Zinc deficiency reduces paclitaxel e!cacy in LNCaPprostate cancer cells

Alison N. Killilea b, Kenneth H. Downing b, David W. Killilea a,*

a Nutrition & Metabolism Center, Children’s Hospital Oakland Research Institute, 5700 Martin Luther King Jr. Way,Oakland, CA 94609-1673, USA

b Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA

Received 11 July 2007; received in revised form 20 August 2007; accepted 21 August 2007

Abstract

Chemotherapeutics used to treat prostate cancer are often from a class of drugs that target microtubule networks, suchas paclitaxel. A previous report indicated that supplemental zinc sensitized prostate cancer cells to paclitaxel-induced apop-tosis, suggesting that increased zinc levels might enhance paclitaxel e!cacy. The e"ect of zinc deficiency on paclitaxel activ-ity is not known though, so we tested this in two prostate cancer cell lines maintained under moderately zinc-deficientconditions. LNCaP and PC3 cell lines were used as models of early and late-stage prostate cancer, respectively. Cells cul-tured in reduced zinc levels did not demonstrate altered cell viability, growth rates, or intracellular zinc content. Addition-ally, zinc deficiency alone had no apparent e"ect on cell cycle kinetics or apoptosis levels. However, the IC50 for paclitaxel-induced cell cycle arrest increased in LNCaP cells from zinc-deficient compared to zinc-replete conditions. Consequently,paclitaxel-induced apoptosis was reduced in LNCaP cells from zinc-deficient compared to zinc-replete conditions. In PC3cells, the e"ects of paclitaxel were independent of zinc status. Reduced extracellular zinc levels were shown to a"ect pac-litaxel activity in a prostate cancer cell line. Given the prevalence of zinc deficiency, determining how chemotherapeuticaction is modulated by zinc adequacy may have clinical importance.! 2007 Elsevier Ireland Ltd. All rights reserved.

Keywords: Apoptosis; Microtubules; Prostate cancer; Paclitaxel; Zinc

1. Introduction

According to 2006 statistics from the AmericanCancer Society, one out of every six men will bediagnosed with prostate cancer during his lifetime;consequentially, prostate cancer will result in nearly10% of all cancer deaths in men [1]. This makes

prostate cancer the second most common cancer(after skin cancer) and second leading cause of can-cer death (after lung cancer) in men from the UnitedStates. If the prostate cancer is detected at an earlystage, treatments including surgery, radiation, andhormone therapy are preferred. However, advancedstages of prostate cancer often become too complexfor surgical removal and refractory to hormonetherapy, so chemotherapy is then a recommendedtherapeutic approach. One of the most commondrug classes of chemotherapy used for the treatment

0304-3835/$ - see front matter ! 2007 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.canlet.2007.08.010

* Corresponding author. Tel.: +1 510 450 7627; fax: +1 510 4507910.

E-mail address: dkillilea@chori.org (D.W. Killilea).

Available online at www.sciencedirect.com

Cancer Letters 258 (2007) 70–79

www.elsevier.com/locate/canlet

of prostate cancer are the taxanes, such as the firstgeneration drug paclitaxel (Taxol, a trademark ofBristol-Myers Squibb). Paclitaxel and related tax-anes suppress normal microtubule dynamics, result-ing in mitotic block and eventual apoptosis [2–4].Paclitaxel has been used to successfully treat a vari-ety of cancers from ovarian, breast, and lung [5].Paclitaxel and other taxanes were later shown toimprove survival in prostate cancer patients, espe-cially in combination with estramustine [6,7]. How-ever, the taxanes have a narrow therapeutic windowand often cause significant negative side-e"ects.Therefore, any approach that enhances the e!cacyof the taxanes would be valuable to reduce the doseof drug necessary for clinical e"ect.

In 2002, Kolenko and colleagues reported thatpaclitaxel activity was enhanced by increased levelsof the micronutrient zinc in cultured prostate cancercell lines [8]. Increased intracellular zinc wasachieved by addition of zinc to the culture mediaalong with a selective ionophore. This treatmentresulted in increased cellular sensitivity to paclitaxelwhen compared to sham-treated cells. This findingwill have clinical importance if elevated zinc avail-ability similarly increases paclitaxel e!cacy in vivo.However, zinc deficiency, not excess, is a major pub-lic health problem; in the United States, approxi-mately 50% of men over the age of 50 are gettingless than the recommended daily allowance (RDA)and over 10% are getting less than the estimatedaverage requirement (EAR), two standard devia-tions below the RDA [9]. In fact, some epidemiolog-ical evidence indicates that inadequate zinc intake isassociated with cancer development, including inthe prostate; for reviews, see [10–13]. This observa-tion prompted us to determine whether reduced zincavailability could cause decreased sensitivity to pac-litaxel in prostate cancer cells.

We found that the ability of paclitaxel to arrestthe cell cycle and to cause apoptosis in a prostatecancer cell line was significantly reduced when cellswere maintained under zinc-deficient conditions.Reduced zinc alone had no apparent a"ect on cellu-lar growth, viability, cell cycle distribution, or apop-tosis levels, indicating that the cell lines adapted wellto zinc-deficient conditions. However, zinc defi-ciency did appear to alter the physiologic contextof cellular response to paclitaxel, resulting inreduced sensitivity. It is too early to speculatewhether this observation is relevant to the successrate of taxanes in treating prostate cancer patients,but if so, then zinc supplementation may be an inex-

pensive and safe co-therapy. It should be noted thatexcessively high zinc supplementation has actuallybeen associated with increased risk of prostate can-cer [14]. Thus we suggest that correcting zinc defi-ciency (rather than supraphysiological zinc dosing)along with taxane chemotherapy may be worthinvestigation.

