Spermine cytotoxicity to human colon carcinoma-derived cells (CaCo-2)

N. Seiler, B. Duranton, F. Gosse and F. RaulCJF INSERM 95-09, Institut de Recherche Contre les Cancers de l'Appareil Digestif (IRCAD),Strasbourg, France

Received 23 September 1999; accepted 3March 2000

Keywords: spermine, CaCo-2 cells, polyamine oxidase, MDL 72527, cell death

Abstract

Spermine is a constituent of all vertebrate cells. Nevertheless, it exerts toxic e¡ects if it accumulatesin cells. Spermine is a natural substrate of the FAD-dependent polyamine oxidase, a constitutiveenzyme of many cell types. It has been reported that the toxicity of spermine was enhanced ifpolyamine oxidase was inhibited.We were interested to examine spermine toxicity to human coloncarcinoma-derived CaCo-2 cells because, in contrast to most tumor cell lines, CaCo-2 cells undergodi¡erentiation, which is paralleled by changes in polyamine metabolism. CaCo-2 cells wereremarkably resistant to spermine accumulation, presumably because spermine is degraded bypolyamine oxidase at a rate su¤cient to provide spermidine for the maintenance of growth.Inactivation of polyamine oxidase increased the sensitivity to spermine. A major reason for theenhanced spermine cytotoxicity at low polyamine oxidase activity is presumably the profounddepletion of spermidine, and the consequent occupation of spermidine binding sites by spermine.Hydrogen peroxide and the aldehydes 3-aminopropanal and 3-acetamidopropanal, the products ofpolyamine oxidase-catalyzed splitting of spermine and N1-acetylspermine, contribute little tospermine cytotoxicity. Activation of caspase by spermine was insigni¢cant, and the formation ofDNA ladders, another indicator of apoptotic cell death, could not be observed. Thus it appears thatcell death due to excessive accumulation of spermine in CaCo-2 cells was mainly nonapoptotic. Thecontent of brush border membranes did not change between days 6 and 8 after seeding, and it wasnot a¡ected by exposure of the cells to spermine. However, the activities of alkaline phosphatase,sucrase, and aminopeptidase in nontreated cells were considerably enhanced during this period, butremained low if cells were exposed to spermine. These changes appear to indicate that di¡erentia-tion is prevented by intoxication with spermine, although other explanations cannot be excluded.

Abbreviations: AdoMetDC, S-adenosylmethionine decarboxylase; DAO, diamine oxidase; MAO,monoamine oxidase; MDL 72527, N1,N4-bis(2,3-butadienyl)-1,4-butanediamine; N1acSpd, N1-acetylspermidine; N1acSpm, N1-acetylspermine; ODC, ornithine decarboxylase; PAO, polyamineoxidase (FAD-dependent); Put, putrescine (1,4-butanediamine); SAO, serum amine oxidase (Cu-containing); Spd spermidine; Spm spermine

Cell Biology and Toxicology. 2000; 16: 117^130.# 2000Kluwer Academic Publishers. Printed in the Netherlands

Introduction

The polyamines spermidine (Spd) andspermine (Spm) are constituents of nearly allvertebrate cells and surpass other biogenicamines in concentration (Cohen, 1998). In thecase of nonphysiological accumulation, theyexert cytotoxic e¡ects. Cytotoxic properties ofthe products of oxidative deaminations of Spmand Spd have been known for more than 30years (Cohen, 1998). Parchment (1996)reviewed known and suspected polyamine-

related mechanisms of apoptosis, with empha-sis on toxic products of polyamine oxidation.

In mammalian tissues and organs, Spm isthe substrate mainly of two oxidases: Diamineoxidases (DAO), are tissue speci¢c copper-containing amine oxidases. They catalyse theoxidative deamination of diamines (putrescine(Put), cadaverine, etc.) most actively, but Spdand Spm are also substrates. Reactionproducts are aldehydes, ammonia, and hydro-gen peroxide (Seiler, 1992) (Figure 1). DAOactivity is highest in gut mucosa and kidney,

Figure 1. Reactions of spermine and its N1-acetyl derivative with diamine oxidase (DAO) and polyamine oxidase (PAO).

118

and low in most other organs. In contrast, theFAD-dependent polyamine oxidase (PAO) isvirtually ubiquitous. It is an enzyme of the so-called interconversion cycle (Seiler, 1987,1995). With removal of the 3-acetamidopropylmoieties of N1-acetylspermidine (N1acSpd)and N1-acetylspermine (N1acSpm), Put andSpd are formed, respectively, wherebyacetamidopropanal and hydrogen peroxideare liberated (Figure 1). Spm, in contrast withSpd, is a good substrate of PAO. The degrada-tion of Spm by PAO in red blood cells isanother important function of this enzyme.Several reports demonstrate that the deregu-

lated (excessive) uptake of polyamines (Poulinet al., 1995; Ientile et al., 1997; Xie et al.,1997), and of structural analogues of the poly-amines (Hu and Pegg, 1997), induces apoptosisin di¡erent cell lines. Inactivation of PAO withN1,N4-bis(2,3-butadienyl)-1,4-butanediamine(MDL 72527), a selective inactivator of PAO(Bey et al., 1985; Bolkenius and Seiler, 1989)enhances the cytotoxicity of Spm in babyhamster kidney cells (Brunton et al., 1991).We were interested to study PAO and PAO-

related aspects of polyamine cytotoxicity inrelation to cell di¡erentiation. CaCo-2 cells, ahuman colon carcinoma-derived cell line(Fogh et al., 1977) appeared a suitable modelfor this purpose, for the following reasons:CaCo-2 cells undergo spontaneous structuraland functional di¡erentiation at con£uence,without stopping growth. Di¡erentiationinc ludes formation of brush bordermembranes, the development of transport sys-tems, and the expression of hydrolases on theapical membrane of the cells (Pinto et al., 1983;Grasset et al., 1984; Rousset, 1986; Blais et al.,1987). Concomitant with structural andfunctional di¡erentiation, polyamine metabo-lism changes: ornithine decarboxylase (ODC)(D'Agostino et al., 1989; Duranton et al.,1998), and S-adenosylmethionine decarboxy-lase (AdoMetDC) (Duranton et al., 1998)activities decrease together with the concentra-

tions of Put and Spd. DAO activity increases(D'Agostino et al., 1989), and the preferentialuptake of polyamines changes from basolateralto apical uptake (D'Agostino et al., 1990). ThusCaCo-2 cells in some respect resemble gutmucosa cells. Although PAO activity had notbeen determined, it was presumed that inanalogy to gut mucosa (Seiler et al., 1980),CaCo-2 cells would exhibit a high PAOactivity.

