-HMOX1 for 48 hours, then treatedwith auranofinfor 8 hours. B, percent apoptosis of sh-NT and sh-HMOX-1–transduced MEC-1 cells following treatment with auranofin for 48 hours. Columns, meanof three independent experiments; bars,SEM. �, apoptosis values significantly greater in sh-HMOX1–transducedcells comparedwith sh-NT–transducedcells(P < 0.001). C, MEC-1 cells (black line) and primary CLL cells (gray line) were exposed to of auranofin (100–500 nmol/L) and ZnBG (10–20 mmol/L)for 48 hours. After treatment, the percentage of apoptotic cells was determined by flow cytometry. Median dose effect and isobologram analyses wereperformedutilizingCalcusyn software.CI values less than1.0 indicate a synergistic interactionbetween the twoagents. D,MEC-1 cells (black line) andprimaryCLL cells (gray line) were treated with auranofin (10–250 nmol/L) and BSO (1–10 mmol/L) for 48 hours. Following this, the percentage of apoptoticcells was determined by flow cytometry. Isobologram analyses were performed as described above.

Fiskus et al.





upregulating chaperone proteins, for example, GRP78, req-uired in restoring the ER function, ATF4 also induces theprodeath transcriptional regulator CHOP (22, 23). Treatmentwith auranofin induced GRP78 levels and increased the mRNAand protein levels of CHOP (Fig. 6A and B). Importantly,auranofin treatment, also dose dependently, increased theratio of CHOP to GRP78 (Fig. 6B). Cotreatment with theantioxidant NAC attenuated the expression of ATF4 and CHOP(Fig. 6C and D), and significantly inhibited apoptosis of CLLcells, as shown above in Fig. 3G and H. We also determined theeffects of treatment with auranofin on the induction of ERstress-based UPR. In MEC-1 cells, treatment with auranofininduced the expression levels of p-PERK, the spliced form ofXBP1 (XBP1-s) and CHOP (Fig. 6E). Auranofin treatment alsoinduced the expression of the ER chaperones GRP78 and

Calreticulin, but not ERp57 (Fig. 6E). In addition, the aurano-fin-induced effects on these protein expressions were attenu-ated by cotreatment with NAC. We also determined whethertreatment with hydrogen peroxide would also induce theexpression of GRP78 and CHOP in CLL cells. Treatment withhydrogen peroxide also induced Nrf2, GRP78, and CHOP levels,which was associated with an increase in the levels of cleavedcaspase-3 (Supplementary Fig. S6). These results suggest thatthe effects of auranofin-induced oxidative response are mim-icked by treatment with hydrogen peroxide. We next deter-mined the effects of shRNA-mediated depletion of CHOP inCLL cells. Knockdown of CHOP by two separate lentivirus-transduced CHOP shRNAs, markedly attenuated CHOPexpression in MEC-1 cells (Fig. 6F). This was associated withan appreciable reduction in auranofin-induced apoptosis in

A B

ββ-Actin

0 0.25 1.0 µmol/L, AF

Primary CLL

GRP78

CHOP

0

2

4

6

8

10

12

GRP78 CHOP

Rel

ativ

e m

RN

A e

xpre

ssio

n

Control

1 µmol/L, AF

Primary CLL (n = 4)

F

0

0.2

0.4

0.6

0.8

1

1.2R

elat

ive

CH

OP

m

RN

A e

xpre

ssio

n

MEC-1

1.0 0.27 0.05β-Actin

CHOP

sh-N

T

shCHOP

#1 #2

C

Primary CLL


β-Actin

CHOP

GRP78

E

1.0 3.12 1.32

1.0 2.60 1.52

0

2

4

6

0 0.25 1.0

CH

OP

/GR

P78

rat

io

µmol/L, AF, 16 h

Primary CLL

Primary CLLMEC-1

Contro

l0.

5 µm

ol/L

AF

0.5

µmol

/L A

F

+ 2.

5 m

mol

/L N

ACCon

trol

1.0

µmol

/L A

F1.

0 µm

ol/L

AF

+ 2.

