Key role for p27Kip1, retinoblastoma protein Rb, and MYCN in polyamine

inhibitor-induced G1 cell cycle arrest in MYCN-amplified humanneuroblastoma cells

Christopher J Wallick1, Ivonne Gamper1,4, Mike Thorne1,5, David J Feith2, Kelsie Y Takasaki1,Shannon M Wilson3,6, Jennifer A Seki1, Anthony E Pegg2, Craig V Byus3 andAndre S Bachmann*,1

1Cancer Research Center of Hawaii, University of Hawaii at Manoa, 1236 Lauhala Street, Honolulu, HI 96813, USA; 2Departmentof Cellular and Molecular Physiology, Pennsylvania State University College of Medicine, Hershey, PA 17033, USA; 3Departmentof Biochemistry, University of California, Riverside, Riverside, CA 92521, USA

Alpha-difluoromethylornithine (DFMO) inhibits the proto-oncogene ornithine decarboxylase (ODC) and is knownto induce cell cycle arrest. However, the effect of DFMOon human neuroblastoma (NB) cells and the exactmechanism of DFMO-induced cell death are largelyunknown. Treatment with DFMO in combinationwith SAM486A, an S-adenosylmethionine decarboxy-lase (AdoMetDC) inhibitor, has been shown to enhancepolyamine pool depletion. Therefore, we analysed themechanism of action of DFMO and/or SAM486A intwo established MYCN-amplified human NB cell lines.DFMO and SAM486A caused rapid cell growth inhibi-tion, polyamine depletion, and G1 cell cycle arrest withoutapoptosis in cell lines LAN-1 and NMB-7. These effectswere enhanced with combined inhibitors and largelyprevented by cotreatment with exogenous polyamines.The G1 cell cycle arrest was concomitant with an increasein cyclin-dependent kinase inhibitor p27Kip1. In a similarfashion, DFMO and DFMO/SAM486A inhibited thephosphorylation of the G1/S transition-regulating retino-blastoma protein Rb at residues Ser795 and Ser807/811.Moreover, we observed a dramatic decrease in MYCNprotein levels. Overexpression of MYCN induces anaggressive NB phenotype with malignant behavior. Weshow for the first time that DFMO and SAM486A induceG1 cell cycle arrest in NB cells through p27Kip1 and Rbhypophosphorylation.Oncogene (2005) 24, 5606–5618. doi:10.1038/sj.onc.1208808;published online 20 June 2005

Keywords: DFMO; SAM486A;MYCN-amplified neuro-blastoma; p27Kip1; Rb; cell cycle

Introduction

Many studies over the years have demonstrated a rolefor polyamines in tumor cell growth, and the associatedincrease in biosynthetic enzyme activities (e.g. ornithinedecarboxylase (ODC) and S-adenosylmethionine de-carboxylase (AdoMetDC)) in tumor tissues relative tosurrounding normal tissues renders a rationale fortargeted drug intervention (Pegg, 1988; McCann andPegg, 1992; Davidson et al., 1999; Meyskens andGerner, 1999; Levin et al., 2000, 2003; Takahashiet al., 2000; Bachmann, 2004).The proto-oncogene ODC is a key enzyme in

polyamine biosynthesis and catalyses the conversionof ornithine (Orn) to putrescine (Put). The latter isfurther converted to the higher polyamines spermidine(Spd) and spermine (Spm), all of which have beenimplicated in tumor cell growth. A second rate-limitingenzyme in polyamine biosynthesis is AdoMetDC, whichprovides the aminopropyl donor decarboxylatedS-adenosylmethionine and is required for the sequentialconversions of Put to Spd and Spm. Specific inhibitorsfor both ODC and AdoMetDC are available andhave been used in polyamine-related studies (Supple-mentary information). Alpha-difluoromethylornithine(DFMO, also known as Eflornithine) (Metcalf et al.,1978) is a synthetic suicide inhibitor of ODC, whichhas been intensively evaluated in human clinical trialsas an anticancer and chemopreventive agent (Porteret al., 1992; Meyskens and Gerner, 1999; Levinet al., 2000, 2003; Carbone et al., 2001; Fabian et al.,2002). SAM486A (also known as CGP48664),a derivative of the first-generation AdoMetDC inhibitormitoguazone (MGBG), exerts potent and specificinhibition of AdoMetDC (Regenass et al., 1994;Svensson et al., 1997). Its efficacy has been assessedin various cancer cells and animal systems (Regenasset al., 1994; Dorhout et al., 1995; Svensson et al.,1997), and has been evaluated in Phase I and Phase IIhuman clinical trials (Eskens et al., 2000; Paridaenset al., 2000; Siu et al., 2002; Pless et al., 2004; van Zuylenet al., 2004).

Received 6 February 2005; revised 29 April 2005; accepted 29 April 2005;published online 20 June 2005

*Correspondence: AS Bachmann;E-mail: abachmann@crch.hawaii.edu4Current address: Institute for Biomedical Technology, Universitatsk-linikum Aachen, RWTH, 52057 Aachen, Germany5Current address: Hawaii Biotech, Inc., Aiea, HI 96701, USA6Current address: University of California, San Francisco, LungBiology Center, San Francisco, CA 94110, USA

Oncogene (2005) 24, 5606–5618& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00

www.nature.com/onc

Neuroblastoma (NB) is the most common extra-cranial tumor of childhood, with approximately 700 newcases per year in the United States. While many infantsexperience complete regression of primary tumors andeven metastatic disease, older children are oftenconfronted with NB metastases that are aggressive andrespond poorly to even the most intense multi-compo-nent drug regimens. MYCN proto-oncogene amplifica-tion occurs in approximately 20% of primary NBtumors, and is strongly associated with advanced stagedisease, rapid tumor progression, and poor prognosis(Brodeur et al., 1984; Seeger et al., 1985; Brodeur, 2003).While DFMO has been shown to induce growthinhibition and differentiation in NB cells (Chapman,1980; Melino et al., 1988; Melino et al., 1991), no relatedNB studies exist with SAM486A alone or combinedwith DFMO. To our knowledge, these two inhibitorshave never been considered for human NB clinical trials.This study was designed (1) to explore the therapeutic

potential of DFMO and SAM486A in human NB cells,and (2) to examine the underlying molecular mechan-isms by which these inhibitors suppress cell growth. Weinitially showed that several NB cell lines are inhibited ina dose-dependent manner. Of those, we selected two celllines (LAN-1 and NMB-7) for further analyses, both ofwhich are MYCN-amplified and mutated in the tumorsuppressor protein p53 (Davidoff et al., 1992; Goldmanet al., 1996; Tweddle et al., 2003). Our findings showedthat both inhibitors alone, or in combination, depletedpolyamine pools to various degrees and caused severegrowth inhibition associated with G1 cell cycle arrest asearly as 2 days after treatment. Significant morphologi-cal changes in the actin cytoskeleton, an increase incyclin-dependent kinase inhibitor p27Kip1, hypophos-phorylation of Rb, and downregulation of MYCNaccompanied these effects. The inhibitor-induced G1arrest in LAN-1 and NMB-7 cells occurred in theabsence of p53, since the tested cell lines are p53deficient. The fact that DFMO and SAM486A effec-tively and rapidly inhibited the growth of the MYCN-amplified LAN-1 and NMB-7 cells provides a strongrationale for their therapeutic use in patients withprogressive or relapsed NB tumors, which often showMYCN amplification and p53-associated chemoresis-tance (Tweddle et al., 2001, 2003) with poor prognosticoutcome.