2. Materials and methods

2.1. Chelators

AG MP-50 (100–200 mesh), BioRex MSZ 501D (25–35 mesh), and Chelex-100 (50–100 mesh) resins were pur-chased from Bio-Rad (Hercules, CA). Duolite C467 (16–50 mesh) resin, tris(2-aminoethyl)amine-agarose (TAEA),diethylenetriamine pentaacetic acid (DTPA), and otherchemicals were purchased from Sigma–Aldrich (St. Louis,MO), unless otherwise indicated.

2.2. Cell culture

LNCaP and PC3 prostate-derived cell lines were pur-chased from American Type Culture Collection (Rock-ville, MD). LNCaP and PC3 cell lines were cultured in90% RPMI-1640 media (Invitrogen, Carlsbad, CA) sup-plemented with 10% fetal bovine serum (FBS; Hyclone,Logan, UT), under standard conditions (37 "C, 5% CO2,100% humidity) with weekly passaging. Cell lines wereadapted to standard or zinc-deficient media for 3 months.Afterwards, a subset of zinc-deficient cultures was trans-ferred to media in which zinc levels were returned to con-trol levels. Cells were continuously cultured under theseconditions during drug sensitivity testing.

2.3. Zinc-deficient media

FBS was depleted of zinc by incubation with immobi-lized TAEA (10% w/v) for 10 min at 4 "C and then elemen-tal content of calcium, copper, iron, magnesium,potassium, rubidium, sodium, sulfur, and zinc was mea-sured as described below. Non-zinc metal levels in allTAEA-stripped FBS were adjusted to levels in controlFBS with appropriate metal salts as needed. Additionally,the zinc levels in the zinc-replete condition were adjustedto control FBS levels with zinc sulfate. All metal salts werecell culture-certified. Then TAEA-stripped FBS was steril-ized using 0.22 lm Millex-GV filters (Millipore, Billerica,MA) and periodically checked for endotoxin (EndosafeEndochrome-K assay; Charles River Laboratories,Charleston, SC). FBS used for control media was preparedin an identical fashion but without TAEA resin (sham).FBS protein content was measured by a commercial

A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79 71

bicinchoninic acid (BCA) assay (Pierce Biotechnology Inc,Rockford, IL) using bovine serum albumin as the standardaccording to manufacturer’s instructions.

2.4. Cell viability assays

Cell viability was assessed for early (resazurin reduc-tion) and end-stage (vital dye exclusion) toxicity. Forearly toxicity, cells cultured in multi-well plates were incu-bated with CellTiter-Blue Cell Viability Assay (PromegaCorporation; Madison, WI) for 2 h according to manu-facturer’s instructions. Fluorescence yield was monitoredusing a microplate fluorescence reader at an excitationof 560 nm and emission at 590 nm. For end-stage toxicity,cells cultured in multi-well plates were incubated with0.2% trypan blue in PBS for 5 min and then scored forexclusion of trypan blue dye.

2.5. Elemental analysis

The elemental content of media and cells was deter-mined by inductively-coupled plasma atomic emissionspectrometry (ICP) [15]. Media or cell pellets were dis-solved in OmniTrace 70% HNO3 (VWR International,West Chester, PA) for 12–16 h at 60 "C with orbital shak-ing. Lysates were then diluted with OmniTrace water(VWR International) to 5% HNO3 and introduced via apneumatic concentric nebulizer using argon as the carriergas into a Vista Pro ICP (Varian Inc., Palo Alto, CA).Elemental values were calibrated using elemental stan-dards and validated using National Institute of Standardsand Technology (NIST)-traceable 1577b bovine liver ref-erence material. The coe!cient of variation (CV) of intra-assay precision for zinc was 4.8% (n = 10 in 1 run) andinter-assay precision for zinc was 7.0% (12 independentruns) for the NIST reference material. Total intra-assayprecision for all elements measured in the NIST referencematerial had CVs ranging from 1.1% to 7.8% (n = 10 in 1run). Cesium (50 ppm) was used for ionization suppres-sion and yttrium (5 ppm) was used as an internal stan-dard. All reagents and plasticware were certified orroutinely tested for trace metal work. Data were collectedand summarized using native software (ICP Expert; Var-ian Inc.).

2.6. Cell cycle analysis

Randomly cycling cell populations were analyzed forcell cycle distribution by propidium iodide staining. Cellswere routinely cultured until approximately 60–70% con-fluent before being exposed to drugs. Cells were harvestedby trypsin–EDTA treatment (37 "C for 5 min), pelletedby centrifugation (1000g for 5 min), and washed in PBS.Pellets were resuspended in staining solution (50 lg/mlpropidium iodide, 0.1 mg/ml sodium citrate, 2 lg/ml ribo-nuclease A, and 0.03% Triton X-100) and vortexed for cell

lysis [16]. DNA content was analyzed using a FACSCali-bur flow cytometer (BD, Franklin Lakes, NJ). Data werecollected using 500,000 events per sample and mean fluo-rescence was summarized using native software (Cell-Quest Pro, BD).