5 m

mol

/L N

AC

Rel

ativ

e m

RN

Aex

pre

ssio

n

G

*

0

5

10

15

20

25

30

35

40

0 0.25%

Ap

op

tosi

sµmol/L, AF, 48 h

MEC-1

sh-NT

shCHOP#2

D

β-Actin

Calreticulin

CHOP

PERK

ERp57

GRP78

hsp70

p-PERK

MEC1

––

+–

–+

++

2.5 mmol/L NAC, 24 h

1.0 µmol/L AF, 24 h

XBP1-s

1.0 1.0 1.37 1.10

1.0 1.06 1.73 1.36

1.0 1.04 1.60 1.18

1.0 1.0 1.50 1.10

76543210

ATF4CHOP

Figure 6. Treatment with auranofin (AF) induces ER stress in CLL cells. A, fold induction of GRP78 and CHOP mRNA in primary CLL (n ¼ 4) treated with1.0 mmol/L of auranofin for 8 hours is presented. B, Western blot analyses of CHOP and GRP78 in primary CLL cells treated with auranofin for 24 hours (top).Ratio of induction of CHOP to GRP78 in the primary CLL cells (bottom). C, relative mRNA expression of ATF4 and CHOP in MEC-1 and primaryCLL cells following treatment auranofin and/or NAC for 8 hours. D, immunoblot analyses of GRP78, CHOP, and b-actin in the cell lysates fromCD19þ primaryCLL cells treated with auranofin and/or NAC for 24 hours. E, immunoblot analyses of MEC-1 cells following treatment with auranofin and/or NAC for24 hours. F, relative CHOPmRNA expression 48 hours posttransduction in nontargeted (sh-NT) or CHOP shRNA-transducedMEC-1 cells (top).Western blotanalysis of CHOP and b-actin in MEC-1 cells before auranofin treatment (bottom). G, percent apoptosis of stably transfected sh-NT or sh-CHOP#2MEC-1 cells treatedwith 0.25mmol/L of auranofin for 48 hours. �, apoptosis values significantly less inMEC-1 sh-CHOPcells treatedwith auranofin comparedwith sh-NT cells (P < 0.02).






MEC-1 cells (Fig. 6G). These findings demonstrate that aur-anofin-induced ROS also leads to induction of a lethal ERstress, associated with increase in the ratio of CHOP to GRP78in CLL cells, whereas abrogation of auranofin-induced CHOPlevels partially undermines anti-CLL activity of auranofin.

Auranofin treatment induces proteotoxic stress withdepletionofHDAC6, ZAP70, andprosurvival hsp90 clientproteins in CLL cells

Next, we determined whether auranofin treatment leads toproteotoxic stress and the disassociation of HDAC6 from thehsp90–HSF1-p97 complex and induction of the heat shock

response (13, 14, 41). Data presented in Fig. 7A are represen-tative of the results obtained in three similarly treated separatesamples of primary CLL cells, which show that treatment withauranofin appreciably increased poly-Ub proteins in primaryCLL cells. Cotreatment with NAC inhibited this accumulation.Auranofin-induced accumulation of poly-Ub proteins was asso-ciated with decreased binding of hsp90 to HDAC6, HSF1, andp97 in CLL cells (Fig. 7B). Auranofin treatment led to a markeddepletion of HDAC6 levels through degradation by the protea-some, which was associatedwith the hyperacetylation of hsp90,detected by immunoblottingwith an anti-acetylated-K69 hsp90antibody (Fig. 7C; ref. 17). Confocal immunofluorescence

B

IgG 0 1.0 µmol/L, AF, 8 h

Primary CLL

IP: hsp90IB:

hsp90

HDAC6

HSF1

p97

C

ββ-Actin

Ac-K-69hsp90

0 1.0 µmol/L, AF

Primary CLL

0 1.0 µmol/L, AF

β-Actin

HDAC6

Primary CLL

D

0.25µmol/L, AF

PhalloidinDAPI MergeHSF-1

Control

1.0µmol/L, AF

Primary CLL E

β-Actin

hsp70

hsp27

hsp40

0 0.25 1.0 µmol/L, AF, 16 h

Primary CLLPrimary CLL

ZAP70

c-RAF

AKT

β-Actin

0 0.25 0.5 1.0 µmol/L, AF, 16 h

F

0

0.2

0.4

0.6

0.8

1

1.2

Control 1.0 µmol/L AF

Exp

ress

ion

of

HD

AC

6 Primary CLL (n = 10)

16 h 16 h

1.0 1.50

G

c-RAF

HDAC6

β-Actin

0 1.0 0 1.0 µmol/L, AF0 0 20 20 nmol/L, CZ

Primary CLL, 8 h

ZAP70

1.0 0.51 3.17 3.07

1.0 0.21 1.85 0.95

1.0 0.22 4.51 1.59

APrimary CLL


β-Actin

Poly Ubproteins

Figure 7. Treatment with auranofin (AF) inhibits hsp90 chaperone function and induces a heat shock response in CLL cells. A, representativeimmunoblot analyses of polyubiquitinated proteins (poly Ub) and b-actin in the cell lysates from three primary CLL samples treated with auranofinalone (1.0 mmol/L) or auranofin and NAC (2.5 mmol/L) for 24 hours. B, immunoblot analysis of HDAC6, HSF1, and p97 expression in hsp90immunoprecipitates following treatment with 1.0 mmol/L of auranofin for 8 hours. The blot was stripped and reprobed for hsp90. C, representativeimmunoblot analysis of HDAC6 expression from primary CLL cells (n ¼ 10) exposed to 1.0 mmol/L of auranofin for 16 hours. The values in the graphunderneath the blot represent quantitation of the levels of HDAC6 (mean � SE) relative to the untreated control (arbitrarily set as 1.0) as determined bydensitometry analysis. The right panel shows the levels of acetylated hsp90 using anti-acetyl-K69 hsp90 antibody, following exposure of a representativeprimary CLL sample to 1.0 mmol/L of auranofin for 16 hours. D, localization of HSF1 in primary CLL cells exposed to auranofin for 6 hours. A representativeimage is shown. E, immunoblot analyses of hsp40, hsp27, hsp70, and b-actin in the cell lysates from primary CLL cells exposed to auranofin for16 hours. F, immunoblot analyses of ZAP70, c-RAF, AKT, and b-actin in the cell lysates from primary CLL cells exposed to auranofin for 16 hours. G,immunoblot analyses of c-RAF, ZAP70, HDAC6, and b-actin in the lysates from primary CLL cells treated with 1.0 mmol/L of auranofin and/or 20 nmol/L ofcarfilzomib (CZ) for 8 hours. The numbers underneath the bands represent densitometry analysis performed on representative immunoblot analyses.

Fiskus et al.





analysis showed that auranofin treatment increased the nuclearlocalization of HSF1, consistent with the disruption of itsbinding to hsp90 (Fig. 7D). This was also associated with theinduction of the levels ofHSPs hsp27, hsp40, andhsp70 (Fig. 7E).By inducing hyperacetylation and inhibition of chaperonefunction of hsp90, treatment with auranofin also dose depen-dently depleted the intracellular levels of the CLL relevant,progrowth and prosurvival hsp90 client proteins ZAP70, c-RAF,and AKT (Fig. 7F; refs. 18, 19, 25). Their decline was due todegradation by the proteasome because cotreatment with theproteasome inhibitor carfilzomib restored the levels of theseclient proteins (Fig. 7G).

DiscussionRecently, treatment with auranofin was discovered to inhibit

TrxR and induce ROS in E. histolytica trophozoites, which wasassociated with high in vitro an in vivo potency of auranofinagainst E. histolytica, and auranofin was granted orphan drugstatus by the FDA (42).With the goal of potential repurposing ofauranofin for the treatment of CLL, our preclinical studiespresented here demonstrate that auranofin also inhibits TrxRand inducesROS in the cultured andprimaryCLL cells. Here,wealso show for the first time that treatment with auranofininduces lethal oxidative and linked to it, proteotoxic and ERstress-based UPR responses in CLL cells. In addition, auranofintreatment is selectively lethal to primary CLL cells, whilesparing the CD34þ normal progenitor and B cells. This maybe due to previously documented higher levels of ROS and Nrf2levels/activity in the primary CLL versus normal B cells, whichrender CLL cells more sensitive to augmented oxidative stressdue to auranofin-inducedROS levels (11). Furthermore, cotreat-ment with NAC abrogated the lethal activity of auranofin.Previous reports indicated that redox-active compounds,including those containing a-b unsaturated carbonyls, isothio-cyanates, arsenic trioxide, and ON01910.Na, that induce ROSand Nrf2 activity are also lethal against CLL cells (5, 11, 43).Auranofin treatment was also effective in inducing apopto-

