[CANCER RESEARCH56, 4936-4941. November 1. 1996]

ABSTRACT

Sphingolipids are hydrolyzed in the gastrointestinal tract to ceramide,sphingosine, and other metabolites that can modulate cell growth, differentiation, and apoptosis. To characterize the effects of dietary sphingolipids on colon carcinogenesis, female CF1 mice were administered 1,2-dimethylhydrazine and then fed an essentially sphingolipid-free dietsupplemented with 0 to 0.1% (w/w) sphlngomyelin (SM) purified from

milk. As was found in a previous pilot study (D. L. Dillehay etaL, J. Nutr.,124: 615—620,1994), SM (@ 0.1%) reduced the number of aberrantcolonic crypt foci (by 70%, P < 0.001) and aberrant crypts per focus (by30%, P < 0.003), which are early indicators of colon carcinogenesis. Inlonger term studies, SM had no effect on colon tumor incidence ormultiplicity; however, up to 31% of the tumors of mice fed SM were

adenomas, whereas all of the tumors of mice fed the diet without SM wereadenocarcinomas. These findings demonstrate that milk SM suppressesthe appearance of more advanced, malignant tumors as well as earlymarkers of colon carcinogenesis. Although the sphingolipid content offoods has not been widely studied, several foods (e.g., milk and soybeans)contain the sphingollpid levels used in these investigalions therefore, thisdass of compounds could be significant contributors to the cancer pre

ventive effects of some foods.

INTRODUCTION

Colon cancer is the second leading cause of cancer mortality in theUnited States. Genetic predisposition and environmental factors, suchas diet, are thought to affect colon carcinogenesis, and a large numberof studies have explored the relationships between specific nutrientsand colon cancer (reviewed in Refs. 1—6).However, relatively few ofthe thousands of compounds in food have been evaluated for theireffects on carcinogenesis, which may account for the current confusion regarding the relationships between diet and cancer.

Sphingolipids are prominent among the components of food thatwarrant investigation because they modulate processes that are important in carcinogenesis: cell growth, differentiation, and apoptosis(7, 8). The sphingolipid backbones (sphingosine, sphmgosine 1-phosphate, sphingosylphosphorylcholine, ceramide, and ceramide l-phosphate) are highly bioactive compounds that have been shown instudies with cells in culture to stimulate (9) or inhibit (10, 11) cellgrowth, enhance (12) or inhibit (13) cell differentiation, and inducecell death (10), in some cases via apoptosis (14). Sphingolipids havesuch a wide range of activities because they are involved in cellstructure, cell-cell and cell-substratum interactions, and as second

Received 7/2/96; accepted 9/3/96.Thecostsof publicationof thisarticleweredefrayedin partby thepaymentof page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

I Supported by funds from the National Dairy Board administered in cooperation with

the National Dairy Council and National Cancer Institute Grant CA61820.2 To whom requests for reprints should be addressed, at Department of Biochemistry,

41 13 Rollins Research Center, Emory University, Atlanta, GA 30322. Phone: (404)727-5978; Fax: (404) 727-3954.

messengers in the signaling pathways for growth factors, cytokines,neurotrophic factors, and other agents (15).

In addition to these intracellular signaling pathways, cells of allregions of the gastrointestinal tract are exposed to ceramides, sphingosine, and other bioactive metabolites during the digestion of dietarysphingolipids (16). To determine if dietary sphingolipids affect intestinal cells, and especially, if they alter the response of colonic cells tocarcinogens, SM3 was purified from powdered milk4 and fed tofemale CF1 mice that had been treated with DMH to induce colontumors (17). SM reduced the number of aberrant colonic foci (an earlymarker of colon carcinogenesis) and decreased the number of animalswith tumors; however, the reduction in tumor incidence was onlymarginally significant (P = 0.08), perhaps because too little SM wasavailable for a thorough feeding study.

In the follow-up investigations described in this report, we haveused a preparation of SM that is available commercially (purifiedfrom burtermilk) to establish more conclusively that feeding SM toDMH-treated CF1 mice reduces the number of aberrant colonic cryptfoci and the number of crypts per focus. The consumption of SM wasalso discovered to shift the types of tumors induced by DMH fromexclusively adenocarcinomas to a combination of adenomas and adenocarcinomas. These findings provide further evidence that thismajor class of dietary lipids could play a role in the associationsbetween diet and cancer.