2.7. Immunocytochemistry

Cells were plated on glass chamber slides precoatedwith polylysine. Cells were routinely cultured untilapproximately 60–70% confluent before being exposedto drugs. Slides were washed with TBS, fixed in 100%methanol for 5 min at !20 "C, rehydrated in TBS, perme-abilized with TBS and 0.1% Triton X-100 (TBST), andthen incubated in blocking solution (TBST and 2% bovineserum albumin). Cells were incubated with 500 lg/mlmonoclonal mouse anti-human b-tubulin antibody(Sigma–Aldrich) for 1 h, washed with TBST, incubatedwith 10 lg/ml donkey-anti-mouse IgG secondary anti-body labeled with Alexa Fluor 488 (Invitrogen), and co-stained with 300 nM 4 0,6-diamidino-2-phenylindole(DAPI). Slides were mounted and imaged by fluorescencemicroscopy. Apoptotic cells were defined as having pino-cytotic nuclei, membrane blebbing, and non-filamentousmicrotubule staining. The identity of the culture condi-tions was randomly coded prior to scoring for apoptosisby a trained observer.

2.8. Statistical analysis

Graphing, regression, and statistical analysis were per-formed using Prism 4.0 software (GraphPad, San Diego,CA). Significance was accepted at p < 0.05.

3. Results

The predominant source of the zinc in culture mediafor LNCaP and PC3 cell lines is FBS, so chelation wasused to deplete zinc levels. A search for a commerciallyavailable chelator with high selectivity for zinc revealedTAEA as a good option (Supplementary Table S1).TAEA at 10% w/v depleted the level of zinc in FBSfrom 32.6 ± 4.0 lM (n = 15) to 11.4 ± 2.7 lM (n = 15),a significant reduction of 65% of zinc compared to con-trol media. However, all other metals analyzed weredepleted by <15%, except for copper which wasdepleted by 30–40%. All TAEA-stripped FBS wasrepleted with copper and other metals as needed, whilezinc-replete media was also supplemented with zincto levels of control media. TAEA-stripped FBS wasroutinely measured for endotoxin, which was neverdetectable above sham FBS levels (data not shown).Additionally, resins with bound metal ions can trapproteins, so TAEA-stripped FBS protein levels wereroutinely measured. Total protein recovery was 95.7 ±3.9% (n = 5).

72 A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79

Freshly thawed LNCaP and PC3 cell lines were pas-saged directly into control or zinc-deficient media and cul-tured for 3 months for adequate adaptation. After thistime, a subpopulation of zinc-deficient cells was trans-ferred to zinc-deficient media supplemented with zinc tothe level of control media (zinc-replete media). The cellswere then tested for changes in cell physiology and drugsensitivity. During this time, no change in cell populationgrowth rates was observed (Fig. 1a and b). These datawere fit to a linear regression model, yielding a high good-ness of fit (r2 > 0.99) for all cells and conditions. Averagenative doubling times were calculated as 1.8–1.9 days forLNCaP cells and 1.6–1.7 days for PC3 cells. Additionally,no change in relative viability between culture conditionswas observed for either cell type when tested for resazurindye reduction (Fig. 1c and d). Cells were plated at varyingdensities to detect changes in viability over a range ofculture confluency. There was no significant loss of cellviability in zinc-deficient compared to zinc-replete condi-tions for either cell type. These findings were corroborated

with experiments showing no di"erence in trypan blue dyeexclusion as an end-stage cell viability marker (data notshown).

LNCaP and PC3 cell lines were also tested for changesin cell physiology following the 3 months adaptation per-iod. There were no changes in cell cycle distribution fromcultures maintained in zinc-deficient conditions in eithercell type (Fig. 2a and b). There were also no detectable dif-ferences in apoptosis rates between culture conditions foreach cell type (Fig. 4a and b – no drug treatment). Addi-tionally, the severity of zinc deficiency was assessed byanalyzing intracellular zinc content. Even after monthsof exposure to reduced extracellular zinc, the levels oftotal intracellular zinc from cells maintained in zinc-defi-cient media showed no consistent variation from controlcells in either cell type (Fig. 2c and d). Linear regressionanalysis indicated that the slopes of the fit data were notsignificantly di"erent from each other and not signifi-cantly di"erent from zero, indicating no change in intra-cellular zinc over time. Thus, zinc-deficient culture

Fig. 1. Zinc deficiency did not reduce cellular growth rates or viability in prostate cell lines. Mean population doublings were plotted as afunction of time in culture for LNCaP (a) and PC3 (b) cells under control (circle), zinc-deficient (ZnDf; open square), or zinc-replete(ZnRp; solid square) conditions, and then fit with linear regression model (r2 > 0.99, all conditions). The slopes for ZnDf and ZnRpconditions were less than 3% lower compared to control for each cell type. Cell viability was determined by resazurin dye reduction inLNCaP (c) and PC3 (d) cells under control (solid), zinc-deficient (ZnDf; open), or zinc-replete (ZnRp; hatched) conditions. Cells wereplated at indicated concentrations and allowed to grow for 3 days before application of CellTiter-Blue reagent for an additional 2 h.Fluorescence was measured at 560/590 nm and plotted in conjunction with culture confluency. PC3 cells in ZnDf media at the 2000 cells/cm2 density were significantly di"erent from control conditions using two-tailed ANOVA with Tukey’s post-test; however, this trend wasnot consistent with other plating densities.