Materials and methods

Chemicals

If not stated otherwise, chemicals were fromSigma Chemical Co., St. Louis, MO, USA, orMerck, Darmstadt, Germany. N1,N12-Diacetylspermine dihydrochloride wassynthesized as described previously (Bolkeniusand Seiler, 1981). MDL 72527 (N1,N4-bis(2,3-butadienyl)-1,4-butanediamine dihydro-chloride) was developed as a selective inactiva-tor of PAO (Bey et al., 1985) at the MerrellDow Research Institute, Strasbourg. Cellculture media were bought from Gibco (LifeTechnologies SARL, Cergy-Pontoise, France).

Cell culture

CaCo-2 cells were obtained from the EuropeanCollection of Animal Cell Culture (Sophia-Antipolis, France). They were routinelycultured in 75-cm2 Falcon £asks in Dulbecco'smodi¢ed Eagle medium (DMEM) at 25mmol/L glucose, supplemented with 10% fetalbovine serum, 1% nonessential amino acids,100 U/ml penicillin, and 100 mg/ml strepto-mycin at 378C in a humidi¢ed atmosphere of5% CO2. For the experiments cells of up to 45passages were used.

119

Determination of cytotoxicity

After trypsinization (0.5% trypsin in 2.6mmol/L EDTA), 46103 cells per well wereseeded in 96-well microplates, and incubatedin 200 ml of the above-mentioned medium,except that 3% horse serum was used insteadof 10% fetal bovine serum, and the mediumwas supplemented with 5 mg/ml transferrin, 10mg/ml insulin, and 5 ng/ml selenium (ITSde¢ned medium, Gibco). The use of lowconcentrations of horse serum avoids thedegradation of important amounts of Spmand Spd by serum amine oxidase (SAO).Medium was changed every other day. Thechemicals were added dissolved in 200 mlculture medium. At the end of the incubationperiod, cell growth was stopped by addition of50 ml trichloroacetic acid (50% w/v), and theprotein content of each well was determined bystaining with sulforhodamine B (Skehan et al.,1990). Absorbance was determined at low cellnumbers (3 days of incubation) at 558 nm, andat high cell numbers (8 days of incubation) at490 nm. The relationship between cell number(protein content per well) and absorbency islinear over a very wide range.

Flow cytometry

Cells (66105 per 10-ml Petri dish) were seededand exposed between 24 h and 72 h afterseeding to 5 mmol/L Spm. Cells attached tothe surface of the plastic culture dish wereharvested after trypsinization. The washedcells were ¢xed in methanol^PBS (9:1) at^208C. The ¢xed cells were washed with PBS,treated with RNAase (25 mg/ml) for 15 min atroom temperature, and stained with propidiumiodide (50 mg/ml) for 30 min on ice. Thestained cells were analyzed by a FACS £owcytometer (Becton Dickinson, San Jose, CA,USA). Cell cycle phase distribution wasquanti¢ed using CellFit software (BectonDickinson).

Determination of polyamines

Cells (66105) were seeded in 10-ml Petridishes, and incubated in supplementedDMEM culture medium for 6 days; mediumwas changed every other day. The chemicals(dissolved in 5 ml fresh medium) were addedon day 6. The following (¢nal) concentrationswere used: Spm 5 mmol/L, MDL 72527 50mmol/L, Spm 1.25 mmol/L (in combinationwith 50 mmol/L MDL72527). After incubationfor 48 h with the drugs, the cell layers werewashed three times with 5 ml portions ofVersene (140 mmol/L NaCl, 3 mmol/L KCl,1.5 mmol/L KH2PO4, 15 mmol/L Na2HPO4,2.6 mmol/L EDTA, pH 7.2), harvested using adisposable cell scraper, and pelleted bycentrifugation. The cell pellets were washedwith 5 ml Versene before they were storedfrozen at ^808C. In order to determine extra-cellular Spm in the pellets, culture dishes withuntreated controls were placed on ice, and 5mmol/L, or 1.25 mmol/L Spm was added tothe culture medium. After 1 min these cellswere washed and harvested as usual. The acidsoluble materials, including the polyamines,were extracted by mixing the cell pellets with1 ml 0.2 mol/L perchloric acid. After centrifu-gation the polyamines were determined in thesupernatants by reversed-phase ion pairHPLC, post-column reaction with o-phthal-aldehyde/2-mercaptoethanol reagent, andmonitoring of £uorescence intensity (Seiler etal., 1990). The acid-insoluble pellets were usedfor protein determinations.

Isolation of brush-border membranes (Schmitzet al., 1973)

Cells were treated, harvested, and stored asdescribed above for the polyamine determina-tions. The frozen cells were homogenized inTris (2 mmol/L) ^ mannitol (50 mmol/L)bu¡er pH 7.1 by sonication. CaCl2 was addedat a ¢nal concentration of 10 mmol/L. Cell

120

debris was precipitated by centrifugation at1200g for 10 min; the brush border membraneswere sedimented at 33 500g (30 min). Thepellets were rehomogenized by sonication in200 ml water.Sucrase activity was determined in brush

border membranes by the method of Messerand Dahlqvist (1966).Alkaline phosphatase activity was assayed

in brush border membranes by the method ofGaren and Levinthal (1960).Aminopeptidase activity was determined in

brush border membranes according to Marouxet al. (1973).