sis in CLL cells exhibiting the biologic and genetic featuresassociated with poor clinical outcome, including high expres-sion of ZAP70, unmutated IGHV genes, and acquired chromo-somal abnormalities such as deletion 17p13 and deletion11q22. Ourfindings also demonstrate that the in vitro coculturewith stromal cells reduces auranofin-induced apoptosis inMEC-1 cells, which has been attributed to reduced "apoptoticpriming," defined as the proximity of a cell to the apoptoticthreshold measured by a functional assay that assesses mito-chondrial depolarization in response to BH3-only peptides(30, 44). However, auranofin treatment retained the ability todose dependently exert lethal activity against CLL cells despitecoculture with NLCs. Furthermore, the capacity to markedlyreduce the in vivo CLL cell burden in the TCL-1 mouse modelsupports the conclusion that auranofin treatment overcomesthe protective effects of stroma not only in vitro, but also in thein vivo setting.IPA analysis of genes differentially expressed in response to

auranofin identified five significantly perturbed biologic net-works with high significance scores, including free radical

scavenging, cellular compromise, and protein degradation.These findings underscore that auranofin-induced oxidativestress and transcriptional activity of Nrf2 play a key role ineliciting the stress-adaptive responses in auranofin-treated CLLcells. Nrf2 activity led to increased expression levels of HMOX-1and GCLM, as well as increase in the ER stress-based UPR-associated CHOP levels. It is noteworthy that auranofin strik-ingly inducedGCLM,HMOX-1, andCHOP levels in primaryCLLcells with or without the deletion of chromosome 17p. Asdescribed in AML cells (45), we found that knockdown of thelevels and/or activity of HMOX-1 synergistically enhancedauranofin-induced apoptosis of CLL cells. Cotreatment withBSO also synergistically increased auranofin-induced apoptosisdue to upregulation of GCLM levels and GCL activity, therebyabrogating the protective effects mediated by auranofin-induced GSH. Cotreatment with auranofin and BSO is likelyto have even greater efficacy in overcoming the prosurvivaleffects of stromal microenvironment in vitro and in vivo, giventhat stromal cells mediate this effect by helping elevate theintracellular levels of GSH in CLL cells (30). Previous reportshave demonstrated that loss of TrxR activity renders trans-formed cells especially susceptible to GSH deprivation (46).Increased oxidative stress caused by simultaneous inhibition ofGSH and thioredoxin metabolism has also been shown tosensitize cancer cells to other therapeutic agents (47, 48).

By inducing ROS levels and oxidative stress, auranofin treat-ment concomitantly inducedER stress response, highlighted byelevated levels of p-PERK, XBP1-s, and CHOP, as well asincrease in the levels of GRP78 and calreticulin. Althoughauranofin-mediated increase in GRP78 levels would be protec-tive, a protracted greater elevation in CHOP relative to GRP78levels could have lethal consequences (24). Protracted CHOPinduction is known to mediate apoptosis, mainly attributed toinduction of BIM with simultaneous inhibition of BCL2 andMCL1 levels (22). Findings presented here demonstrating thatthe shRNA-mediated knockdown of CHOP modestly inhibitsauranofin-induced apoptosis suggest that auranofin-inducedER stress is a secondary response to auranofin-induced ROSthat can amplify the cell death pathway, rather than being amediator of it in CLL cells. Auranofin-induced increase in thelevels of poly-Ubproteins led to hyperacetylation and inhibitionof chaperone function of hsp90. The resulting depletion in thelevels of HDAC6, ZAP70, AKT, and c-RAF would also increasethe apoptotic priming as well as attenuate growth and reducethe survival of CLL cells. Collectively, our findings underscorethat auranofin-induced oxidative stress is also linked to pro-teotoxic and ER stress-based UPR, which could amplify thelethal effects of auranofin in CLL cells.

Research on phosphine gold compounds for elucidatingtheir full range of biologic activity and therapeutic potentialis ongoing (31, 42). Findings presented here point toward arepurposing potential for auranofin in the treatment of CLL. Tothis end, under the umbrella of an FDA-approved Investiga-tional New Drug, a clinical trial of auranofin in treatment-refractory CLL, is underway (4).

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.