A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79 73

conditions appeared to be mild and well-tolerated byLNCaP and PC3 cells and could be used to test the influ-ence of zinc status on chemotherapeutic activity.

The e"ect of paclitaxel on LNCaP and PC3 cell cycledistribution was then investigated as a function of zincstatus. Paclitaxel disrupts normal microtubule functionduring mitosis and thereby stalls the cell cycle in G2/Mphase. Cells were exposed to a range of paclitaxel concen-trations for 16 h and then analyzed for cell cycle distribu-tion by propidium iodide staining of DNA contentquantitated by flow cytometry. In PC3 cells, paclitaxeldemonstrated similar e!cacy in all groups (Fig. 3b). How-ever in LNCaP cells, paclitaxel demonstrated decreasede!cacy in zinc-deficient conditions, resulting in a signifi-cant increase in IC50 compared to control and zinc-repletecells (Fig. 3a). The calculated IC50 values of cells understandard media conditions were similar to literaturereported values [17]. In four independent experiments,the average IC50 for paclitaxel-induced G2/M arrest inzinc-deficient conditions in LNCaP cells was 2.4-foldgreater than matched controls. In four independent exper-

iments, the average IC50 for paclitaxel-induced G2/Marrest in zinc-deficient conditions in PC3 cells was 0.9-foldgreater than matched control.

The eventual consequence of stalled mitosis is apopto-sis. Therefore, paclitaxel-stimulated apoptosis was deter-mined in LNCaP and PC3 cell lines as a function ofzinc status. Cells were exposed to 25 nM paclitaxel for24 h and then visually scored for apoptosis by fluores-cence microscopy. In LNCaP cells, paclitaxel treatmentresulted in a reduced number of cells with pinocytoticnuclei in zinc-deficient compared to zinc-replete condi-tions; the number of cells positive for apoptosis wasreduced by approximately 50% in zinc-deficient condi-tions (Fig. 4a). In PC3 cells, paclitaxel treatment underthe same conditions resulted in similar e!cacy in allgroups (Fig. 4b). This pattern was also seen when analyz-ing cell cycle distribution described above for increases incells containing sub-G1 DNA content, indicative of lateapoptosis (data not shown). In LNCaP cells, paclitaxeltreatment resulted in fewer cells with sub-G1 DNA con-tent as compared to control and zinc-replete cells. In

Fig. 2. Zinc deficiency did not alter cell cycle distribution or intracellular iron levels in LNCaP and PC3 cells. Representative datashow means ± SD of cells in randomly-cycling populations within each phase of the cycle after 3 months acclimation to control(solid), zinc-deficient (ZnDf; open), or zinc-replete (ZnRp; hatched) conditions (a, LNCaP; b, PC3). No significant di"erences wereobserved in either cell type from zinc-deficient or -replete conditions compared to control using one-way ANOVA with Tukey’s post-test. Cells from same conditions were also sampled for total intracellular zinc levels as determined by ICP. Representative data showmeans ± SD of zinc content normalized to 1 · 106 cells (c, LNCaP; d, PC3). No consistent trend was detectable in either cell type ormedia conditions. Linear regression analysis of data indicated that slopes were not significantly di"erent from each other and werenot significantly non-zero.

74 A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79

PC3 cells, paclitaxel demonstrated similar e!cacy in allgroups. These data paralleled the findings from paclit-axel-induced changes in cell cycle distribution.

4. Discussion

A report showing that excess zinc improved thee!cacy of paclitaxel in prostate cancer cells [8] stim-ulated our interest in whether paclitaxel would be lesse!cacious when zinc was limiting. This question hasimportant public health implications because zincdeficiency is surprisingly common [9,18]. Zinc defi-ciency is especially pronounced in the elderly andunderprivileged groups due in part to excessive con-sumption of calorie-rich, micronutrient poor diets

[19,20]. Zinc is an essential micronutrient that isrequired for a vast number of physiological func-tions, including reproductive health [18]. In men,the prostate may be particularly sensitive to zinc defi-ciency due to the high levels of zinc required forproper function [21]. In fact, zinc uptake in prostateepithelial cells can be up to 10 times higher than othercells types in the body; interestingly, zinc levels inprostate epithelial cells rapidly decline upon trans-formation to a cancerous phenotype [22,23]. Zinchomeostasis may play important regulatory roles inboth normal growth and carcinogenesis within theprostate, though the exact details remain obscured.

To study the e"ects of zinc deficiency on prostatecancer cells, culture media with reduced zinc content

Fig. 3. Zinc deficiency was associated with reduced paclitaxel-induced cell cycle arrest in LNCaP cells. Subconfluent cultures were exposedto increasing paclitaxel (taxol) concentrations for 16 h and then analyzed for G2/M arrest by flow cytometry. Representative data showsummarized mean data fit to sigmoidal dose-response curves for cells from control (circle/solid line), zinc-deficient (ZnDf; open square/dotted line), or zinc-replete (ZnRp; solid square/solid line) conditions. In LNCaP cells, the dose–response curve of paclitaxel in zinc-deficient conditions was significantly right-shifted relative to control conditions, whereas zinc-replete conditions were similar to control (a).The mean ± SEM IC50 were determined to be 4.6 ± 1.8, 18.3 ± 1.7, and 5.1 ± 1.7 nM for control, ZnDf, or ZnRp conditions, respectively.In PC3 cells, the dose–response curve of paclitaxel in all conditions was similar (b). The mean ± SEM IC50 were determined to be6.9 ± 1.6, 4.2 ± 2.0, and 5.5 ± 1.6 nM for control, ZnDf, or ZnRp conditions respectively.