Results

Expression of ODC, AdoMetDC, and MYCN in NB cells

To characterize the NB cell lines LAN-1 and NMB-7,we verified the presence of ODC, AdoMetDC, andMYCN at the RNA level using reverse-transcriptase–polymerase chain reaction (RT–PCR). Total RNA wasisolated from subconfluent monolayer cell cultures andequal amounts of RNA were reverse-transcribed intocDNA and then analysed by PCR using specific DNAprimers. ODC, AdoMetDC, and MYCN mRNAs wereexpressed in both cell lines, as indicated by the PCR

products of expected sizes at 400, 450, and 220 bp,respectively (Figure 1a). These results confirm thatLAN-1 and NMB-7 cells produce mRNA for bothtarget enzymes and that both cell lines are of theMYCN-amplified NB subtype.To determine the presence of ODC, AdoMetDC, and

MYCN at the protein level, cell lysates containing equalamounts of protein were analysed by Western blotting.ODC (53 kDa) and processed AdoMetDC a chain(31 kDa) proteins were detected at the expected size inLAN-1 and NMB-7 cells (Figure 1b, left and middlepanels), using an ODC-specific antibody and anAdoMetDC-recognizing rabbit serum, respectively. Asthe latter interacted with several other proteins, weverified the presence of AdoMetDC in LAN-1 cellstreated with 5mM DFMO and 10 mM SAM486A. Thetreatment of cells with both polyamine inhibitors for 3days strongly induced AdoMetDC protein expression,as indicated by an increase of the 31 kDa band, while theintensity of a second band at 75 kDa did not change(Figure 1c). The MYCN protein (67 kDa) was alsostrongly expressed in LAN-1 and NMB-7 cells(Figure 1b, right panel).The presence and functionality of ODC and Ado-

MetDC was further demonstrated by measuring theirrespective enzyme activities in lysates of LAN-1 andNMB-7 cells harvested at early (day 1), mid (day 2), andlate (day 3) log phase. As shown in Figure 1d (ODC)and Figure 1e (AdoMetDC), enzyme activities weredetected in both cell lines. Specific activities were highestduring the early-to-mid log growth phase, with amarked decline in activity on day 3. Protein concentra-tions from cell lysates on days 1, 2, and 3 indicatedlogarithmic cell growth (not shown). These resultstogether clearly show that (1) mRNAs for ODC,AdoMetDC, and MYCN are present in LAN-1 andNMB-7 cells, (2) mRNAs are properly translated intoproteins, and (3) ODC and AdoMetDC are functionallyactive enzymes.

Effects of DFMO and SAM486A on ODC andAdoMetDC activities and polyamine pools

Next, we determined the effects of DFMO andSAM486A on enzymatic activities of ODC and Ado-MetDC in inhibitor-treated cells. Cell lines LAN-1 andNMB-7 were left untreated (control) or treated with5mM DFMO, 10 mM SAM486A, or the combination ofboth inhibitors for 2 days (mid-log phase) and enzymeactivities were measured in cell lysates. In comparisonwith untreated cells, the ODC enzyme activity was lowerafter DFMO treatment and was further decreased bycombined inhibitor treatment, while SAM486A aloneincreased ODC activity (Table 1). Preincubation ofextracts from untreated or DFMO-treated cells with100 mM DFMO resulted in a complete loss of ODCactivity (data not shown), indicating that the residualODC activity represents a new steady state in thepresence of the inhibitor rather than DFMO-resistantODC protein. Under identical experimental conditions,AdoMetDC activity was determined. In contrast to

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5607

Oncogene

ODC, we found that treatment with DFMO,SAM486A, or the combination of both inhibitorsincreased AdoMetDC activity (Table 1). SAM486Aacts as a competitive inhibitor of AdoMetDC in the celland is capable of stabilizing the enzyme (Svensson et al.,1997), as shown in Figure 1c. Therefore, AdoMetDCactivity is elevated in the in vitro assay of SAM486A-treated cells because dilution of the cellular contents toprepare the cytosolic lysate reduces the concentrationof SAM486A below the level necessary for inhibitionand reveals the stabilization of the enzyme. Sequentialdilution and concentration of cell lysates to remove theresidual inhibitor further increased the measured Ado-MetDC activity nearly 100-fold in SAM486A-treatedcells (not shown). This contrasts with the irreversibleinhibition of ODC by the suicide inhibitor DFMO,which is readily apparent in the in vitro assay. Thefurther increase in AdoMetDC activity observed in cells

020406080

100120140

0 2Time (d)

OD

C s

peci

fic

activ

ity

(pm

ol C

O2/3

0 m

in/m

g pr

ot)

LAN-1

NMB-7

0

20

40

60

80

100

120

140

Ado

Met

DC

spe

cifi

c ac

tivity

(p

mol

CO

2/30

min

/mg

prot

)

LAN-1

NMB-7L

AN

-1

β -actin

MYCN

NM

B-7

220

400

ODC 400

400

LA

N-1

NM

B-7

β-actin β-actin

bp

450 AdoMetDC

LA

N-1

NM

B-7

400

bp bp

LA

N-1

NM

B-7

ODC

GAPDH

LA

N-1

NM

B-7

AdoMetDC

GAPDH

MYCN

LA

N-1

NM

B-7

GAPDH

GAPDH

AdoMetDC

kDa - DFMO/SAM486A

LAN-1

31

75

+

1 3 0 2Time (d)1 3

a

b

c d e

Figure 1 Expression and activity studies of ODC, AdoMetDC, and MYCN in NB cells. (a) Total RNA was isolated from MYCN-amplified cell lines LAN-1 and NMB-7 and gene-specific primers for ODC, AdoMetDC, andMYCN were used in RT–PCR analyses.b-Actin was included as a normalization standard. The data are representative of similar results from three independent experiments.(b) Equal amounts of total protein from LAN-1 and NMB-7 lysates were analysed by SDS–PAGE and Western blotting and probedfor ODC, AdoMetDC, and MYCN. GAPDH was used as loading control. (c) Lysates of untreated or DFMO/SAM486A-treatedLAN-1 cells were probed for AdoMetDC. A nearly identical result was obtained with a different rabbit serum against AdoMetDC.Enzyme activities of ODC (d) and AdoMetDC (e) were determined in LAN-1 and NMB-7 cell lysates in a time-course study. Cells(1� 105 cells/ml) were seeded in 10 cm culture dishes and harvested after 1, 2, and 3 days. Specific enzyme activities are expressed aspmol CO2/30min/mg protein. Each sample was measured in duplicate and data represent the mean of three independentexperiments7s.d.

Table 1 Effects of polyamine inhibitors on ODC and AdoMetDCactivities in human neuroblastoma cells

Cellline

Treatmenta Specific enzyme activities (% of control)b

ODC AdoMetDC

LAN-1 Control 10078 100712DFMO 2672 689794SAM486A 240727 16979DFMO/SAM486A 2272 490746

NMB-7 Control 100711 100721DFMO 2578 582767SAM486A 251742 200716DFMO/SAM486A 2371 552764

aDFMO¼ 5mM; SAM486A¼ 10mM. Cells were analysed 2 days aftertreatments. bValues are presented as percentage (%) of specific enzymeactivities (pmol CO2/30min/mg protein) relative to untreated controlcells. Each sample was measured in duplicate and data represent themean of three independent experiments7s.d.