MATERIALS AND METHODS

Animals. Female CF1 mice (5 weeks of age and virus antibody free) wereobtained from Charles Rivers Laboratories (Portage, MI) and kept at23°C±2°Cwith 50—60%relative humidity and a 12-h light/dark cycle. Micewere housed in microisolator cages (five per cage) with hardwood chipbedding and had free access to water and, for the first week (for acclimatiza

tion), were fed Purina Rodent Chow 5001 (Ralston Purina, St. Louis, MO) adlibitum. The animals were weighed weekly and monitored closely for signs of

disease. All experimental protocols involving animals were approved by theInstitutional Animal Care and Use Committee and conducted according to

National Research Council guidelines (18).Sphingomyelin. SM was isolated from powdered, low-fat milk as de

scribed previously (16) and from buttermilk (uncultured); the latter was purchased from Matreya, Inc. (Pleasant Gap, PA). The two SM preparations werecompared by thin-layer chromatography and high-performance liquid chromatography (16) and by mass spectrometry using a JEOL JMS-SX1O2/

SX1O2A/E, five-sector, tandem (MS1-MS2-MS3) mass spectrometer (19, 20).

Full-scan negative ion FAB mass spectra were acquired using MS1 andfrit-FAB in which the solvent was 2:1 CHCI3/methanol containing 1% trieth

anolamine. Precursor ions were produced by bombarding the samples with6-keV xenon atoms at a gun emission current of 5 mA, which were thenaccelerated to 10 keV. Tandem mass spectrometric experiments (MS-MS

3 The abbreviations used are: SM, sphingomyelin; DMH, l,2-dimethylhydrazine;

FAB, fast atom bombardment; MS. mass spectrometry.4 Sphingolipids can be isolated from powdered milk because they are mainly part of the

membranes rather than the milkfat droplet, and a large portion is retained in “lowfat―milkandmilkproducts.

4936

Sphingomyelin Consumption Suppresses Aberrant Colonic Crypt Foci and Increasesthe Proportion of Adenomas versus Adenocarcinomas in CF1 Mice Treated with1,2-Dimethythydrazine: Implications for Dietary Sphingolipids and ColonCarcinogenesis'

Eva M. Schmelz, Dirck L. Dilehay, Sonji K. Webb, Alex Reiter, Jeanette Adams, and Alfred H. Merrill, Jr.2Departments ofBiochemistry fE. M. S.. A. H. MI. Pathology ID. L. Di, and Chemistry and the Enwrv Mass Spectrometry Center (A. R., J. A.), and Division ofAnimal Resources

(D. L D., S. K. WI. Emory University, Atlanta, Georgia 30322

@A 687.@-@ 616@ 771

.@ 714

gi@ 642@@ @799

@ @4A@A44A4@ AAIIkk@600 . .700 ‘800

@.i 687i@ 728C@ 616 700 771@U@ 714@@@ 785

I @:AAi4A41k@700 ‘ ‘ ‘

m/z 800

INHIB@ON OF COLON CARCINOGENESIS BY SPHINGOMYELIN

experiments) were performed by mass- and energy-selecting the precursor ionsusing MS1 and then colliding the ions with helium in a collision cell locatedbetween MS1 and MS2. The MS-MS spectra were acquired by linking MS2and MS3 together in a scan in which the magnets and electrostatic analyzerswere held at a constant ratio of magnetic field strength to electrical fieldstrength (B:E).

Experimental Diets and Carcinogen Treatment. The mice were randomly divided into the experimental groups and fed a semipurified AIN 76Adiet (Ralston Purina) after 7 days of acclimatization. The sphingolipid contentof this diet is <0.005% (wlw; Ref. 17) and serves as an essentially sphingolipid-free control. l,2-Dimethylhydrazine2 HC1(DMH) was purchased fromSigma Chemical Co. (St. Louis, MO) and diluted before each injection in 1 mtviEDTA. Fifty mice/group (in the carcinogenesis study; 15/groupin the aberrantcrypt study) were injected i.p. with 0.5 ml of DMH (20 mg of DMH/kg body

weight in the carcinogenesis study; 40 mg/kg for the aberrant crypt study) oncea week for 6 weeks. These dosages of DMH are comparable to previous studiesand were selected to target a tumor incidence of approximately 50% (21, 22).One week after the fmal injection of DMH, the diets were supplemented with0, 0.025, 0.05, and 0.1% SM (by weight). The SM was ground into a finepowder and thoroughly mixed into the AIN 76A diet at the stated percentages,

which were confirmed by high-performance liquid chromatography (23). Thediets were prepared monthly and stored at 4°C(SM is stable under theseconditions).