Fig. 4. Zinc deficiency was associated with reduced paclitaxel-induced apoptosis in LNCaP cells. Cultures were acclimated for 3 monthsunder control (solid) or zinc-deficient (ZnDf; open) conditions. Subconfluent cultures were scored for apoptosis by immunocytochemistryafter treatment with 25 nM paclitaxel (taxol) for 24 hours. In LNCaP cells, the percent of apoptosis in the culture was significantly greaterin zinc-deficient relative to control or zinc-replete conditions (a). In PC3 cells, the percent of apoptosis in the cultures was similar in allconditions (b). Two-way ANOVA with Bonferroni post test indicated that paclitaxel caused a significant increase in apoptosis in all cellsand zinc status (not indicated), but paclitaxel treatment resulted in significantly less apoptosis in ZnDf LNCaP compared to controlLNCaP cells (asterisk).

A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79 75

was prepared. Previous approaches to deplete zincin culture media have employed dialysis against che-lators such as EDTA and DTPA [24,25]. However,these chelators are not selective for zinc and causesevere depletion of many divalent metals. In somecases, the investigators repleted at least some ofthe lost non-zinc metals, but it is di!cult to assesswhether exogenously added metals return to correctfunctional pools. Our laboratory and others havepreviously extracted zinc from FBS with EDTA-analogs immobilized on beads (e.g. Chelex-100),which had somewhat better specificity for zinc thanother methods [26,27]. We continued to evaluateother immobilized chelators and found TAEAextracted a sizable fraction of zinc with only a mod-est depletion of other metals. To our knowledge, theselectivity of TAEA for zinc has not been previouslyreported. Additionally, TAEA did not substantiallytrap protein or add detectable amounts of endo-toxin. The main drawback is that zinc extractioncapacity for TAEA (60–70%) is somewhat lowerthan EDTA-analogs (often >90%) in FBS, but thelevel of deficiency was su!cient for this study.Therefore, TAEA was used to prepare the zinc-defi-cient culture media.

Both LNCaP and PC3 prostate cell linesappeared to adapt to moderately zinc-deficient con-ditions. No losses in cellular growth, viability, cellcycle distribution, or apoptosis rates were detectedunder zinc-deficient conditions alone. Moreover,the total intracellular zinc content was not signifi-cantly altered despite months of culturing in zinc-deficient conditions. We previously found that othercell types did respond with reduced intracellular zinclevels when cultured in culture media containingzinc-depleted FBS prepared using Chelex-100[27,28]. This is likely due to cell type di"erencesand a more severe zinc deficiency. In the currentstudy, we calculated that the number of moles ofzinc in the zinc-deficient media was still over 100-fold greater than the moles of zinc in the combinedintracellular volume of plated cells. Thus cells cul-tured in zinc-deficient media would still be able tomeet intracellular zinc requirements by increasingzinc intake and/or reducing zinc export. Moreover,other reports have shown that specific cell types aree!cient in maintaining their intracellular zinc levels[29,30]. The lack of severe phenotypic alteration dueto zinc-deficient conditions was beneficial forthis study because it provided a cell model thatwas uncomplicated by additional toxicities forevaluating the chemotherapeutic drug paclitaxel.

The major phenotypic di"erence we detected inzinc-deficient media was a loss of sensitivity to pac-litaxel-induced cell cycle arrest and apoptosis; zincdeficiency alone, but zinc deficiency was necessaryfor the e"ect since cells cultured in zinc-deficientmedia repleted with zinc responded to paclitaxelsimilarly to control cells.

Zinc-dependent loss of paclitaxel activity wasobserved in LNCaP, but not PC3, cell lines. Themeans by which zinc deficiency decreased paclit-axel sensitivity in LNCaP cells are unclear, thoughearlier studies have also shown di"erentialresponses of LNCaP and PC3 cells to alterationsin zinc homeostasis. For example, Feng andcolleagues reported a greater inhibition of cellgrowth, delayed cell cycle kinetics, and increasedapoptosis for LNCaP compared to PC3 cells whenexogenous zinc was added to culture media [31].Other groups have shown increased LNCaP cellsensitivity to apoptosis sensitizing agents and hor-mones [32–34]. Several of the known di"erencesbetween LNCaP and PC3 cells may explain thesefindings. First, LNCaP cells are androgen-sensi-tive, whereas PC3 cells are androgen-insensitive.Androgens are known to regulate prostate cellzinc homeostasis in vitro and in vivo [35], thusandrogen-sensitivity may result in LNCaP cellsbeing more a"ected by zinc-deficient conditions.Also, LNCaP cells are p53+/+, whereas PC3 cellsare p53!/!. p53 is a known zinc-dependentenzyme and zinc deficiency has been shown toinactivate p53 activity [27], which may makeLNCaP cells more sensitive to zinc-deficient con-ditions. Additionally, LNCaP cells have constitu-tively low levels of the anti-apoptotic nuclearfactor-jB (NF-jB), whereas PC3 cells have consti-tutively high NF-jB levels [36]. NF-jB bindingwas shown to be reduced in zinc deficiency [27],so cells with low NF-jB levels might be more sen-sitive to the e"ects of reduced zinc. However, it isnot clear which, if any, of these di"erences play amajor role in the di"erential sensitivity to paclit-axel during zinc deficiency.