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5608

Oncogene

treated with DFMO is likely because the treatment ofcells with DFMO is known to induce AdoMetDC.To study the effects of inhibitors on intracellular

polyamine pools, LAN-1 and NMB-7 cells were treatedwith DFMO, SAM486A, or both inhibitors combined,for 2 and 3 days (Table 2). Inhibition of ODC byDFMO decreased Put and Spd pools in both cell lines,while Spm pools decreased only slightly. Inhibition ofAdoMetDC with SAM486A dramatically increased Putlevels in both cell lines, indicating an efficient blockadein the polyamine biosynthetic pathway between Put andSpd, and reduced Spd and Spm levels. Combinedinhibition of both enzymes decreased polyamine poolsto various degrees, and exogenous supplementation ofSpd (10 mM) to cell cultures led to a substantial increasein Spd. These observations are comparable with trendsseen in other cell types (Kramer et al., 1989, 2001;Svensson et al., 1997). In untreated control cells, thelevels of Put and Spd decreased during exponential cellgrowth (day 0 to day 3), which is also consistent witha decrease in ODC and AdoMetDC enzyme activities(Figure 1d and e). Spm levels decreased in untreatedLAN-1 cells and remained the same in NMB-7 cells(Table 2).

Effects of DFMO and SAM486A on cell growthand morphology

To determine the effects of DFMO and SAM486A onthe growth of NB cells, LAN-1 and NMB-7 cells wereexposed to 5mM DFMO, 10 mM SAM486A, or thecombination of both inhibitors for 3 days, and the cellnumber was determined. As depicted in Figure 2a,LAN-1 and NMB-7 cell growth slowed substantially,and the combination of both inhibitors led to near-totalcessation of cell growth after 3 days. Exogenoussupplementation of cell cultures with Spd (10 mM) clearlyalleviated the growth-inhibitory effects, thus indicatingthat the inhibition of cell growth is due to polyaminepool depletion. Identical treatment of cells with 0.5mMDFMO and 1 mM SAM486A (alone, and in combina-tions) resulted in dose-dependent growth inhibition

(not shown). Moreover, we showed that DFMO andSAM486A inhibited the growth of several other NB celllines including SK-N-SH, SH-SY5Y (both MYCN-nonamplified and p53 wild type), LAN-5, and IMR-32(both MYCN-amplified and p53 wild type) in a dose-dependent manner (not shown).The observed growth inhibition of LAN-1 and NMB-

7 cells was accompanied by dramatic alterations in cellmorphology (Figure 2b). After 3 days of treatment with5mM DFMO, the triangular-shaped LAN-1 cellschanged into spindle-like cells with neurite-like exten-sions. Under identical treatment, NMB-7 cells formedlarge polynucleate cell aggregates with only very shortextensions. Treatment of these two cell lines with 10 mMSAM486A resulted in similar cell aggregates, but in theabsence of neurite-like extensions. The combination ofboth inhibitors led to a mixed-type cell morphology, andsupplementation of these cells with Spd (10 mM) at thebeginning of the experiment alleviated the inhibitoreffects, producing normal, triangular-shaped cells com-parable to untreated control cells.To further study these morphological changes, LAN-

1 and NMB-7 cells were treated with 5mM DFMO,10 mM SAM486A, or the combination of both inhibitors,and the actin cytoskeleton was visualized using phalloi-din. Concomitant staining with 40,6-diamidino-2-pheny-lindole (DAPI) allowed the visualization of cell nuclei.As shown in Figure 3, the treatment with DFMO and/orSAM486A clearly disturbed the actin cytoskeleton inboth cell lines. Furthermore, the DAPI-stained nucleireveal the presence of polynucleate aggregates in bothcell lines, which confirms that the large cell aggregatesseen in Figure 2b are indeed composites of several cells.In summary, these data show that in NB cells, DFMO

and SAM486A cause rapid growth inhibition andinduce morphological changes within 3 days. Similareffects were already apparent after 1–2 days (notshown). By comparison, the treatment of humanmelanoma cells (MALME-3) with polyamine inhibitorsrequired more extensive incubation times (8–12 days)and led to less obvious rearrangements in cell structures(Kramer et al., 2001). Furthermore, human ovarian

Table 2 Effects of polyamine inhibitors on polyamine pools in human neuroblastoma cells

Cell line Treatmenta Polyamine pools (nmol/mg total protein)b

Day 0 Day 2 Day 3

Put Spd Spm Put Spd Spm Put Spd Spm

LAN-1 Control 1.2 11.9 15.3 0.6 9.4 11.8 0.1 7.1 9.4DFMO 0.3 2.2 8.7 0.0 1.9 8.1SAM486A 48.5 5.5 4.2 43.7 5.0 3.5DFMO/SAM486A 0.3 2.6 9.0 0.1 1.4 6.9DFMO/SAM486A/Spd 0.0 21.1 4.2 0.4 16.9 3.5

NMB-7 Control 5.6 13.9 11.2 2.6 12.7 11.2 0.8 8.4 12.6DFMO 0.1 1.9 9.3 0.0 1.4 8.6SAM486A 69.6 5.6 3.9 259.5 5.5 3.6DFMO/SAM486A 0.7 5.1 7.4 0.6 1.9 9.8DFMO/SAM486A/Spd 0.1 22.6 6.9 0.1 40.6 3.4

aDFMO¼ 5mM; SAM486A¼ 10mM; Spd¼ 10mM. bData represent the mean values of two independent experiments

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5609

Oncogene

1

2

3

4

5

Control

DFMO

SAM486A

DFMO +

SAM486A

DFMO +

SAM486A +

Spd

0

20

40

60

80

100

120

140

1 2 5%

Cel

l Gro

wth

0

20

40

60

80

100

120

140

% C

ell G

row

th

LAN-1 NMB-7

LAN-1 NMB-7

43 1 2 543

a

b

Figure 2 Effects of polyamine inhibitors DFMO and SAM486A on the growth of human NB cells. (a) LAN-1 and NMB-7 cells weregrown in the absence (1) or presence of 5mM DFMO (2), 10mM SAM486A (3), 5mM DFMO plus 10mM SAM486A (4), and 5mMDFMO plus 10 mM SAM486A, supplemented with 10mM Spd (5) for 3 days. Cell numbers are expressed as percentage (%) of controlcell growth. Each sample was measured in duplicate and the data represent the mean of three separate determinations7s.d. (b)Representative light micrographs of LAN-1 and NMB-7 cells corresponding to (a). Exogenous supplementation with 10mM Spdrestored cell growth and the triangular-shaped cell morphology of LAN-1 and NMB-7 cells. Light micrographs were routinely taken todocument most of the experiments throughout this study and the pictures are representative of at least ten (controls and inhibitors) orsix (inhibitors plus Spd) separate experiments

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5610

Oncogene

cancer cells (SKOV-3), treated with DFMO orSAM486A, were not inhibited in their growth andshowed no visible changes in cell morphology (Bach-mann, 2004).