Polyamine oxidase

The assay of PAO activity relied on the deter-mination of hydrogen peroxide by horseradishperoxidase-catalyzed formation of quinonefrom phenol, and reaction of the quinone withaminoantipyrine. The assay conditions wereadopted from Hayashi et al. (1989). However,the glycine homogenization bu¡er contained10 mmol/L pargyline, 10 mmol/L amino-guanidine, and 0.02% sodium azide to inhibitmonoamine oxidase (MAO), DAO, andcatalase, respectively. N1,N12-diacetylspermine(2.5 mmol/L in assay) was used as substrate,because it is selective for PAO (Bolkenius andSeiler 1981). The molar extinction coe¤cientof the reaction product of quinone and amino-antipyrine at 500 nm (6390 L/(mol cm)) wasused for the calculation of the reaction rate.

Determination of caspase activity

Cells £oating in the culture medium weresedimented by centrifugation, and combinedwith the attached cells that were harvested asusual by scraping. Caspase activity wasdetermined by using the commercial ApopainAssay Kit (Bio-Rad, Ivry Sur Seine, France).The method relies on the £uorimetric determi-nation of the rate of release of 7-amino-4-

tri£uoromethylcoumarine (AFC) by celllysates from the £uorogenic peptide carbo-benzoxy-Asp-Glu-Val-Asp-AFC. For thedetermination of the selectivity of proteaseactivity the speci¢c apopain inhibitor acetyl-Asp-Glu-Val-Asp-chloromethylketone wasadded to the reaction mixture. Fluorescencewas activated at 390 nm; £uorescence emissionwas determined at 550 nm.

Determination of apoptosis-associated DNAfragmentation

The detached cells, £oating in the culturemedium, were used for the isolation of DNA.They were collected by centrifugation andstored at ^808C. Cells remaining attachedserved as controls. Apoptotic DNA wasseparated from genomic DNA using theSuicide Track DNA ladder isolation kit (Onco-gene Research Products, Cambridge, MA,USA). The DNA fragments were separated bygel electrophoresis and stained with ethidiumbromide.

Proteins

Proteins in homogenates of cells and brushborder membranes were determined accordingto Lowry et al. (1951).

Statistics

The signi¢cance of di¡erences between treatedand nontreated cells was established by theStudent^Neumann^Keuls multiple compari-son test.

Results

Polyamine oxidase activity

Although the formation of hydrogen peroxide,a product of several oxidases, was determined,

121

the presence of inhibitors of MAO and DAOin the incubation medium, and the use ofN1,N12-diacetylspermine as substrate ensuredthat exclusively PAO activity was determinedunder the assay conditions. As appears fromFigure 2, PAO was found in CaCo-2 cells asearly as 3 days after seeding (earlier time pointswere not studied). It increased with di¡erentia-tion, but even 15 days after seeding PAOactivity was only 58% above the 3-day value.

Exposure of the cells to 50 mmol/L MDL72527 between days 6 and 8 after seedinginactivated 75% of the oxidase. Inhibition ofPAO may have been more complete than isindicated by this value. Disregarding competi-tive inhibition (in addition to inactivation) byMDL 72527, a partial recovery of activeenzyme from the inactivated PAO during thepreparation of the cell homogenate may haveoccurred.Exposure of CaCo-2 cells to 5 mmol/L Spm

during 48 h had no e¡ect on PAO activity.

Cytotoxicity of spermine

Under the experimental conditions the CaCo-2cells grew exponentially for 9 days, with anapproximate doubling time of 2 days (Figure3). Spm at 5 mmol/L completely prevented thegrowth of CaCo-2 cells. After 8 days ofexposure to 2.5 mmol/L Spm the cell numberwas lower than in controls by about 60%(Figure 3A).

Figure 2. Polyamine oxidase activity in CaCo-2 cells. Cells weregrown in 10-ml Petri dishes, as described in Materials andMethods. The washed cells were stored at ^808C until PAOdeterminations were carried out. For the enzyme assay of 3-day-old cells the contents of 3 dishes were combined; for theassay of 5-day-old cells the contents of 2 dishes were combined.The error bars indicate+SD (n = 3).

Figure 3. Growth of CaCo-2 cells in the absence and presence of1.25, 2.5, and 5 mmol/L spermine (A) and in the absence andpresence of 50 mmol/L MDL 72527 and combinations of 50mmol/L MDL + 0.31, 0.625 and 1.25 mmol/L spermine (B).Ordinate is absorbance at 490 nm; the error bars indicate +SD(n = 6^8). 4000 cells were seeded in 96-well microplates andgrown under the conditions described in Materials andMethods. Drugs were added to 200 ml culture medium on day1 after seeding. Cells were exposed to the drugs throughout theincubation period; drug-free and drug-containing medium waschanged every other day.