The molecular mechanisms through which zincdeficiency diminishes paclitaxel e!cacy in theLNCaP cell line are not yet known, though severalpossibilities can be proposed. Zinc levels modulatekey apoptosis-related proteins such as bax, bcl-2,and caspases [37–41], so reduced zinc could alterapoptotic cascades. However, these perturbationswould occur downstream to the e"ects on cell cyclekinetics and thus not explain the changes seen in

76 A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79

cells cultured in zinc-deficient media. It has beenproposed that microtubule dynamics themselvesmay be disrupted by altered intracellular zinc levels,perhaps due to reduced tubulin polymerization rates[42,43]. However, intracellular zinc levels were notdecreased in our study; moreover, zinc deficiencyalone had no e"ect on cell cycling. Yet we cannotrule out that zinc deficiency might have caused adecrease in the labile zinc pool, comprising a tinyfraction of the total zinc in the cell, which was suf-ficient to a"ect paclitaxel action. Another reportedconsequence of zinc deficiency was increased oxida-tive stress levels [27,28,41,44,45]. While paclitaxelbinding to microtubules could be influenced byredox imbalances, this has not been previouslyreported. However, the zinc-deficient conditions inour study were more moderate than in previousstudies, so oxidative stress levels were expected tobe lower. Also, the mechanisms regulating cellcycling are very sensitive to DNA damage, yet wesaw no change in cell cycle kinetics for either celltype in zinc-deficient compared to control condi-tions. Moreover, preliminary experiments withLNCaP and PC3 cells did not show any increasein protein carbonyl levels in zinc-deficient comparedto control conditions (data not shown). Yet theinvolvement of low level or transient bursts of oxi-dants cannot be ruled out.

An alternative mechanism for zinc-dependentloss of paclitaxel e!cacy could involve changesin signal transduction caused by decreased levelsof extracellular zinc. Hershfinkel, Moran, and col-leagues have reported that at least some cell typescontain membrane-bound receptors that senseextracellular zinc levels and couple with intracellu-lar signal transduction pathways without necessi-tating a concomitant rise or fall in intracellularzinc levels [46]. In their studies, zinc deficiencyconditions triggered an extracellular zinc receptorresulting in activation of the mitogen-activatedprotein kinase (MAPK) pathway; these findingssupport previous work showing increased MAPKsignaling with exposure to reduced zinc levels[41]. MAPKs phosphorylate a variety of targetproteins in the cell, including microtubule-associ-ated proteins (MAPs). MAPs bind to the microtu-bule network and serve to regulate microtubuledynamics [47]. Phosphorylation of MAPs resultsin dissociation from the microtubule networkand thereby increased microtubule dynamics [48].Increased microtubule dynamics can reduce theability of drugs like paclitaxel to bind microtu-

bules, thereby reducing paclitaxel e!cacy [49].Currently, no studies have reported the presenceof the extracellular zinc receptor in LNCaP orPC3 cells, so the plausibility of this mechanismis unclear at this time.

This study shows that some prostate cancer cellscultured in zinc-deficient conditions demonstratereduced sensitivity to the chemotherapeutic paclit-axel. There are no data as to whether zinc statushas any consequence on the e!cacy of paclitaxelor other taxane compounds in individuals beingtreated for prostate cancer. If zinc availability doesattenuate paclitaxel activity in vivo, this would haveimportant clinical implications since zinc is a com-mon micronutrient deficiency in the United States.Adjusting zinc intake by dietary means or modestzinc supplementation might prove to be a safe andinexpensive approach to improving success withpaclitaxel. This would also highlight the importanceof the nutritional context for evaluation of thera-peutic e!cacy, an issue that in our assessment isgreatly underappreciated.

Acknowledgements

The authors are indebted to Dr. Bruce N. Amesfor his suggestions and support, Dr. Mark Shigena-ga for providing analysis of endotoxin, and AnureetTiwana and Lily Lou for their technical assistance.Financial support from Department of DefenseProstate Cancer Exploratory Grant PC041140 toB.N.A and D.W.K.

Appendix A. Supplementary data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.canlet.2007.08.010.

References

[1] American Cancer Society, Cancer facts & figures 2007,American Cancer Society, Atlanta (GA), 2007.

[2] P.B. Schi", J. Fant, S.B. Horowitz, Promotion of microtu-bule assembly in vitro by Taxol, Nature 277 (1979) 665–667.

[3] K.H. Downing, Structural basis for the action of drugs thata"ect microtubule dynamics, Emer. Ther. Targ. 4 (2000)219–237.

[4] H. Xiao, P. Verdier-Pinard, N. Fernandez-Fuentes, B. Burd,et al., Insights into the mechanism of microtubule stabiliza-tion by Taxol, Proc. Natl. Acad. Sci. USA 103 (2006) 10166–10173.

[5] M.A. Jordan, L. Wilson, Microtubules as a target foranticancer drugs, Nat. Rev. 4 (2004) 253–265.

A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79 77

[6] D.J. Vaughn, A.W. Brown Jr., W.G. Harker, et al., Mul-ticenter phase II study of estramustine phosphate plusweekly paclitaxel in patients with androgen-independentprostate carcinoma, Cancer 100 (2004) 746–750.