Prolonged cell growth inhibition after combined treatmentwith DFMO and SAM486A

To examine whether the combined treatment withDFMO and SAM486A produces NB cell populationswhich lose their proliferative capacity permanently,LAN-1 and NMB-7 cells were either left untreated(control) or treated with 5mM DFMO plus 10 mMSAM486A, and cell numbers were determined after 3days (Figure 4). As expected, the growth of cells treatedwith both inhibitors was strongly suppressed. On day 3of the experiment, cells were washed free of inhibitorswith PBS, suspended in new medium (NM) withoutinhibitors, and seeded again at the initial cell density.While the untreated LAN-1 and NMB-7 cells grewnormally and reached similar cell densities after anadditional 3 days (day 6 of the experiment), the initiallytreated LAN-1 and NMB-7 cells resumed proliferationonly slowly and reached about 30 and 58% of control

cell growth, respectively, on day 10 of the experiment(Figure 4). This result indicates that the combinedpolyamine inhibitor treatment leads to sustained growtharrest in both MYCN-amplified NB cell lines as, evenafter removal of inhibitors, their proliferative capacity isslowly and only partially restored. Similar experimentsusing DFMO and SAM486A individually revealed thatthe sustained growth arrest was primarily due to theaction of DFMO and not SAM486A (not shown).

Effects of DFMO and SAM486A on cell cycledistribution

Depletion of polyamines has previously been shown tocause G1 cell cycle arrest in cancer cells (Ray et al., 1999;Kramer et al., 2001; Singh et al., 2001). To investigatewhether the observed growth inhibition of NB cells isdue to the interference with cell cycle kinetics, LAN-1and NMB-7 cells were treated with 5mM DFMO, 10 mMSAM486A, or the combination of both inhibitors,harvested, and stained with propidium iodide (PI) ondays 2 and 3 of the experiment. As shown in Table 3,both inhibitors individually produced changes in the cellcycle distribution at days 2 and 3 of the experiment,

Figure 3 Representative fluorescence micrographs of the actin cytoskeleton and nuclei of human NB cells. LAN-1 and NMB-7 cells inthe absence or presence of 5mM DFMO, 10mM SAM486A or the combination of both inhibitors, are shown 3 days after treatment.The actin cytoskeleton (red) was visualized with Texas Red-X phalloidin and cell nuclei (blue) were stained using DAPI. Strongcytoskeletal changes are visible after treatment with polyamine inhibitors

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5611

Oncogene

as indicated by the accumulation of cells in G1 and areduction of cells in S phase and in G2/M phase. Thisprofile, which is typical for cell cycle arrest at G1, wasenhanced by the combination of both inhibitors, withthe majority of cells in G1 (B71–73%), a reducednumber of cells in S phase (B20–28%), and a smallportion in G2/M (B2–7%). Cotreatment with both inhi-bitors plus Spd (10 mM) prevented G1 arrest (Table 3),which is in agreement with the growth inhibition datashown in Figure 2. The polyamine depletion-induced G1cell cycle arrest did not involve apoptosis, as judged by(i) the absence of typical morphological changes suchas nuclear condensation, membrane blebbing, and lossof cell adherence (Figures 2 and 3), (ii) the absence of

PARP cleavage (not shown), and (iii) the lack of anapoptosis-associated G1 subpeak (Table 3). Similarobservations were previously reported using normalrat intestinal epithelial cells (IEC-6) (Ray et al., 1999).

Strong and rapid changes in cell cycle regulatory proteinsp27Kip1, Rb, and MYCN

To this end, our data showed that the treatment ofLAN-1 and NMB-7 cells with DFMO and/orSAM486A inhibits polyamine-dependent cell growthand causes G1 cell cycle arrest, and these effects couldbe largely prevented by the addition of Spd. To furtherinvestigate the mechanism that regulates G1 arrest inLAN-1 and NMB-7 cells, and to define the changes inassociated protein responses, we studied the cell cycle-related proteins p21Waf1/Cip1, p27Kip1, p53, Rb, andMYCN in NB cell lysates. We found that LAN-1and NMB-7 cells treated with 5mM DFMO increasedendogenous levels of cdk inhibitor protein p27Kip1

relative to untreated control cells after 2 days oftreatment. This effect was enhanced on day 3 of theexperiment and was further increased in the presence ofboth inhibitors (Figure 5a). By comparison, treatmentwith 10 mM SAM486A alone did not increase p27Kip1

levels. The result suggests that p27Kip1 plays a role in theobserved G1 cell cycle arrest in NB cells (Table 3), andthat this effect is primarily due to the action of DFMO,and not SAM486A. To test for the presence of the cdkinhibitor protein p21Waf1/Cip1, identical experiments wereconducted, but the protein could not be detected (notshown). Low levels of p21Waf1/Cip1 in NB cells have beenpreviously described (McKenzie et al., 2003). Similarly,the tumor-suppressor protein p53 was not detected (notshown), which is in agreement with previous findingsthat cell lines LAN-1 and NMB-7 are p53 deficient. Incontrast, p53 was readily expressed in identical Westernblot control experiments and detected by immunofluore-scence using the p53 wild-type cell line SK-N-SH (notshown).The retinoblastoma protein Rb can be regulated

through the action of p27Kip1. Rb phosphorylationreduces the levels of functional Rb/E2F repressorcomplexes and activates E2F-responsive promotersimportant for moving the cell through the G1/S phaseof the cell cycle. To study the effects on Rb phosphory-lation, LAN-1 and NMB-7 cells were treated withDFMO and/or SAM486A and analysed on days 1, 2,and 3 using two phospho-specific Rb (p-Rb) antibodiesas well as an antibody which recognizes total Rbprotein. As shown in Figure 5b, Rb was phosphorylatedat residue Ser795 and residues Ser807/811 in untreatedLAN-1 and NMB-7 cells, while the treatment withDFMO or the combination of DFMO and SAM486Aresulted in strong hypophosphorylation of Rb on days 2and 3 of the experiment. The effect of SAM486A alonewas much less pronounced. Rb hypophosphorylationwas further confirmed with total Rb showing a visiblemobility shift from a hyperphosphorylated form (upperband) to a hypophosphorylated form (lower band). Thetreatment with both inhibitors for 3 days completely

0

10

20

30

40

50

0 2 5 6 7 10 11Time (d)

Via

ble

Cel

l Num

ber

x 10

4 /ml

NM

0

10

20

30

40

50

60

70

80

Via

ble

Cel

l Num

ber

x 10

4 /ml

NM

1 43 98

0 2 5 6 7 10 11Time (d)

1 43 98

b

Figure 4 Prolonged cell growth inhibition of inhibitor-treatedhuman NB cells. Growth of cell lines (a) LAN-1 and (b) NMB-7 inthe absence (&) or the presence (�) of 5mM DFMO plus 10 mMSAM486A. On day 3, both untreated and treated cells were washedtwice with PBS, reseeded in NM (solid black arrow) withoutinhibitors at the starting cell density (dashed black arrow), and theexperiment was continued for an additional 7 days. Cell numberswere determined using a hemocytometer and trypan blue reagent,and each data point is representative of three separate determina-tions7s.d.