Authors' ContributionsConception and design: N. Saba, M. Ghias, C.P. Austin, P. Huang, A. Wiestner,K.N. BhallaDevelopment of methodology: N. Saba, C.P. Austin, A. Melnick, A. Wiestner,K.N. BhallaAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.):W.Fiskus, N. Saba,M. Ghias, J. Liu, L. Chauhan, R. Rao,K. Maddocks, P. Huang, A. Wiestner, K.N. BhallaAnalysis and interpretation of data (e.g., statistical analysis, biostatis-tics, computational analysis): W. Fiskus, N. Saba, M. Shen, M. Ghias, S.D.Gupta, L. Chauhan, R. Rao, S. Gunewardena, C.P. Austin, P. Huang, A. Wiestner,K.N. Bhalla

Writing, review, and/or revision of the manuscript: W. Fiskus, N. Saba,M. Shen, C.P. Austin, J. Byrd, A. Melnick, A. Wiestner, K.N. BhallaAdministrative, technical, or material support (i.e., reporting or organiz-ing data, constructing databases): N. Saba, M. Ghias, K. Schorno, C.P. AustinStudy supervision: N. Saba, A. Wiestner, K.N. Bhalla

The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received July 22, 2013; revised January 7, 2014; accepted February 6, 2014;published OnlineFirst March 5, 2014.

References1. Zenz T, Mertens D, Kuppers R, Dohner H, Stilgenbauer S. From

pathogenesis to treatment of chronic lymphocytic leukaemia. Nat RevCancer 2010;10:37–50.

2. Gribben JG, O'Brien S. Update on therapy of chronic lymphocyticleukemia. J Clin Oncol 2011;29:544–50.

3. Zenz T, Gribben JG, Hallek M, Dohner H, Keating MJ, Stilgenbauer S.Risk categories and refractoryCLL in the era of chemoimmunotherapy.Blood 2012;119:4101–7.

4. Weir SJ, DeGennaro LJ, Austin CP. Repurposing approved and aban-doned drugs for the treatment and prevention of cancer throughpublic-private partnership. Cancer Res 2012;72:1055–8.

5. Chapman CM, Sun X, Roschewski M, Aue G, Farooqui M, Stennett L,et al. ON 01910.Na is selectively cytotoxic for chronic lymphocyticleukemia cells through a dualmechanism of action involving PI3K/AKTinhibition and induction of oxidative stress. Clin Cancer Res2012;18:1979–91.

6. Trachootham D, Zhang H, Zhang W, Feng L, Du M, Zhou Y, et al.Effective elimination of fludarabine-resistant CLL cells by PEITCthrough a redox-mediated mechanism. Blood 2008;112:1912–22.

7. Zhou Y, Hileman EO, Plunkett W, Keating MJ, Huang P. Free radicalstress in chronic lymphocytic leukemia cells and its role in cellularsensitivity to ROS-generating anticancer agents. Blood 2003;101:4098–104.

8. Bindoli A, Fukuto JM, Forman HJ. Thiol chemistry in peroxidase catal-ysis and redox signaling. Antioxid Redox Signal 2008;10:1549–64.

9. TrachoothamD, Alexandre J, HuangP. Targeting cancer cells byROS-mediated mechanisms: a radical therapeutic approach? Nat Rev DrugDiscov 2009;8:579–91.

10. Malhotra D, Portales-Casamar E, Singh A, Srivastava S, Arenillas D,Happel C, et al. Globalmapping of binding sites forNrf2 identifiesnoveltargets in cell survival response through ChIP-Seq profiling and net-work analysis. Nucleic Acids Res 2010;38:5718–34.

11. Wu RP, Hayashi T, Cottam HB, Jin G, Yao S, Wu CC, et al. Nrf2responses and the therapeutic selectivity of electrophilic compoundsin chronic lymphocytic leukemia. Proc Natl Acad Sci U S A 2010;107:7479–84.

12. Avery S. Molecular targets of oxidative stress. Biochem J 2011;434:201–10.

13. Boyault C, Zhang Y, Fritah S, Caron C, Gilquin B, Kwon SH, et al.HDAC6 controls major cell response pathways to cytotoxic accumu-lation of protein aggregates. Genes Dev 2007;21:2172–81.

14. Matthias P, Yoshida M, Khochbin S. HDAC6 a new cellular stresssurveillance factor. Cell Cycle 2008;7:7–10.

15. Kawaguchi Y, Kovacs JJ, McLaurin A, Vance JM, Ito A, Yao TP. Thedeacetylase HDAC6 regulates aggresome formation and cell viabilityin response to misfolded protein stress. Cell 2003;115:727–38.

16. Rao R, Nalluri S, Kolhe R, Yang Y, Fiskus W, Chen J, et al. Treatmentwith panobinostat induces glucose-regulated protein 78 acetylationand endoplasmic reticulum stress in breast cancer cells. Mol CancerTher 2010;9:942–52.