122

Exposure of CaCo-2 cells to 50 mmol/LMDL 72527 reduced the cell number by about60% after 8 days; addition of Spm at a concen-tration of 1.25mmol/L resulted in complete cellloss, and 0.625 mmol/L Spm prevented theincrease of the cell number (Figure 3B).Cytotoxicity was also determined by

exposure of the cells to Spm and combinationsof Spm and the PAO inactivator MDL 72527for two days. Exposure between days 1 and 3to 1.25 mmol/L Spm caused a signi¢cantgrowth retardation (Figure 4A). In contrast,Spm even at a concentration of 2.5 mmol/Ldid not signi¢cantly a¡ect the number of cellsif they were exposed between days 6 and 8 tothe drug, and at 5 mmol/L Spm the cellnumber was lower by only about 50%compared with controls (Figure 4B).Two days of exposure to MDL 72527 had no

signi¢cant e¡ect on the growth rate of CaCo-2cells. However, addition of Spm in the presenceof 50 mmol/L MDL 72527 to the culturemedium at concentrations 40.3 mmol/Lreduced the cell number dose-dependently in3-day-old cells (Figure 4A) and less e¡ectivelyin 8-day-old cells (Figure 4B). At 2.5 mmol/LSpm, cell loss was complete both in 3- and 8-day-old cells.On the light-microscopic level, morpho-

logical alterations of the cells due to exposureto 5 mmol/L Spm between days 1 and 3 wereunimpressive. Cells di¡ered from controlsmainly by the lower number and smaller isletsand by the presence of isolated cells with adi¡use borderline and of cell fragments.The time course of Spm cytotoxicity to

con£uent CaCo-2 cells was determined asfollows: Cells were exposed to Spm/MDL72527 combinations on day 7 after seeding of76104 cells per well. At 1, 2, 3, 5, 7, 24, and 48h of exposure to the drug combination, theculture media were exchanged against fresh(drug-free) media, and incubation was contin-ued for a total of 48 h. At this time the cellprotein content of all wells was determined. As

is shown in Figure 5, after about 7 h ofexposure with all Spm/MDL 72527 combina-tions, steady-state growth rates were attainedby the CaCo-2 cells

In spite of a high DAO activity in con£uentCaCo-2 cells (D'Agostino et al., 1989),addition of 50 mmol/L or 1 mmol/L of theDAO inhibitor aminoguanidine to MDL72527 and Spm-exposed cells had only a smallprotective e¡ect (not shown).

Figure 4. Cytotoxic e¡ects of spermine and of combinations of50 mmol/L MDL 72527 + spermine. (A) CaCo-2 cells exposedto the chemicals between 24 and 72 h after seeding. (B) CaCo-2cells exposed to the chemicals between days 6 and 8 afterseeding (50 mmol/L MDL 72527 had no signi¢cant e¡ect oncell growth). For experimental details see Materials andMethods. The error bars indicate +SD (n = 6^12).*Statistically signi¢cant (p50.01) di¡erence between numbersof treated and untreated cells.

123

Cell cycle phase distribution

Exponentially growing cells that remainedattached to the surface of the plastic culturedish accumulated in the presence of 5 mmol/LSpm in the G2 phase of the cell cycle (Table 1),indicating a retarded G2/M-progression.

Polyamines

The polyamine concentrations of CaCo-2 cellswere determined on day 8 after seeding, i.e., incells that were exposed to the drugs for 48 h.Cells were either untreated or were exposed to50 mmol/LMDL 72427 or 5 mmol/L spermineor 50 mmol/L MDL 72527 + 1.25 mmol/LSpm. The data are summarized in Table 2. Inthe assessment of these and some other data ofthis work, one has to remember that only cellsthat remained attached to the surface of theplastic dishes were analyzed. The polyaminedata therefore represent the less damagedportion of the cells.

In 8-day-old CaCo-2 cells Put concentra-tions were low, in agreement with a low ODCactivity (D'Agostino et al., 1989). N1acSpdconcentrations were close to the detection limitof the method, and N1acSpm was undetect-able. The most conspicuous e¡ects of exposureof the cells to Spm and the PAO inactivatorwere the following:

(a) At 50 mmol/L MDL 72527, Put decreasedto close to the detection limit and Spd byabout 70%. Spm increased by a signi¢cantdegree (19%), and the concentrations ofN1acSpd and N1acSpm increased to anextent that the polyamine-derived positivecharge remained the same as in controlcells.

(b) The presence of 5 mmol/L Spm in theculture medium more than doubled itsintracellular concentration. (The amountof extracellular Spm in the pellet wasnegligibly low in comparison with its intra-cellular content, as was evident from con-trol experiments). Spd concentrationremained unchanged. The concentrationof N1acSpd increased dramatically, that ofN1acSpm to a value observed after inhibi-tion of PAO. Owing to the increase of theSpm concentration, the polyamine-derivedpositive charges more than doubled, com-pared with control cells.

Figure 5. Time course of the cytotoxic e¡ects of spermine and ofcombinations of spermine with 50 mmol/L MDL 72527. 4000CaCo-2 cells were seeded in 96-well microplates. The chemicalswere added to the culture medium on day 7 after seeding. Atvarious time points of incubation the drug-containing mediawere exchanged against fresh medium, and incubation wascontinued for a total of 48 h. The protein content of the wellswas determined at this time. The error bars indicate +SD (n =6); if they are not visible they are smaller than the symbols.

Table 1. Cell cycle phase distribution of exponentially growingCaCo-2 cells. E¡ect of exposure to spermine between 24 h and72 h after seeding

Percentage of cells in

Treatment G1 S G2

None (control) 45.5 23.7 30.8Spermine (5 mmol/L) 28.4* 27.1 44.5*

*Statistically signi¢cant di¡erence (p50.01) between treatedand untreated cells.

124

(c) The combined treatment of the cells withSpm and the PAO inhibitor resulted in aprofound (96%) depletion of Spd, and theaccumulation of N1acSpm above 1 nmol/mg protein. N1acSpd concentration wasnot di¡erent from that of cells exposed tothe PAO inactivator alone.

Di¡erentiation markers

Alkaline phosphatase, sucrase, and amino-peptidase are located in the brush bordermembrane of CaCo-2 cells. They are di¡eren-

tiation markers of these cells (Pinto et al.,1983). Their activities increased very consider-ably during incubation between days 6 and 8after seeding (Table 3). Exposure of the cells to5 mmol/L Spm had no e¡ect on the yield ofbrush border membranes but it reduced theactivities of the brush border membraneenzymes below the pre-treatment values. Thecombination of 1.25 mmol/L Spm + 50 mmol/LMDL 72527 had, in agreement with its lowertoxicity compared with 5 mmol/L Spm, asmaller but highly signi¢cant e¡ect on theevolution of the brush border enzymes.