Analyses of Aberrant Colonic Crypts. After 4 weeks of feeding theexperimental diets (5 weeks after the last injection of DMH), the mice werekilled by CO2 asphyxiation, and the large intestines were removed, flushedwith cold Tris [0.05 MTris (hydroxymethyl) amino methane-HC1,pH 7.6, at4°C],opened longitudinally, fixed overnight in 10%neutral buffered formalin,and stained with 0.2% methylene blue for 20 to 40 rain. Aberrant crypt foci andthe number of crypts per focus were measured by light microscopy at 40- or100-fold magnification (24).

Carcinogenesis Study. After 34 weeks of feeding, blood was found inalmost all of the feces sampled; therefore, all the animals were killed by CO2

asphyxiation.The largeintestineswereremoved,flushedwithcoldTris (0.05M, pH 7.6 at 4°C), and opened longitudinally; the tumor number, location, and

size were noted. The tumors were evaluated histologically after being fixed in

formalin, embedded in paraffin, sectioned at 5 @m,and stained with H&E. Ina blinded fashion, the tumors were classified as adenomas or adenocarcinomas,as described by Elsayed and Shamsuddin (25). Serum was collected from the0 and 0.1% SM groups and analyzed for standard clinical chemistry parametersby a commercial lab.

Statistical Analyses. Statistical analyses were conducted using the Instatsoftware (version 1.0; Instat, San Diego, CA). Differences in the weight andnumber of aberrant foci and tumors were evaluated by Student's t test afterANOVA analysis. The data on tumor composition comparisons were evaluatedby Fisher's exact test. Regression analyses were used for the calculation ofcorrelations. Differences were considered significant at P 0.05.

RESULTS

Characterization of the SM Used in These Studies. As part ofthe follow-up to our previous study (17), the molecular subspecieswere analyzed for the SM isolated by our laboratory from powderedmilk (16) and a preparation from buttermilk (uncultured) that isavailable commercially (Matreya, Inc.). The mass spectrometric analyses of these samples are shown in Fig. 1, A and B. All of themolecular species from both samples were also collisionally activatedin MS-MS experiments to confirm that they were SMs and to identifythe components of isomers that produce the same m/z. One exampleof a MS-MS spectrum of the precursor ion of m/z 616 from thefull-scan FAB spectrum of the buttermilk sample is shown in Fig. 1C.This MS-MS spectrum was easily interpreted to be the spectrum of the(M-86) ion ofN-palmitoyl-(4E)-sphingosyl phosphorylcholine (d18:1/16:0; see structure in Fig. 1D) plus small amounts of the M-60fragment ion for the d18:l/14:1 homologue (19). As has been reportedpreviously (26), milk SM is composed of mostly sphingosine (d18:l)withsaturatedfattyacids(16:0,m/z616;22:0,nilz700;23:0,m/z714;and 24:0, m/z 728). Within each cluster there are also small amountsof SM with sphinganine (d18:0) rather than sphingosine (d18:1). Thehigher m/z ions are mainly from the M-60 and M-15 fragmentations of

a)UC(5

C-C(5C0

a)>Cs

a)

C

(M-86)

C

i@I

D

600

Fig. 1. Full-scan negative ion FAB mass spectra of SMs from powdered milk (A) and buttermilk (B). C, the high-energy MS-MS spectrum of precursor ion of m/z 616 from thefull-scan FAB spectrom of SMs in buttermilk (see B). The structure shown in D was deduced directly from interpretation of the MS-MS spectrum (19).

4937

0 LG/K

.1HL.@ I 1

0 100 200 300 400 500 600

0

HOf@L@@

OH

@0 HN.C I

K L M

]

A

L@I1ri.1B

INHIB@ON OF COLON CARCINOGENESIS BY SPHINGOMYELIN

and ranged in size from 1—5mm in diameter. Macroscopically, thetumors were sessile or pedunculated; microscopically, 57 colonictumors were observed: 10 benign polypoid adenomas, as indicated bythe presence of an intact muscularis mucosa (Fig. 4A); and 47 adenocarcinomas (usually large), which were characterized by invasionof neoplastic cells through the muscularis mucosa (Fig. 4B).

The difference between the mice fed SM-containing diets versusthe control surfaced when the distribution of tumor types was cornpared (Fig. 5). All of the tumors of the mice fed the control diet wereadenocarcinomas, whereas every group fed SM had a combination ofadenomas and adenocarcinomas. This shift in the proportion of adenomas versus adenocarcinomas was significant (P = 0.043) for the0.05% SM group and marginally significant (P = 0.075) for the0.025% SM group.