17. Yang Y, Rao R, Shen J, Tang Y, Fiskus W, Nechtman J, et al. Role ofacetylation and extracellular location of heat shock protein 90alpha intumor cell invasion. Cancer Res 2008;68:4833–42.

18. Trepel J, Mollapour M, Giaccone G, Neckers L. Targeting the dynamicHSP90 complex in cancer. Nat Rev Cancer 2010;10:537–49.

19. Moulick K, Ahn JH, Zong H, Rodina A, Cerchietti L, Gomes DaGamaEM, et al. Affinity-based proteomics reveal cancer-specific networkscoordinated by Hsp90. Nat Chem Biol 2011;7:818–26.

20. RaoR, FiskusW,YangY,HergerB, Joshi R, Fernandez P, et al. HDAC6inhibition enhances 17-AAG-mediated abrogation of HSP90 Chaper-one function in human leukemia cells. Blood 2008;112:1886–93.

21. Castro JE, PradaCE, Loria O, Kamal A, Chen L, Burrows FJ, et al. ZAP-70 is a novel conditional heat shock protein 90 (Hsp90) client: inhibitionof Hsp90 leads to ZAP-70 degradation, apoptosis, and impairedsignaling in chronic lymphocytic leukemia. Blood 2005;106:2506–12.

22. Tabas I, Ron D. Integrating the mechanisms of apoptosis induced byendoplasmic reticulum stress. Nat Cell Biol 2011;13:184–90.

23. Walter P, Ron D. The unfolded protein response: from stress pathwayto homeostatic regulation. Science 2011;334:1081–6.

24. Lin JH, Li H, Yasumura D, Cohen HR, Zhang C, Panning B, et al. IRE1signaling affects cell fate during the unfolded protein response. Sci-ence 2007;318:944–9.

25. Rao R, Nalluri S, Fiskus W, Savoie A, Buckley K, Ha K, et al. Role ofCAAT/enhancer binding protein homologous protein in panobinostat-mediated potentiation of bortezomib-induced lethal endoplasmicreticulum stress in mantle cell lymphoma cells. Clin Cancer Res2010;16:4742–54.

26. Burger JA, Ghia P, Rosenwald A, Caligaris-Cappio F. The microenvi-ronment in mature B-cell malignancies: a target for new treatmentstrategies. Blood 2009;116:3367–75.

27. Woyach JA, Johnson AJ, Byrd JC. The B-cell receptor signalingpathway as a therapeutic target in CLL. Blood 2012;120:1175–84.

28. Wiestner A. Emerging role of kinase targeted strategies in chroniclymphocytic leukemia. Blood 2012;120:4684–91.

29. Davids MS, Deng J, Wiestner A, Lannutti BJ, Wang L, Wu CJ, et al.Decreased mitochondrial apoptotic priming underlies stroma-mediat-ed treatment resistance in chronic lymphocytic leukemia. Blood2012;120:3501–9.

30. ZhangW, Trachootham D, Liu J, Chen G, Pelicano H, Garcia-Prieto C,et al. Stromal control of cystine metabolism promotes cancer cellsurvival in chronic lymphocytic leukaemia. Nat Cell Biol 2012;14:276–86.

31. Gandin V, Fernandes AP, RigobelloMP, Dani B, Sorrentino F, Tisato F,et al. Cancer cell death induced by phosphine gold(I) compoundstargeting thioredoxin reductase. Biochem Pharmacol 2010;79:90–101.

32. Kim HS, Zhang X, Choi YS. Activation and proliferation of folliculardendritic cell-like cells by activated T lymphocytes. J Immunol 1994;153:2951–61.

33. Burger JA, Tsukada N, Burger M, Zvaifler NJ, Dell'Aquila M, Kipps TJ.Blood-derived nurse-like cells protect chronic lymphocytic leukemia Bcells from spontaneous apoptosis through stromal cell-derived factor-1. Blood 2000;96:2655–63.

34. Bichi R, Shinton SA, Martin ES, Koval A, Calin GA, Cesari R, et al.Human chronic lymphocytic leukemia modeled in mouse by targetedTCL1 expression. Proc Natl Acad Sci U S A 2002;99:6955–60.