Table 2. Polyamine concentrations (pmol/mg protein) in con£uent CaCo-2 cells. E¡ects of 48 h exposure to spermine and thepolyamine oxidase inhibitor MDL 72527

Treatment Putrescine N1-Acetylspermidine Spermidine N1-Acetylspermine Spermine

None (control) 40+30 56 2 070+54 50.5 6 797+133

Spermine (5mmol/L) 515* 213+20* 2 020+52 114+18* 16 009+142*

MDL 72527 (50 mmol/L) 517* 39+2* 649+18* 118+3* 8 082+240*

Spermine (1.25 mmol/L)+ MDL 72527 (50 mmol/L) 51* 37+1* 90+6* 1 020+25* 12 737+1745*

Cells were grown for 6 days in 10-ml Petri dishes, then exposed to the chemicals for 48 h, and harvested as described in Materials andMethods; values are mean+SD (n = 3). To determine the amount of extracellular spermine in the cell pellets, controls were cooled onice before harvesting, and were then exposed for 1 min to 1.25 and 5 mmol/L Spm, followed by the usual treatment. The amounts ofextracellular Spm were around 5% of the amount of intracellularly accumulated Spm (not shown). The asterisk indicates a statisticallysigni¢cant di¡erence between treated and control cells (p50.01).

Table 3. Brush border membranes, and alkaline phosphatase, sucrase, and aminopeptidase activities in CaCo-2 cells. E¡ects ofspermine and the polyamine oxidase inhibitor MDL 72527

Alkaline Amino-Brush border Sucrase phosphatase peptidase

membrane protein ööööööööööööööööööööööTreatment (mg/mg cell protein) (mUnits/mg brush border membrane protein)

None (control 6 days) 0.46+0.05 0.7+0.1 76+14 6.8+0.6

None (control 8 days) 0.47+0.01 12.7+0.4 152+4 13.4+0.4

Spermine (5 mmol/L) 0.45+0.03 0.1+0.1** 56+3* 5.2+0.2*

Spermine (1.25 mmol/L)+ MDL 72527 (50 mmol/L) 0.44+0.01 2.5+0.6*6 94+5* 8.2+0.4*

Cells were grown in 10-ml Petri dishes for 6 days, and then exposed to the chemicals for 48 h. After washing and harvesting the cellswere stored at ^808C until enzyme assays were performed; values are mean+SD (n = 3). *Statistically signi¢cant di¡erence betweentreated and 8-day control cells; 6indicates a signi¢cant di¡erence between treated and 6-day control cells (p50.01).

125

Activation of proteases

Eight-day-old cells were treated with Spm,MDL 72527, and combinations of Spm andthe PAO inactivator, as described in the legendto Figure 6. Protease activity was determinedin the cell lysates by measuring the rate of 7-amino-4-tri£uoromethylcoumarine (AFC)release from the £uorogenic peptide carbo-benzoxy-Asp-Glu-Val-Asp-AFC. This andrelated peptides are substrates of caspase-3,and of other members of the caspase family(Nicholson et al., 1995). As shown in Figure 6,protease activity, as measured by hydrolysis ofthe caspase substrate, was only about doubledby 5 mmol/L Spm and Spm (2 mmol/L) +MDL 72527 (50 mmol/L). Addition of theapopain inhibitor acetyl-Asp-Glu-Val-Asp-chloromethylketone (a variation of an inhibitordescribed by Nicholson et al. (1995)) to theassay mixture revealed that the CaCo-2 cellhomogenate contained only 20^25% apopain.

This percentage did not change signi¢cantlywhen the protease activities were enhanced by5 mmol/L Spm (not shown).

Apoptosis-associated DNA fragmentation

The formation of oligonucleosome-sized DNAfragments due to double strand breaks is ahallmark of apoptosis (Wyllie, 1980; Wyllie etal., 1980). In spontaneously detached cells of 8-day-old CaCo-2 cells, typical DNA ladders ofapoptotic cells were detected. However, inCaCo-2 cells, which were detached from theplastic surface after exposure to 5 mmol/LSpm either between days 1 through 3 orbetween days 6 and 8 after seeding, no sig-ni¢cant increase of apoptosis-associated DNAfragmentation was observed.

Discussion

A number of explanations have been o¡eredfor the growth inhibitory and cytotoxic e¡ectsof the polyamines. The formation of hydrogenperoxide and of aldehydes due to oxidativedeaminations of the polyamines by serumamine oxidase (SAO) (Cohen, 1998) has beena frequent source of misinterpretations ofresults in experiments with cell cultures thatcontained ruminant serum and polyamines.(Oxidative deaminations by SAO are analo-gous to those catalyzed by DAO but areexclusive for the aminopropyl moieties (Seiler,1992).) Intracellular formation of hydrogenperoxide due to polyamine oxidation is stillunder consideration as an apoptotic e¡ect ofpolyamine catabolism (Parchment, 1996). Itwas, however, pointed out by Brunton et al.(1991) that high concentrations of Spm exertdirect toxic e¡ects on the cells. He et al. (1993)demonstrated that inhibition of cell growthdue to accumulation of polyamines in mouseFM3A cells correlates with the decrease ofMg2+ and ATP, and the consequent impair-

Figure 6. Protease activity of CaCo-2 cells: e¡ect of exposurebetween days 6 and 8 after seeding to 5 mmol/L spermine, thePAO inhibitor MDL 72527, and a combination of thesecompounds. Protease activity was determined by the release of7-amino-4-tri£uoromethylcoumarin (AFC) from the caspasesubstrate carbobenzoxy-Asp-Glu-Val-Asp-AFC. The errorbars indicate +SD (n = 3).