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5612

Oncogene

dephosphorylated Rb and also decreased the total levelsof Rb (Figure 5b). Hypophosphorylation of Rb is aknown consequence of p27Kip1 induction, and this is instrong agreement with our findings. In summary, theresults suggest that polyamine inhibitor-induced G1 cellcycle arrest inMYCN-amplified NB cells is mediated byan increase in p27Kip1 and subsequent hypophosphory-lation of Rb.Downregulation of MYCN expression either by

antisense treatment or by retinoic acid (RA) has beenshown to inhibit cell growth and/or induce neuronaldifferentiation of NB cells (Matsuo and Thiele, 1998;Galderisi et al., 1999, 2003; Lu et al., 2003). Therefore,we were curious to see whether DFMO and SAM486Aalso affect MYCN protein levels in LAN-1 and NMB-7

cells. As shown in Figure 5c, DFMO and both inhibitorsin combination, and to a lesser extent SAM486A alone,decreased MYCN levels in both cell lines as early asday 1 and, most profoundly, on day 3 of the experiment.The quantification of data showed that both drugscombined reduced MYCN levels to 6% and 5% ofcontrols in LAN-1 and NMB-7 cells, respectively.

Discussion

In the present paper, we investigated for the firsttime the effects of the clinically well-established poly-amine inhibitors DFMO and SAM486A in twoMYCN-amplified NB cell lines. We found that both inhibitors

Table 3 Effects of polyamine inhibitors on cell cycle phase distribution in human neuroblastoma cells

Cell line Treatmenta Time (d) Cell cycle (% Total cells)b Histogram

G1 S G2/M

LAN-1 Control 0 50.075.1 38.178.0 11.974.3Control 2 47.972.8 45.872.7 6.370.9 aDFMO 2 62.1710.4 26.477.5 11.573.0SAM486A 2 58.778.3 31.776.8 9.673.2DFMO/SAM486A 2 73.475.6 19.574.9 7.172.5 bDFMO/SAM486A/Spd 2 45.674.8 45.974.1 8.570.8 cControl 3 52.073.6 37.774.5 10.472.0DFMO 3 63.1712.2 30.878.2 6.274.1SAM486A 3 53.375.1 37.473.5 9.372.0DFMO/SAM486A 3 70.579.6 25.576.0 4.173.5DFMO/SAM486A/Spd 3 56.775.9 34.375.6 9.070.4

NMB-7 Control 0 45.074.0 41.573.3 13.570.9Control 2 47.073.7 41.573.9 11.570.4 dDFMO 2 60.373.9 31.376.5 8.472.9SAM486A 2 54.174.3 36.676.0 9.373.2DFMO/SAM486A 2 71.673.5 21.772.6 6.772.4 eDFMO/SAM486A/Spd 2 45.776.6 45.776.1 8.772.8 fControl 3 56.776.4 35.777.3 7.671.0DFMO 3 58.872.6 31.076.3 10.273.9SAM486A 3 56.476.5 35.277.0 8.571.4DFMO/SAM486A 3 69.274.1 28.472.6 2.472.3DFMO/SAM486A/Spd 3 61.0710.1 31.979.0 7.171.1

aDFMO¼ 5mM; SAM486A¼ 10mM; Spd¼ 10mM. bData represent mean values from three separate experiments7s.d.

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5613

Oncogene

interfere with ODC and AdoMetDC enzyme activities,cause polyamine depletion, and, ultimately, disturbcell cycle progression at the G1/S transition, whichleads to growth arrest. The combination of bothinhibitors generally intensified these effects (Table 3,Figures 2a and 5). We also found that the growthinhibition of cells treated with a combination of DFMOand SAM486A was sustained even after removal ofinhibitors, thus suggesting prolonged growth arrest(Figure 4).To further explore the underlying molecular mechan-

isms by which DFMO and SAM486A induce G1 cellcycle arrest, we examined the expression patterns of cellcycle-associated proteins in response to inhibitor treat-ments. Most significantly, we found that the proteinlevels of p27Kip1 sharply increased in the presence ofDFMO alone or combined with SAM486A (Figure 5a).Our results further showed that this increase in p27Kip1 istightly correlated with hypophosphorylation of Rb(Figure 5b). Since Rb is a known regulator of G1/S

transition, it is conceivable that the polyamine inhibitor-induced G1 cell cycle arrest in LAN-1 and NMB-7 cellsis regulated by Rb hypophosphorylation in response top27Kip1. Similarly, it was shown that p27Kip1 is a keymediator of RA-induced growth arrest in NB cells(Matsuo and Thiele, 1998). Our findings are also inpartial agreement with results obtained with other cellsystems, such as normal rat cells IEC-6 (Ray et al.,1999), human breast cancer cells MDA-MB-468 (Singhet al., 2001), and human melanoma cells MALME-3(Kramer et al., 2001). However, in those cell lines,polyamine inhibitor-induced G1 arrest was primarilyregulated by p21Waf1/Cip1 and p53, while DFMO andDFMO/SAM486A in LAN-1 and NMB-7 cells increa-sed p27Kip1 in the absence of both p21Waf1/Cip1 and p53.Furthermore, in contrast to the rapid growth inhibitionobserved in NB cells (Figure 2), IEC-6 and MALME-3cells required treatments of up to 10 days with DFMO(Ray et al., 1999) or 8–12 days with DFMO plus MDL-73811 (another AdoMetDC inhibitor) (Kramer et al.,

p27Kip1

GAPDH

LAN-1 NMB-7

DFMO SAM486A

Day 1 Day 2 Day 3 Day 1 Day 2 Day 3

ppRb (hyper)

p-Rb (Ser795)

GAPDH

LAN-1 NMB-7

p-Rb (Ser807/811)

DFMO SAM486A

Day 1 Day 2 Day 3 Day 1 Day 2 Day 3

pRb/ppRbpRb (hypo)

MYCN

LAN-1 NMB-7

DFMO SAM486A

Day 1 Day 2 Day 3 Day 1 Day 2 Day 3

100 69 57 9 100 14 90 7 100 20 132 6 100 70 76 22 100 31 70 11 100 17 62 5 % Control

β-actin

-- - - - - - - - - - - -

-----------+ + + + + + + + + + + +++++++++++++

-- - - - - - - - - - - -

-----------+ + + + + + + + + + + +++++++++++++

-- - - - - - - - - - - -

-----------+ + + + + + + + + + + +++++++++++++

1 1.2 0.8 1.4 0.7 1 0.7 1.5 0.6 1 1 1.3 0.8 1.2 0.9 1 0.5 0.8 0.6 1.5 0.5 0.9 0.9 0.6

a

b

c

Figure 5 Western blot analyses of inhibitor-treated human NB cells. Effect on (a) cyclin-dependent kinase inhibitor p27Kip1, (b)retinoblastoma protein Rb (phosphorylations at Ser795 and Ser807/811, and detection of total Rb), and (c) MYCN in LAN-1 andNMB-7 cells treated with 5mM DFMO, 10 mM SAM486A or the combination of both inhibitors. Cells were analysed on days 1, 2, and3. Equal amounts of total protein were loaded in all experiments. Proteins were scanned and the ratio of hypophosphorylated Rb(pRb) (lower band) : hyperphosphorylated Rb (ppRb) (upper band) was calculated (b). MYCN bands were quantified and normalizedrelative to b-actin. Values are presented as percentage (%) of controls of days 1, 2, or 3. (c). The data are representative of at least threeindependent experiments