DISCUSSION

The results of these studies support the hypothesis that dietarysphingolipids suppress colon carcinogenesis, perhaps through theirhydrolysis to the bioactive backbones. The ability of sphingoid basesto affect transformation was initially proposed by Hannun et al. (27),based on the discovery that sphingosine is a potent inhibitor of proteinkinase C. Borek et a!. (28, 29) next found that sphingosine andsphinganine inhibit the induction of transformed foci in C3H 10T½cells exposed to -y-irradiation and phorbol esters, and many types ofcells in culture are now known to undergo change in growth, differentiation, and apoptosis upon exposure to various sphingolipids (8,30). In the first in vivo studies (3 1, 32), topical application of sphingosine to mouse skin inhibited the induction of ornithine decarboxylase by phorbol esters, a marker of tumor promotion. Subsequentexperiments did not find that sphingosine blocked the development ofskin tumors (33), but it is possible that the high amounts of sphingosine and ceramide that are already in skin (34) complicate the use ofthe skin tumor model. The administration of a number of sphingolipids i.p. or iv. have been shown to block the growth and metastasis of

r@-ir@@iControl Powdered Buttermilk

milk SM SM

Fig. 3. Effect of SM on aberrant crypt foci (A) and number of aberrant crypts per focus(B). The significance of the differences from the control were: ***, P < 0.001; and **,P < 0.01.

60@

50.

.—

@ 40@

0

30@

—0-— 0% SM

S 0.025% SM

. 0.05% SM

—0— 0.1% SM

0 10 20 30

Time course (weeks)Fig. 2. Weights shown are the means (n = 50/group) of mice fed diets supplemented

with varying amounts of SM in the shown percentages of the diet (w/w); bars, SE.

this same series (i.e., m/z 642, 726, 740, and 754 for M-60, and m/z687, 771, 785, and 799 for M-l5). Thus, the MS-MS analyses provided unequivocal structure assignments and confirmed that bothsamples were composed of identical molecular species.

The SM preparations were also pure by thin-layer chromatographyand analyses of the sphingoid base backbones by high-performanceliquid chromatography (data not shown), which confirmed that sphingosines (d18:l ; rather than sphinganines, dl8:0) were the majorsphingoid base backbones.

General Characteristics of the Mice Fed SM. The mice fedvarying amounts of SM (from buuermilk) for a long term exhibited nodifferences in weight gain (Fig. 2). The pause in growth during the10th week was due to a problem in the handling of the water bottles,but the animals quickly returned to their normal weight without anysigns of illness after this was corrected. At the end of the study, thebody weights were 53.2 ± 1.3, 51.1 ± 1.0, 51.6 ± 1.1, and53.1 ± 1.6 g (mean ±SE) for the groups supplemented with 0,0.025%, 0.05%, and 0.1% SM, respectively. Analyses of blood collected from mice in the 0 and 0.1% SM groups at the end of the studydid not reveal differences in blood urea nitrogen, creatinine, glucose,serum glutamate oxaloacetate transaminase, serum glutamate pyruvate transaminase, cholesterol, alkaline phosphatase, total protein,albumin, globulin, or electrolytes (data not shown). Thus, as in ourprevious study (17), SM feeding did not appear to have deleteriouseffects on the animals.

Effects of Dietary SM on the Appearance of Aberrant ColonicCrypts. At 0.1% of the diet, buttermilk and powdered milk SMreduced the appearance of aberrant colonic crypt foci by 66 and 70%,respectively (P < 0.001; Fig. 3). There were approximately 30%fewer aberrant crypts per colonic focus in both SM groups(P = 0.0035 for milk powder SM and P = 0.0003 for buttermilk SM).Most of the aberrant crypts were localized in the distal part of thecolon, some were in the middle, and only a few were detected in theanterior portion (and these were seen only in the 0% SM group). Therewas no correlation between the weight of mice and the number ofaberrant crypts (data not shown).