126

ment of protein synthesis. The activation ofcaspases in leukemia cells by Spm (Stefanelliet al., 1998) hinted at a toxic mechanism that isdirectly related to apoptosis, because caspasesare a family of cysteine proteases responsiblefor many characteristics of apoptotic cell death(Thornberry et al., 1997; Haunstetter andIzumo, 1998). The mechanism of caspaseactivation by Spm has not yet been clari¢ed.If cells were exposed for 48 h, the resistance

of CaCo-2 cells to Spm was remarkably high(Figure 4). In contrast with FM3A cells, whichaccumulated under comparable conditionsexclusively in G1 (He et al., 1993), CaCo-2 cellstended to accumulate in the G2 phase of thecell cycle. Evidently di¡erent cell types reactdi¡erently to Spm accumulation. Sincecalmodulin is irreplaceable in G1/S and G2/M progression (Rasmussen and Means, 1989),e¡ects of high intracellular concentrations ofSpm on cell cycle progression and toxicitymight be due to its calmodulin antagonistfunction (Walters and Johnson, 1988).The emphasis of our work was on the role of

PAO in Spm cytotoxicity. The increase in PAOactivity during cell maturation was onlymodest. Exposure to Spm had no e¡ect onPAO activity. These observations suggest thatPAO is a constitutive enzyme of CaCo-2 cells.PAO activity is a plausible reason for the low

cytotoxicity of Spm, as appears from thecomparison of cells with active and inactivatedPAO. In cells with active PAO, depletion ofSpd was prevented, most probably owing to itsformation from both Spm and N1acSpm(Figure 1). However, the metabolic trans-formation of Spm to Spd in CaCo-2 cells wasobviously not fast enough to prevent theexcessive accumulation of Spm within the cells.It is proposed that at least a part of the Spmtoxicity can be explained by the occupation ofSpd and Mg2+ binding sites by Spm, whichprevents physiological functions of Spd andMg2+, e.g., in protein synthesis and othergrowth related processes. The inability of cells

to grow with normal or slightly elevated Spmpools, but their inability to grow with Put andSpd pools selectively depleted by 2-(di£uoro-methyl)ornithine, has been extensivelydocumented (McCann et al., 1987). It isevidence in favour of important roles of Spdin growth-related processes.

The PAO activity of CaCo-2 cells is lowerthan the activities reported of other humancancer cells (Flahey and Wallace 1990;Lamond and Wallace 1994). Since the di¡er-ence in PAO activity between noncon£uentand di¡erentiating cells was small, PAOactivity is presumably only one among severalreasons for the di¡erence in sensitivity to Spmexposure between days 1 to 3 and days 6 to 8.Other parameters, such as di¡erences of trans-port rates, and of vital metabolic activities(protein, DNA, and RNA synthesis) are pre-sumably contributing as well. Furthermore,MDL72527 appears to increase Spm cyto-toxicity not only by selective inactivation ofPAO, but in addition by a direct cytotoxice¡ect, as became obvious from long-termincubations with this agent (Figure 3).Recently an apoptotic e¡ect of 150 mmol/LMDL 72527 on transformed hematopietic cells(but not on myeloid progenitors) has beenreported (Dai et al., 1999).

The observed e¡ects of PAO inactivation byMDL 72527 on the polyamine pattern ofcon£uent CaCo-2 cells are in agreement withexpectations, and can be explained as follows:The decrease of Put indicates that most of thePut in CaCo-2 cells is formed from Spd, not bydecarboxylation of ornithine. The increase ofSpm is due to the fact that at low Put concen-trations spermidine synthase does not competewith spermine synthase for the limited amountof the decarboxylation product of S-adenosyl-methionine, so that Spm formation prevails.(For short reviews of the principles of poly-amine metabolism and regulation see, e.g.,Seiler, 1987, 1990; Pegg, 1989).

The increase of N1acSpd and N1acSpm

127

concentrations in the presence of 5 mmol/LSpm is indicative of the induction of thecytosolic spermidine/spermine N1-acetyl-transferase, and a high rate of polyamineacetylation. The Spd which is displaced bySpm from its binding sites, or formed fromSpm, is exposed to acetylation and elimination.There may be several reasons why Put is alsolow: (a) Since the concentration of Spm is high,and since N1acSpm is a better substrate ofPAO than N1acSpd (Bolkenius and Seiler,1981), oxidative splitting of N1acSpd may becompetitively inhibited. (b) The cells mayexcrete Put because Spm and Spd have ahigher binding a¤nity than Put and displaceit from anionic binding sites.Spd is a poor substrate of PAO but Spm is a

reasonably good substrate (Ho« ltta« , 1977; Bolk-enius and Seiler, 1981; Seiler et al., 1995). Athigh PAO activity the formation of cytotoxichydrogen peroxide and of 3-aminopropanal(and its acetyl derivative) should be enhanced,if intracellular Spm concentrations increase,especially since N1acSpd and N1acSpm arerapidly formed and are the best substrates ofPAO (Bolkenius and Seiler, 1981). If the intra-cellular formation of the above cytotoxicme t abo l i t e s wa s a maj o r c au s e o f(programmed) cell death, as was postulated(Parchment, 1996), Spm should be more toxicin the absence than in the presence of the PAOinactivator. However, the opposite is true.The presence of 50 mmol/L or 1 mmol/L

aminoguanidine had only a small cyto-protective e¡ect to intoxication by Spm,although CaCo-2 cells exhibit an importantDAO activity (D'Agostino et al., 1989). Itfollows that in CaCo-2 cells oxidativedeaminations contribute little to Spm cyto-toxicity, in agreement with previous experiencewith another cell line (Brunton et al., 1991).The intracellular inactivation of hydrogenperoxide and of the aldehydes by catalase andaldehyde dehydrogenase, respectively, isobviously su¤ciently fast to prevent cell death.