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5614

Oncogene

2001), respectively, and showed almost no alterations incell morphology. This suggests that DFMO andSAM486A are more potent in NB cells and, possibly,also more successful in killing NB tumors.Downregulation of MYCN is of great interest with

regard to developing advanced NB therapies. It haspreviously been shown that DFMO blocks the expres-sion of c-Myc (Tabib and Bachrach, 1994) and thesefindings are extended by our own results, which showstrong downregulation of the c-Myc-related proteinMYCN in response to DFMO or DFMO/SAM486A(Figure 5c). It has also been shown that MYCN canregulate p27Kip1 in NB cells by targeting p27Kip1 to theproteasome (Nakamura et al., 2003), and it is thereforepossible that the observed increase in p27Kip1 is inresponse to a decrease in MYCN. Figure 6 shows thepossible involvement of polyamines in the regulationof p27Kip1, Rb, and MYCN in LAN-1 and NMB-7 cellsbased on our observations. The regulation of ODCby MYCN has been previously proposed (Lutz et al.,1996; Ben-Yosef et al., 1998; Lu et al., 2003) and is alsosubject to current investigations in our laboratory. Theprecise mechanism by which polyamines regulateMYCN in NB cells remains to be further clarified.In conclusion, our findings provide new insights into

the regulation of ODC, AdoMetDC, and polyamines inNB cells. Furthermore, we show for the first time that

the proteins p27Kip1, Rb, and MYCN play a key rolein polyamine inhibitor-induced G1 cell cycle arrest inMYCN-amplified NB cells. Unlike DFMO and DFMO/SAM486A, we found that SAM486A alone had almostno effect on p27Kip1, Rb, and MYCN despite itsprofound effects on polyamine pools, G1 arrest, andNB cell growth and morphology. The reason for this isunclear. We suspect that other cell signaling moleculesare involved in the observed SAM486A-induced growtharrest, which are yet to be identified. It is possible thatthe exceptionally high putrescine levels, which occuronly in SAM486A-treated cells (Table 2), may be thereason for the observed differences.The presented results are in support of our hypothesis

that the combined administration of DFMO andSAM486A provides an effective means for the treatmentof NB tumors (Bachmann, 2004). Our findings areparticularly relevant because DFMO and SAM486Acause rapid and prolonged growth inhibition of twoMYCN-amplified NB cell lines, which (1) are represen-tative of MYCN-amplified tumors with aggressivebehavior, and (2) are deficient in p53. Mutated p53is often associated with relapsed and more progressivetumors, which circumvent p53-mediated apoptosis andexhibit chemoresistance (Hopkins-Donaldson et al.,2002). Since DFMO and SAM486A have been recentlyevaluated in Phase III and Phase II human trials,respectively (Levin et al., 2003; Pless et al., 2004), thereis considerable information available regarding theirconcurrent side effects, pharmacokinetics, and maxi-mum tolerated doses. Therefore, DFMO andSAM486A, possibly in combination with cytotoxicdrugs at reduced doses, and concomitant with a strictpolyamine-limited diet, should be considered for thetreatment of late-stage NB patients with unfavorableoutcome.

Materials and methods

Chemicals

The ODC inhibitor DFMO (Metcalf et al., 1978) andthe AdoMetDC inhibitor SAM486A (Regenass et al.,1994; Svensson et al., 1997) were provided by Dr PatrickWoster (Wayne State University, MI, USA) and Novartis(Basel, Switzerland), respectively. Spd and aminoguanidinewere obtained from Sigma Chemical Co. (St Louis, MO,USA).

Cell lines and treatment of cultured cells

Human NB cell lines were provided by Dr Robert Seeger(LAN-1; University of California, Los Angeles, CA, USA),Dr Nai-Kong Cheung (NMB-7; Memorial Sloan-KetteringCancer Center, NY, USA), and the American Type CultureCollection (SK-N-SH; ATCC, Manassas, VA, USA). Cellswere maintained in RPMI 1640 (Biosource, Rockeville, MD,USA) containing 10% heat-inactivated fetal bovine serum(FBS) (Invitrogen, Carlsbad, CA, USA), penicillin (100 IU/ml)and streptomycin (100 mg/ml). If cells were treated withexogenous Spd, 1mM of aminoguanidine was included as aninhibitor of serum polyamine oxidation. Cells were seeded2–3 h before treatment with 5mM DFMO and/or 10mM

Rb- P

AdoMetDC ODC MYCN

G1 Cell Cycle Progression

Polyamines

SAM486A

?

p27Kip1

DFMO

ODC

Figure 6 Proposed interplay between polyamines and cell cycleregulatory proteins inMYCN-amplified NB cells. Polyamines playa key role in cell cycle regulation (Wallace et al., 2003), but proteinswhich are directly regulated by polyamines are still poorly defined.High polyamine titers promote cell cycle progression and cellproliferation, presumably via the blocking of cyclin-dependentkinase inhibitor p27Kip1, which leads to hyperphosphorylation ofretinoblastoma protein Rb. The blockade of polyamine biosyn-thesis with specific inhibitors DFMO and SAM486A leads to G1cell cycle arrest in response to p27Kip1 upregulation and subsequenthypophosphorylation of Rb. MYCN is strongly downregulated inresponse to polyamine depletion and is known to regulate p27Kip1

and ODC gene expression in NB cells

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5615

Oncogene

SAM486A and analysed after 1, 2, and/or 3 days. For growthinhibition, polyamine pool, and cell cycle studies, 10 mM Spdwas added together with inhibitors, and cells were assayedon days 2 and/or 3.

RT–PCR analysis

Total RNA was isolated from control or treated NB cells usingthe RNeasy mini kit (Qiagen Inc., Valencia, CA, USA). ThecDNA synthesis and standard RT–PCR analysis wereperformed with DNAse-treated total RNA using the followingprimer sets: ODC-fwd, 50-gtatgctgctaataatggag-30; ODC-rev,50-ttacgccggtgatctcttca-30; AdoMetDC-fwd, 50-agccttctcaccaagggtacc-30; AdoMetDC-rev, 50-cctggcttgaagacttctac-30; MYCN-fwd, 50-cacaaggccctcagtacctc-30; MYCN-rev, 50-gatcagctcgctg-gactgag-30; b-actin-fwd, 50-gctcgtcgtcgacaacggctc-30; b-actin-rev, 50-ccaacatgatctgggtcatcttctc-30 (loading control).

ODC and AdoMetDC activity assays

ODC and AdoMetDC enzyme activities were determinedusing a standard assay as previously reported (Coleman andPegg, 1998; Shantz and Pegg, 1998). In brief, NB cells werewashed twice with ice-cold PBS and harvested in 50mMsodium phosphate buffer (pH 7.2), 0.1mM EDTA, 2.5mMDTT, supplemented with 1� protease inhibitor cocktail(Roche Molecular Biochemicals, Indianapolis, IN, USA).For the measurement of AdoMetDC activity, cells wereharvested in the above buffer supplemented with 2.5mM Put.Cells were lysed by three freeze/thaw cycles and cell lysateswere prepared by centrifugation for 20min at 14 000 r.p.m. and41C. ODC activity was measured in a reaction mixture thatcontained 20mM L-ornithine (47.70mCi/mmol; NEN LifeScience Products, Boston, MA, USA). AdoMetDC activitywas assayed in a reaction mixture that contained 8.2 mMS-adenosyl-L-[carboxy-14C]methionine (58.0–61.0mCi/mmol;Amersham Biosciences, Piscataway, NJ, USA) and 140.1mMcold S-adenosylmethionine. Extracts from untreated orDFMO-treated LAN-1 cells were incubated in 100mMDFMO for 30min at 371C prior to the ODC activityassay to test for DFMO-resistant ODC. Untreated orSAM486A-treated LAN-1 extracts were assayed forAdoMetDC activity after multiple rounds of sequentialdilution with harvest buffer, followed by concentration ofthe extract (Centricon YM-10, Millipore Corporation, Bed-ford, MA, USA) to remove the inhibitor. This process resultedin an approximately 400-fold dilution of the residual inhibitorin the extract without a significant change in the overallprotein concentration.