[7] A. Athanasiadis, D. Tsavdaridis, S.K. Rigatos, I. Athana-siadis, N. Pergantas, G.P. Stathopoulos, Hormone refrac-tory advanced prostate cancer treated with estramustine andpaclitaxel combination, Anticancer Res. 23 (2003) 3085–3088.

[8] R.G. Uzzo, P. Leavis, W. Hatch, et al., Zinc inhibitsnuclear factor-kappaB activation and sensitizes prostatecancer cells to cytotoxic agents, Clin. Cancer Res. 8 (2002)3579–3583.

[9] A. Moshfegh, J. Goldman, L. Cleveland, What we eat inAmerica, NHANES 2001–2002: usual nutrient intakes fromfood compared to dietary reference intakes, U.S. Depart-ment of Agriculture, Agricultural Research Service, Wash-ington, DC, 2005.

[10] A.R. Kristal, J.L. Stanford, J.H. Cohen, K. Wicklund, R.E.Patterson, Vitamin and mineral supplement use is associatedwith reduced risk of prostate cancer, Cancer Epidemiol.Biomarkers Prev. 8 (1999) 887–892.

[11] B.N. Ames, P. Wakimoto, Are vitamin and mineral defi-ciencies a major cancer risk?, Nat Rev. Cancer 2 (2002) 694–704.

[12] A.S. Prasad, O. Kucuk, Zinc in cancer prevention, CancerMetastasis Rev. 2 (2002) 291–295.

[13] E. Ho, Zinc deficiency, DNA damage and cancer risk, J.Nutr. Biochem. 15 (2004) 572–578.

[14] M.F. Leitzmann, M.J. Stampfer, K. Wu, G.A. Colditz, W.C.Willett, E.L. Giovannucci, Zinc supplement use and risk ofprostate cancer, J. Natl. Cancer Inst. 95 (2003) 1004–1007.

[15] D.W. Killilea, H. Atamna, C. Liao, B.N. Ames, Ironaccumulation during cellular senescence in human fibroblastsin vitro, Antioxid. Redox Signal. 5 (2003) 507–516.

[16] A. Krishan, Rapid flow cytofluorometric analysis of cellcycle by propidium iodide staining, J. Cell Biol. 66 (1975)188–193.

[17] C. Perez-Stable, 2-Methoxyestradiol and paclitaxel havesimilar e"ects on the cell cycle and induction of apoptosis inprostate cancer cells, Cancer Lett. 231 (2006) 49–64.

[18] Dietary Reference Intakes for Vitamin A, Vitamin K,Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manga-nese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc.Standing Committee on the Scientific Evaluation of DietaryReference Intakes, Food and Nutrition Board, Institute ofmedicine, National Academy Press, Washington, DC, 2001,pp. 442–501.

[19] P. Wakimoto, G. Block, Dietary intake, dietary patterns,and changes with age: an epidemiological perspective, J.Gerontol. A, Biol. Sci. Med. Sci. 56 (2001) 65–80.

[20] B.N. Ames, Low micronutrient intake may accelerate thedegenerative diseases of aging through allocation of scarcemicronutrients by triage, Proc. Natl. Acad. Sci. USA 103(2006) 17589–17594.

[21] P. Boyle, G. Severi, G.G. Giles, The epidemiology ofprostate cancer, Urol. Clin. North Am. 30 (2003) 209–217.

[22] L.C. Costello, R.B. Franklin, Novel role of zinc in theregulation of prostate citrate metabolism and its implicationsin prostate cancer, Prostate 35 (1998) 285–296.

[23] L. Huang, C.P. Kirschke, Y. Zhang, Decreased intracellularzinc in human tumorigenic prostate epithelial cells: a possible

role in prostate cancer progression, Cancer Cell Int. 6 (2006)10.

[24] R.B. Pilz, R.C. Willis, J.E. Seegmiller, Regulation of humanlymphoblast plasma membrane 5’-nucleotidase by zinc, J.Biol. Chem. 257 (1982) 13544–13549.

[25] G.G. Mackenzie, C.L. Keen, P.I. Oteiza, Zinc status ofhuman IMR-32 neuroblastoma cells influences their suscep-tibility to iron-induced oxidative stress, Dev. Neurosci. 24(2002) 125–133.

[26] D.J. Tate Jr., M.V. Miceli, D.A. Newsome, Zinc protectsagainst oxidative damage in cultured human retinal pigmentepithelial cells, Free Radic. Biol. Med. 26 (1999) 704–713.

[27] E. Ho, B.N. Ames, Low intracellular zinc induces oxidativeDNA damage, disrupts p53, NFkappa B, and AP1 DNAbinding, and a"ects DNA repair in a rat glioma cell line,Proc. Natl. Acad. Sci. USA 99 (2002) 16770–16775.

[28] E. Ho, C. Courtemanche, B.N. Ames, Zinc deficiencyinduces oxidative DNA damage and increases p53 expressionin human lung fibroblasts, J. Nutr. 133 (2003) 2543–2548.

[29] C.T. Walsh, H.H. Sandstead, A.S. Prasad, P.M. Newberne,P.J. Fraker, Zinc: health e"ects and research priorities for the1990s, Environ. Health Perspect. 102 (1994) 5–46.