The activities of the di¡erentiation markerenzymes increase in CaCo-2 cells veryconsiderably between days 6 and 8 of incuba-tion. This increase was completely preventedby 5 mmol/L Spm; the activity of sucrase waseven lower than the pre-treatment value (Table3), while no change was observed in theamount of brush border protein. Mostprobably, Spm retards di¡erentiation. Anotherpotential explanation for the low activity of thebrush border enzymes is that Spm impairstheir synthesis and/or their processing, withouthaving other e¡ects on the phenotype. He et al.(1993) had reported a reduced proteinsynthesis rate in cells exposed to high Spmconcentrations. Whether the e¡ect of Spm onprotein synthesis is general or selective forcertain proteins is for future study.

In contrast to L1210 leukemia cells (Stefa-nelli et al., 1998) caspase activation in CaCo-2cells by Spm was unimpressive. Moreover, nosigni¢cant enhancement of apoptosis-asso-ciated DNA fragmentation was observed indetached cells, compared with spontaneouslydetached control cells. Thus it appears that theexcessive accumulation of Spm in CaCo-2 cellscauses mainly nonapoptotic cell death.

Colon carcinomas have signi¢cantly lowerPAO activities than the surrounding tissue(Linsalata et al., 1997). Based on our observa-tions, and those of others (Brunton et al., 1991;Stefanelli et al., 1998) it is to be expected thatcytotoxic substrates of PAO (of which Spm isan example) preferentially kill tumor cells.Selective toxicity should be further enhancedif the cytotoxic compounds are substrates ofpolyamine uptake systems, since tumor cellsusually exhibit higher polyamine transportrates than the analogous nontransformed cells(Seiler and Dezeur, 1990; Seiler et al., 1996).Owing to its nephrotoxicity and centralnervous system toxicity (Seiler, 1991), Spm isnot a suitable candidate for testing this tumortargeting principle in animal models, but itmay serve as a lead compound.

128

References

Bey P, Bolkenius FN, Seiler N, Casara P. N-2,3-Butadienyl-1,4-butanediamine derivatives: potent irreversible inactivatorsof mammalian polyamine oxidase. J Med Chem. 1985;28:1^2.

Blais A, Bissonnette B, Berteloot A. Common characteristicsfor Na+-dependent sugar transport in CaCo-2 cells andhuman fetal colon. J Membr Biol. 1987;99:113^25.

Bolkenius FN, Seiler N. Acetylderivatives as intermediates inpolyamine catabolism. Int J Biochem. 1981;13:287^92.

Bolkenius FN, Seiler N. Polyamine oxidase inhibitors. In:Sandler M, Smith HJ, eds. Design of enzyme inhibitors asdrugs. Oxford: Oxford University Press: 1989:245^56.

Brunton VG, Grant MH, Wallace H.M. Mechanism ofspermine toxicity in baby hamster kidney (BHK) cells.Biochem J. 1991;280:193^8.

Cohen SS. A guide to the polyamines. New York: OxfordUniversity Press; 1998:69^93.

D'Agostino L, Daniele B, Pignata S, et al., Ornithinedecarboxylase and diamine oxidase in human coloncarcinoma cell line CaCo-2 in culture. Gastroenterology.1989;97:888^94.

D'Agostino L, Pignata S, Daniele B, et al., Polyamine uptake byhuman colon carcinoma cell line CaCo-2. Digestion.1990;46(supplement 2):352^9.

Dai H, Kramer DL,Yanag C, Murti KG, Porter CW, ClevelandJL. The polyamine oxidase inhibitor MDL 72527 selectivelyinduces apoptosis of transformed hematopoietic cellsthrough lysosomotropic e¡ects. Cancer Res. 1999;59:4944^54.

Duranton B, Keith G, Gosse F, Bergmann C, Schlei¡er R, RaulF. Concomitant changes in polyamine pools and DNAmethylation during growth inhibition of human coloncarcinoma cells. Exp Cell Res. 1998;243:319^25.

Flahey KA,Wallace HM. Polyamine oxidase activity in humancolonic carcinoma cell line. Biochem Soc Trans. 1990;18:1225.

Fogh J, Fogh JM, Orfeo T. One hundred and twenty sevencultured human tumor cell lines producing tumors in nudemice. J Natl Cancer Inst. 1977;59:221^6.

Garen A, Levinthal C. A ¢ne structure genetic and chemicalstudy of the enzyme alkaline phosphatase of E. coli. 1.Puri¢cation and characterization of alkaline phosphatase.Biochim Biophys Acta. 1960;38:470^83.

Grasset E, Pinto M, Dussaulx E, Zweibaum A, Dejeux JF.Epithelial properties of human colonic carcinoma cell lineCaco-2: electrical parameters. Am J Physiol. 1984;247:C260^7.

Haunstetter A, Izumo S. Apoptosis: basic mechanisms andimplications for cardiovascular disease. Circ Res. 1998; 82:1111^29.

Hayashi H, Yoshida H, Hashimoto F, Okazeri, S. Changes inpolyamine-oxidizing capacity of peroxisomes under variousphysiological conditions in rats. Biochim Biophys Acta.1989;991:10^16.

He Y, Kashiwagi K, Fukuchi J, Terao K, Shirahata A, IgarashiK. Correlation between the inhibition of cell growth by

accumulation of polyamines and the decrease of magnesiumand ATP. Eur J Biochem. 1993;217:89^96.

Ho« ltta« E. Oxidation of spermidine and spermine in rat liver:puri¢cation of polyamine oxidase. Biochemistry. 1977;16:91^100.

Hu RH, Pegg AE. Rapid induction of apoptosis by deregulateduptake of polyamine analogues. Biochem J. 1997;328:307^16.