Polyamine pool analysis

Intracellular polyamines were extracted from cell pellets aspreviously described (Gilbert et al., 1991). In brief, cells werewashed and centrifuged. After sonication and centrifugation,250 ml of supernatant, 100 ml of saturated sodium carbonate,and 300 ml of 25mg/ml dansyl chloride (5-dimethylamino-naphthalene-1-sulfonyl chloride) (Fluka, Switzerland) weremixed and analysed via reverse-phase HPLC using a C18reverse-phase analytical column (Waters, Milford, MA, USA).The effluent from the column was monitored at 510 nm(excitation at 330 nm). Polyamine levels were determined usingappropriate standards. The samples were normalized in 0.2 Nsodium hydroxide and the amount of total protein per samplewas measured using the Bradford assay (70). Polyamine levelsare expressed as nmol per mg of total cellular protein.

Flow cytometry

Cell samples were stained with PI using the proceduredescribed by Krishan (1975) and subjected to flow-cytometricanalysis using a FACScan flow cytometry instrument (BectonDickinson, San Jose, CA, USA). Cell cycle distribution wasdetermined using the ModFit software.

Western blot analysis

Cell lysates were prepared in RIPA buffer (20mM Tris-HCl,pH 7.5, 0.1% sodium lauryl sulfate, 0.5% sodium deoxycho-late, 135mM NaCl, 1% Triton X-100, 10% glycerol, 2mMEDTA), supplemented with a protease inhibitor cocktail(Roche Molecular Biochemicals, Indianapolis, IN, USA) andphosphatase inhibitors sodium fluoride (NaF; 20mM) andsodium vanadate (Na3VO4; 0.27mM). The total proteinconcentration was determined using the protein assay dyereagent from Bio-Rad Laboratories (Hercules, CA, USA). Celllysates in SDS-sample buffer were either boiled for 5min orincubated at room temperature for 60min, and equal amountsof total protein analysed by 10% SDS–polyacrylamide gelelectrophoresis (PAGE) and Western blotting using PVDFmembranes. The antibodies used in this study are: mousemonoclonal p21Waf1/Cip1 (sc-817), rabbit polyclonal p27Kip1

(sc-528), mouse monoclonal p53 (sc-126), rabbit polyclonalMYCN (sc-791), and goat polyclonal ODC (sc-21515) fromSanta Cruz Biotechnology (Santa Cruz, CA, USA); rabbitpolyclonal phospho-Rb (Ser795), rabbit polyclonal phospho-Rb (Ser807/811), and mouse monoclonal Rb (4H1) from CellSignaling Technology, Inc. (Beverly, MA, USA); mousemonoclonal b-actin (A5316) from Sigma Chemical Co. (StLouis, MO, USA), and mouse monoclonal glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (6C5) from Ambion, Inc.(Austin, TX, USA). The AdoMetDC-recognizing rabbit serumwas described previously (Shantz et al., 1994). Secondary anti-mouse and anti-rabbit antibodies (Amersham Biosciences,Piscataway, NJ), and secondary anti-goat antibody (SantaCruz Biotechnology, Santa Cruz, CA, USA) were coupled tohorseradish peroxidase (HRP). Proteins were detected usingthe ECL or ECL Plus reagents (Amersham Biosciences,Piscataway, NJ, USA) and Kodak BioMax XAR film (FisherScientific, Pittsburgh, PA, USA), and bands quantified usinga Bio-Rad Fluor-S Multi Imager and Quantity One QuantitationSoftware, Version 4 (Bio-Rad Laboratories, Hercules, CA, USA).

Microscopy

Cells were grown in cell culture dishes, treated with inhibitorsas described above and light micrographs were taken after 3days using an inverted phase-contrast microscope (NikonDiaphot, Nikon Corporation, Tokyo, Japan). To visualize theactin cytoskeleton, NB cells were grown on glass coverslips in12-well plates and at the end of the treatment stained withTexas Red-X phalloidin according to the manufacturer’sinstructions (Molecular Probes, Eugene, OR, USA). Cellswere washed, fixed in 3% p-formaldehyde for 10min, andexposed to 0.1% Triton X-100 for 5min. Fixed cells werewashed again, blocked in 1% BSA/PBS for 30min, andincubated with phalloidin for 20min in the dark at roomtemperature. Washed coverslips were mounted using Vecta-shields Mounting Medium with DAPI (Vector Laboratories,Inc., Burlingame, CA, USA) and samples analysed with anepifluorescence microscope (Zeiss Axioplan).

Acknowledgements

We are grateful to Dr Randal Wada (Cancer ResearchCenter of Hawaii) for providing us with the MYCN antibody

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5616

Oncogene

in initial studies, and to Drs Carl-Wilhelm Vogel, BonnieWarn-Cramer, Darren Park, Patricia Lorenzo, and MattTuthill (Cancer Research Center of Hawaii) for their support,advice, and stimulating discussions during the course of thiswork. Crystal Fo and Craig Coleman are thanked for theirexcellent technical support of this project. We give specialthanks to Dr Patrick Woster (Wayne State University, MI,USA) for providing the ODC inhibitor DFMO. We also thank

Novartis (Basel, Switzerland) for providing the AdoMetDCinhibitor SAM486A. This work was supported by the HawaiiCommunity Foundation, grants #20022113 and #20041684 toAS Bachmann. KY Takasaki received support from the Celland Molecular Biology Graduate Program of the University ofHawaii at Manoa. SM Wilson was supported by the NIHgrant R01CA79909 to CV Byus. Additional funds wereprovided by the NIH grant R01CA18138 to AE Pegg.

References

Bachmann AS. (2004). Hawaii Med. J., 63, 371–374.Ben-Yosef T, Yanuka O, Halle D and Benvenisty N. (1998).

Oncogene, 17, 165–171.Brodeur GM. (2003). Nat. Rev. Cancer, 3, 203–216.Brodeur GM, Seeger RC, Schwab M, Varmus HE and BishopJM. (1984). Science, 224, 1121–1124.

Carbone PP, Pirsch JD, Thomas JP, Douglas JA, Verma AK,Larson PO, Snow S, Tutsch KD and Pauk D. (2001). CancerEpidemiol. Biomarkers Prev., 10, 657–661.

Chapman SK. (1980). Life Sci., 26, 1359–1366.Coleman CS and Pegg AE. (1998). Methods Mol. Biol., 79,41–44.

Davidoff AM, Pence JC, Shorter NA, Iglehart JD and MarksJR. (1992). Oncogene, 7, 127–133.

Davidson NE, Hahm HA, McCloskey DE, Woster PMand Casero Jr RA. (1999). Endocr. Relat. Cancer, 6,69–73.

Dorhout B, te Velde RJ, Ferwerda H, Kingma AW,de Hoog E and Muskiet FA. (1995). Int. J. Cancer, 62,738–742.