35. Johnson AJ, Lucas DM,Muthusamy N, Smith LL, Edwards RB, De LayMD, et al. Characterization of the TCL-1 transgenic mouse as apreclinical drug development tool for human chronic lymphocyticleukemia. Blood 2006;108:1334–8.

Fiskus et al.





36. SivinaM, Hartmann E, Vasyutina E, Boucas JM, Breuer A, KeatingM J,et al. Stromal cells modulate TCL1 expression, interacting AP-1components and TCL1-targeting micro-RNAs in chronic lymphocyticleukemia. Leukemia 2012;26:1812–20.

37. AppletonSD,ChretienML,McLaughlinBE, VremanHJ,StevensonDK,Brien JF, et al. Selective inhibition of heme oxygenase, without inhi-bition of nitric oxide synthase or soluble guanylyl cyclase, by metallo-porphyrins at low concentrations. Drug Metabol Disposit 1999;27:1214–9.

38. Chou TC, Talalay P. Quantitative analysis of dose-effect relationships:the combined effects of multiple drugs or enzyme inhibitors. AdvEnzyme Reg 1984;22:27–55.

39. Krejsa CM, Franklin CC, White CC, Ledbetter JA, Schieven GL,Kavanaugh TJ. Rapid activation of glutamate cysteine ligase followingoxidative stress. J Biol Chem 2010;285:16116–24.

40. Scarbrough PM, Mapuskar KA, Mattson DM, Gius D, Watson WH,Spitz DR. Simultaneous inhibition of glutathione- and thioredoxin-dependent metabolism is necessary to potentiate 17AAG-inducedcancer cell killing via oxidative stress. Free Radic Biol Med 2012;52:436–43.

41. Akerfelt M, Morimoto RI, Sistonen L. Heat shock factors: integrators ofcell stress, development and lifespan. Nat Rev Mol Cell Biol 2010;11:545–55.

42. Debnath A, Parsonage D, Andrade RM, He C, Cobo ER, Hirata K, et al.A high-throughput drug screen for Entamoeba histolytica identifies anew lead and target. Nat Med 2012;18:956–60.

43. Biswas S, Zhao X, Mone AP, Mo X, Vargo M, Jarjoura D, et al. Arsenictrioxide and ascorbic acid demonstrate promising activity againstprimary human CLL cells in vitro. Leuk Res 2010;34:925–31.

44. Ni Chonghaile T, Sarosiek KA, Vo TT, Ryan JA, Tammareddi A, MooreVdel G, et al. Pretreatment mitochondrial priming correlates with clinicalresponse to cytotoxic chemotherapy. Science 2011;334:1129–33.

45. Rushworth SA, MacEwan DJ. HO-1 underlies resistance of AML cellsto TNF-induced apoptosis. Blood 2008;111:3793–801.

46. Mandal PK, Schneider M, K€olle P, Kuhlencordt P, F€orster H, Beck H,et al. Loss of thioredoxin reductase 1 renders tumors highly suscep-tible to pharmacologic glutathione deprivation. Cancer Res 2010;70:9505–14.

47. Fath MA, Ahmad IM, Smith CJ, Spence J, Spitz DR. Enhancement ofcarboplatin-mediated lung cancer cell killing by simultaneous disrup-tion of glutathione and thioredoxin metabolism. Clin Cancer Res2011;17:6206–17.

48. Sobhakumari A, Love-Homan L, Fletcher EV, Martin SM, Parsons AD,Spitz DR, et al. Susceptibility of human head and neck cancer cells tocombined inhibition of glutathione and thioredoxin metabolism. PLoSONE 2012;7:e48175.






Published OnlineFirst March 5, 2014.Cancer Res Warren Fiskus, Nakhle Saba, Min Shen, et al. against Chronic Lymphocytic LeukemiaReticulum Stress and Exerts Potent Preclinical Activity Auranofin Induces Lethal Oxidative and Endoplasmic

Updated version

10.1158/0008-5472.CAN-13-2033doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2014/08/22/0008-5472.CAN-13-2033.DC1

Access the most recent supplemental material at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/early/2014/04/12/0008-5472.CAN-13-2033To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-13-2033

http://cancerres.aacrjournals.org/content/suppl/2014/08/22/0008-5472.CAN-13-2033.DC1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/early/2014/04/12/0008-5472.CAN-13-2033


auranoﬁn induces lethal oxidative and endoplasmic ... · transduced into mec-1 cells. forty-eight...

Documents