Ientile R, Ginoprelli T, Cannavo G, Picerno I, Piedimonte G.E¡ect of beta-endorphin on cell growth and cell death inhuman peripheral blood lymphocytes. J Neuroimmunol.1997;80:87^92.

Lamond S, Wallace HM. Polyamine oxidase activity in humancancer cells. Biochem Soc Trans. 1994;22:396S.

Linsalata M, Cavallini A, Di Leo A Polyamine oxidase activityand polyamine levels in human colorectal cancer and innormal surrounding mucosa. Anticancer Res. 1997;17:3757^60.

Lowry OG, Rosebrough NJ, Farr AL, Randall RJ. Proteinmeasurements with the Folin phenol reagent. J Biol Chem.1951;193:265^75.

Maroux S, Louvard D, Baratti J. The aminopeptidase from hogintestinal brush border. Biochim Biophys Acta. 1973;321:282^95.

McCann PP, Pegg AE, Sjoerdsma A, eds. Inhibition of poly-amine metabolism. Orlando: Academic Press; 1987.

Messer M, Dahlqvist A. A one-step ultramicromethod for theassay of intestinal disaccharidases. Anal Biochem. 1966;193:265^75.

Nicholson DW, Ali A, Thornberry NA, et al. Identi¢cation andinhibition of the ICE/CED-3 protease necessary formammalian apoptosis. Nature. 1995;376:37^43.

Parchment, R.E. An overview of the known and suspectedmechanisms for control of apoptosis by polyamines andtheir metabolites. In: Nishioka K, ed. Polyamines in cancer:basic mechanisms and clinical approaches. New York:Springer; 1996: 99^150.

Pegg AE. Polyamine metabolism and its importance in neo-plastic growth and as a target for chemotherapy. CancerRes. 1989;48:759^74.

Pinto M, Robin-Leon S, Appay MD, et al. Enterocyte-likedi¡erentiation and polarization of the human coloncarcinoma cell line CaCo-2 in culture. Biol Cell. 1983;7:323^30.

Poulin R, Pelletier G, Pegg AE. Induction of apoptosis byexcessive polyamine accumulation in ornithine decarboxy-lase-overproducing L1210 cells. Biochem J. 1995;311:723^7.

Rasmussen CD, Means AR. Calmodulin, cell growth and geneexpression. TINS. 1989;12:433^8.

Rousset L. The human colon carcinoma HT-29 and Caco-2:two in vitro models for the study of intestinal di¡erentiation.Biochimie. 1986;68:1035^40.

Schmitz J, Preiser H, Maestracci D, Ghosh BK, Cerda JJ,Crane RK. Puri¢cation of human intestinal brush bordermembrane. Biochim Biophys Acta. 1973;323:98^112.

Seiler N. Functions of polyamine acetylation. Can J PhysiolPharmacol. 1987;65:2024^35.

Seiler, N. Polyamine metabolism. Digestion. 1990;46 (supple-ment 2):319^30.

129

Seiler N. Pharmacological properties of the natural polyaminesand their depletion by biosynthesis inhibitors as atherapeutic approach. In: Jucker E, ed. Progress in drugresearch. Basel: BirkhÌuser; 1991:107^59.

Seiler N. Polyamine catabolism and elimination by thevertebrate organism. In: Dowling RH, Fo« lsch UR, Lo« serC, eds. Polyamines in the gastrointestinal tract. Dordrecht:Kluwer Academic Publishers; 1992:65^85.

Seiler N. Polyamine oxidase, properties and functions. In: Yu P,Tipton KF, Boulton AA, eds. Current neurochemical andpharmacological aspects of biogenic amines. Amsterdam:Elsevier; 1995:333^44. (Progress in brain research; vol 106).

Seiler N, Dezeure F. Polyamine transport in mammalian cells.Int J Biochem. 1990;22:211^18.

Seiler N, Bolkenius FN, Kno« dgen B, Mamont P. Polyamineoxidase in rat tissues. Biochim Biophys Acta. 1980;615:480^8.

Seiler N, Sarhan S, Grau¡el C, Jones R, Kno« dgen B, Mouli-noux JP. Endogenous and exogenous polyamines in supportof tumor growth. Cancer Res. 1990;50:5077^83.

Seiler N, Moulinoux JP, Havouis R, Toujas L. Characterizationof amine oxidase activities in macrophages from humanperipheral blood. Biochem Cell Biol. 1995;73:275^81.

Seiler N, Delcros JG, Moulinoux JP. Polyamine transport inmammalian cells. An update. Int J Biochem Cell Biol. 1996;28:843^61.

Skehan P, Storeng R, Scudiero D, et al. New colorimetriccytotoxicity assay for anti-cancer drug screening. J NatlCancer Inst. 1990;82:1107^12.

Stefanelli C, Bonavita F, Stanic I, et al. Spermine causescaspase activation in leukaemia cells. FEBS Lett. 1998;437:233^6.

Thornberry NA, Rosen A, Nicholson D. Control of apoptosisby proteases. Adv Pharmacol. 1997;41:155^77.

Walters JD, Johnson JD. Inhibition of cyclic phosphodiesteraseand calcineurin by spermine, a calcium independentcalmodulin antagonist. Biochim Biophys Acta. 1988;957:138^42.

Wyllie AH. Glucocorticoid-induced thymocyte apoptosis isassociated with endogenous activation. Nature. 1980;284:555^6.

Wyllie AH, Kerr JRF, Curie AR. Cell death: the signi¢cance ofapoptosis. Int Rev Cytol. 1980;68:251^306.

Xie X, Tome M, Gerner EW. Loss of intracellular putrescinepool-size regulation induces apoptosis. Exp Cell Res. 1997;230:386^92.

Address for correspondence: Dr N Seiler, CJF INSERM 95-09,Institut de Recherche Contre les Cancers de l'Appareil Digestif(IRCAD), 1 place de l'hoª pital, BP 426, 67091 Strasbourg,Cedex, France.

130