Eskens FA, Greim GA, van Zuylen C, Wolff I, Denis LJ,Planting AS, Muskiet FA, Wanders J, Barbet NC, Choi L,Capdeville R, Verweij J, Hanauske AR and Bruntsch U.(2000). Clin. Cancer Res., 6, 1736–1743.

Fabian CJ, Kimler BF, Brady DA, Mayo MS, Chang CH,Ferraro JA, Zalles CM, Stanton AL, Masood S, GrizzleWE, Boyd NF, Arneson DW and Johnson KA. (2002). Clin.Cancer Res., 8, 3105–3117.

Galderisi U, Di Bernardo G, Cipollaro M, Peluso G, CascinoA, Cotrufo R and Melone MA. (1999). J. Cell. Biochem., 73,97–105.

Galderisi U, Jori FP and Giordano A. (2003). Oncogene, 22,5208–5219.

Gilbert RS, Gonzalez GG, Hawel III L and Byus CV. (1991).Anal. Biochem., 199, 86–92.

Goldman SC, Chen CY, Lansing TJ, Gilmer TM and KastanMB. (1996). Am. J. Pathol., 148, 1381–1385.

Hopkins-Donaldson S, Yan P, Bourloud KB, Muhlethaler A,Bodmer JL and Gross N. (2002). Oncogene, 21, 6132–6137.

Kramer DL, Chang BD, Chen Y, Diegelman P, Alm K, BlackAR, Roninson IB and Porter CW. (2001). Cancer Res., 61,7754–7762.

Kramer DL, Khomutov RM, Bukin YV, Khomutov AR andPorter CW. (1989). Biochem. J., 259, 325–331.

Krishan A. (1975). J. Cell Biol., 66, 188–193.Levin VA, Hess KR, Choucair A, Flynn PJ, Jaeckle KA,Kyritsis AP, Yung WK, Prados MD, Bruner JM, Ictech S,Gleason MJ and Kim HW. (2003). Clin. Cancer Res., 9,981–990.

Levin VA, Uhm JH, Jaeckle KA, Choucair A, Flynn PJ, YungWKA, Prados MD, Bruner JM, Chang SM, Kyritsis AP,Gleason MJ and Hess KR. (2000). Clin. Cancer Res., 6,3878–3884.

Lu X, Pearson A and Lunec J. (2003). Cancer Lett., 197,125–130.

Lutz W, Stohr M, Schurmann J, Wenzel A, Lohr A andSchwab M. (1996). Oncogene, 13, 803–812.

Matsuo T and Thiele CJ. (1998). Oncogene, 16, 3337–3343.

McCann PP and Pegg AE. (1992). Pharmacol. Ther., 54,195–215.

McKenzie PP, Danks MK, Kriwacki RW and Harris LC.(2003). Cancer Res., 63, 3840–3844.

Melino G, Farrace MG, Ceru MP and Piacentini M. (1988).Exp. Cell Res., 179, 429–445.

Melino G, Piacentini M, Patel K, Annicchiarico-Petruzzelli M,Piredda L and Kemshead JT. (1991). Prog. Clin. Biol. Res.,366, 283–291.

Metcalf BW, Bey P, Danzin C, Jung MJ, Casara P and VevertJP. (1978). J. Am. Chem. Soc., 100, 2551–2553.

Meyskens Jr FL and Gerner EW. (1999). Clin. Cancer Res., 5,945–951.

Nakamura M, Matsuo T, Stauffer J, Neckers L and Thiele CJ.(2003). Cell Death Differ., 10, 230–239.

Paridaens R, Uges DR, Barbet N, Choi L, Seeghers M, vander Graaf WT, Groen HJ, Dumez H, Buuren IV, Muskiet F,Capdeville R, Oosterom AT and de Vries EG. (2000). Br. J.Cancer, 83, 594–601.

Pegg AE. (1988). Cancer Res., 48, 759–774.Pless M, Belhadj K, Menssen HD, Kern W, Coiffier B, Wolf J,Herrmann R, Thiel E, Bootle D, Sklenar I, Muller C, ChoiL, Porter C and Capdeville R. (2004). Clin. Cancer Res., 10,1299–1305.

Porter CW, Regenass U and Bergeron RJ. (1992).Falk Symposium on Polyamines in the Gastrointestinal Tract,Dowling Rh, Folsch Ur, Loser C (eds). KluwerAcademic Publishers: Dordrecht, The Netherlands, pp.301–322.

Ray RM, Zimmerman BJ, McCormack SA, Patel TBand Johnson LR. (1999). Am. J. Physiol., 276, C684–C691.

Regenass U, Mett H, Stanek J, Mueller M, Kramer D andPorter CW. (1994). Cancer Res., 54, 3210–3217.

Seeger RC, Brodeur GM, Sather H, Dalton A, Siegel SE,Wong KY and Hammond D. (1985). N. Engl. J. Med., 313,1111–1116.

Shantz LM and Pegg AE. (1998). Methods Mol. Biol., 79,45–49.

Shantz LM, Viswanath R and Pegg AE. (1994). Biochem. J.,302, 765–772.

Singh R, Pervin S, Wu G and Chaudhuri G. (2001).Carcinogenesis, 22, 1863–1869.

Siu LL, Rowinsky EK, Hammond LA, Weiss GR,Hidalgo M, Clark GM, Moczygemba J, Choi L, LinnartzR, Barbet NC, Sklenar IT, Capdeville R, Gan G, PorterCW, Von Hoff DD and Eckhardt SG. (2002). Clin. CancerRes., 8, 2157–2166.

Svensson F, Mett H and Persson L. (1997). Biochem. J., 322,297–302.

Tabib A and Bachrach U. (1994). Biochem. Biophys. Res.Commun., 202, 720–727.

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5617

Oncogene

Takahashi Y, Mai M and Nishioka K. (2000). Int. J. Cancer,85, 243–247.

Tweddle DA, Malcolm AJ, Bown N, Pearson AD and LunecJ. (2001). Cancer Res., 61, 8–13.

Tweddle DA, Pearson AD, Haber M, Norris MD, Xue C,Flemming C and Lunec J. (2003). Cancer Lett., 197, 93–98.

van Zuylen L, Bridgewater J, Sparreboom A, Eskens FA, deBruijn P, Sklenar I, Planting AS, Choi L, Bootle D, MuellerC, Ledermann JA and Verweij J. (2004). Clin. Cancer Res.,10, 1949–1955.

Wallace HM, Fraser AV and Hughes A. (2003). Biochem. J.,376, 1–14.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Regulation of p27Kip1, Rb, and MYCN by polyamine inhibitorsCJ Wallick et al

5618

Oncogene

Supplementary Information

Chemical structures of α-difluoromethylornithine (DFMO, also known as Eflornithine) and 4-(aminoiminomethyl)-2,3-dihydro-1H-inden-1-one-diaminomethylenehydrazone (SAM486A, also known as CGP48664). A, DFMO (MW 218.6) is a specific suicide inhibitor of ornithine decarboxylase (ODC). B, SAM486A (MW 321.1) inhibits S-adenosylmethionine decarboxylase (AdoMetDC). Both ODC and AdoMetDC are sentinel enzymes in the polyamine biosynthetic pathway.

H2N

NH2

CHF2COOH

NH

NH2

NNH

NH

NH2

A DFMO

B SAM486A