[30] R.G. Uzzo, P.L. Crispen, K. Golovine, P. Makhov, E.M.Horwitz, V.M. Kolenko, Diverse e"ects of zinc on NF-kappaB and AP-1 transcription factors: implications forprostate cancer progression, Carcinogenesis 27 (2006) 1980–1990.

[31] J.Y. Liang, Y.Y. Liu, J. Zou, R.B. Franklin, L.C. Costello,P. Feng, Inhibitory e"ect of zinc on human prostaticcarcinoma cell growth, Prostate 40 (1999) 200–207.

[32] X.Z. Wang, J.R. Beebe, L. Pwiti, A. Bielawska, M.J. Smyth,Aberrant sphingolipid signaling is involved in the resistanceof prostate cancer cell lines to chemotherapy, Cancer Res. 59(1999) 5842–5848.

[33] N.I. Ivanov, S.P. Cowell, P. Brown, P.S. Rennie, E.S. Guns,M.E. Cox, Lycopene di"erentially induces quiescence andapoptosis in androgen-responsive and -independent prostatecancer cell lines, Clin. Nutr. 26 (2007) 252–263.

[34] S.H. Zhuang, K.L. Burnstein, Antiproliferative e"ect of1alpha,25-dihydroxyvitamin D3 in human prostate cancercell line LNCaP involves reduction of cyclin-dependentkinase 2 activity and persistent G1 accumulation, Endocri-nology 139 (1998) 1197–1207.

[35] Y. Liu, R.B. Franklin, L.C. Costello, Prolactin and testos-terone regulation of mitochondrial zinc in prostate epithelialcells, Prostate 30 (1997) 26–32.

[36] K. Shimada, M. Nakamura, E. Ishida, M. Kishi, S.Yonehara, N. Konishi, Contributions of mitogen-activatedprotein kinase and nuclear factor kappa B to N-(4 hydroxy-phenyl)retinamide-induced apoptosis in prostate cancer cells,Mol. Carcinog. 35 (2002) 127–137.

[37] P. Feng, J.Y. Liang, T.L. Li, et al., Zinc induces mitochon-dria apoptogenesis in prostate cells, Mol. Urol. 4 (2000) 31–36.

[38] G. Untergasser, H. Rumpold, E. Plas, M. Witkowski, G.Pfister, P. Berger, High levels of zinc ions induce loss ofmitochondrial potential and degradation of antiapoptoticBcl-2 protein in in vitro cultivated human prostateepithelial cells, Biochem. Biophys. Res. Commun. 279(2000) 607–614.

[39] N. Ganju, A. Eastman, Zinc inhibits Bax and Bak activationand cytochrome c release induced by chemical inducers of

78 A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79

apoptosis but not by death-receptor-initiated pathways, CellDeath Di"er. 10 (2003) 652–661.

[40] M. Kondoh, E. Tasaki, M. Takiguchi, M. Higashimoto, Y.Watanabe, M. Sato, Activation of caspase-3 in HL-60 cellstreated with pyrithione and zinc, Biol. Pharm. Bull. 28 (2005)757–759.

[41] M.P. Zago, G.G. Mackenzie, A.M. Adamo, C.L. Keen, P.I.Oteiza, Di"erential modulation of MAP kinases by zincdeficiency in IMR-32 cells: role of H(2)O(2), Antioxid.Redox Signal. 7 (2005) 1773–1782.

[42] J.E. Hesketh, Zinc-stimulated microtubule assembly andevidence for zinc binding to tubulin, Int. J. Biochem. 14(1982) 983–990.

[43] P.I. Oteiza, L.S. Hurley, B. Lonnerdal, C.L. Keen, E"ects ofmarginal zinc deficiency onmicrotubule polymerization in thedeveloping rat brain, Biol. Trace Elem. Res. 24 (1990) 13–23.

[44] P.I. Oteiza, M.S. Clegg, M.P. Zago, C.L. Keen, Zincdeficiency induces oxidative stress and AP-1 activation in3T3 cells, Free Radic. Biol. Med. 28 (2000) 1091–1099.

[45] M.S. Clegg, L.A. Hanna, B.J. Niles, T.Y. Momma, C.L.Keen, Zinc Deficiency-Induced Cell Death, IUBMB Life 57(2005) 661–669.

[46] M. Hershfinkel, A. Moran, N. Grossman, I. Sekler, A zinc-sensing receptor triggers the release of intracellular Ca2+and regulates ion transport, Proc. Natl. Acad. Sci. USA 98(2001) 11749–11754.

[47] E. Mandelkow, E.M. Mandelkow, Microtubules and micro-tubule-associated proteins, Curr. Opin. Cell Biol. 7 (1995)72–81.

[48] S. Illenberger, G. Drewes, B. Trinczek, et al., Phosphoryla-tion of microtubule-associated proteins MAP2 and MAP4by the protein kinase p110mark. Phosphorylation sites andregulation of microtubule dynamics, J. Biol. Chem. 271(1996) 10834–10843.

[49] W.B. Derry, L. Wilson, I.A. Khan, R.F. Luduena, M.A.Jordan, Taxol di"erentially modulates the dynamics ofmicrotubules assembled from unfractionated and purifiedbeta-tubulin isotypes, Biochemistry 36 (1997) 3554–3562.

A.N. Killilea et al. / Cancer Letters 258 (2007) 70–79 79