Effect of SM on DMH-induced Colon Carcinogenesis. Allgroups had the same tumor incidence and multiplicity (24.5, 19.6,25.0, and 25.5%, and 1.3, 1.3, 1.3 and 1.2 tumors per tumor-bearinganimal for 0, 0.025, 0.05, and 0.1% SM, respectively). The colonictumors were all found in the middle and distal portions of the colon

20

10

0

H

z

2

4938

INHIBITION OF COLON CARCINOGENESIS BY SPHINGOMYELIN

A B

I•ig.4. 1Iitniici@.@itpli 1 tilL t\@ )t @@I()fl1LtU11@O1@.@.@‘@Ionic@ cntiinin@ t pI\ p@1 .idcnoni.i \@tb in Intt@t111u@@uIt!l\iuiic@.t@ @1@ it •I( (). I@.inwith invasion (arrow) through the muscularis mucosa (Al) at 100-fold magnification.

kf@:4@@ - C.>@(_@-. - . . . .

implanted tumors (35—37).As far as we are aware, our pilot experiments (17) constitute the only previous investigation of the effects ofdietary sphingolipids on cancer.

The current investigation was possible because SM (from butter

a

milk) has been recently made available commercially in the largequantities (>100 g) that are needed for long-term feeding studies.This preparation is structurally identical to the SM that we isolatedfrom powdered milk and caused the same percentage of reduction inaberrant colonic crypts at 0.1% SM (66%) as the preparation frompowdered milk (70%), which also agrees well with the 50% reductionby 0.05% SM in the previous feeding experiment (17). Furthermore,the SM preparations are highly pure (based on MS-MS and otheranalyses); therefore, their effects are likely to be due to SM per se andnot a contaminant of the preparations.

Aberrant colonic crypt foci are one of the earliest morphologicalchanges of the rodent colonic cells to dysplasia. It is thought that apercentage of these lesions will eventually develop into adenomas andadenocarcinomas (38—40),which makes them useful biomarkers forstudies of agents that may inhibit (or enhance) colon carcinogenesis.Nonetheless, caution must be exercised in extrapolating findings withthis biomarker because several agents that have affected the numberof aberrant crypt foci have not had the same effect on development oftumors (41, 42). It has been suggested that the number of aberrantcrypts per focus correlates better with the subsequent development oftumors (42). Our studies with SM have found significant reductions inboth the number of aberrant colonic crypt foci and the number ofaberrant crypts per focus.

Unlike our pilot study, which indicated that SM decreased thenumber of tumors induced by DMH, this investigation found that SMincreased the portion of tumors that are histologically characterized as

4939

Adenomas

5.

0@

5,

10

15

0 0.025 0.05 0.1

% Sphingomyelin in diet

Fig. 5. Number of tumors classified histologically as adenomas or adenocarcinomas inCF1 mice treated with DMH and fed SM in the diet at percentages (w/w) shown. Thestatisticalsignificanceofthedifferencesbetweenthedesignatedgroupsrelativeto the0%SM control were: *, P < 0.08; and **, P < 0.05.

Adenocarcinomas

INHIBITION OF COLON CARCINOGENESIS BY SPHINGOMYELIN

Intestinal Colonic cellLumen

hydrolyzed in the colon (perhaps, involving intestinal microflora) tothe lipid backbones ceramide and sphingosine (and possibly sphingosylphosphorylcholine, lyso-SM), and these are taken up by coloniccells (16). These sphingolipids and their rnetabolites are known toaffect a wide spectrum of cellular targets and cell behaviors, from theregulation of growth and differentiation to apoptosis. The immediatetargets of these sphingolipids are still being elucidated (recentlyreviewed in Refs. 8 and 15), but they have been suggested to includegrowth inhibition due to the inhibition of protein kinase C (10, 27) andinduction of pRb dephosphorylation by sphingosine (48) or ceramide(14); inhibition of growth and induction of apoptosis by activation ofphosphoprotein phosphatase(s) (49) and kinases (50) by ceramide;and increases in intracellular calcium (and growth stimulation) bysphingosine, sphingosine 1-phosphate, and sphingosylphosphorylcholine (9, 47, 5 1). Therefore, the simplest explanation for the findings ofour two studies is that the sphingosine and/or ceramide derived fromdietary SM suppress the development of aberrant colonic crypts andadenocarcinomas by inhibiting growth or inducing differentiationand/or apoptosis. When other mitogenic sphingolipid rnetabolites(such as lyso-SM) are produced, they may diminish these suppressiveeffects. Unfortunately, it is not possible to test these hypotheses byfeeding the SM metabolites (lyso-SM, ceramide, or sphingosine)because these compounds are more rapidly taken up by the uppergastrointestinal tract and do not seem to reach the colon (l6).6

Dudeja et aL (52) have noted that one of the first changes in rat coloniccells induced by DMH is a decrease in membrane fluidity that appears tobe due to increases in membrane SM. The elevation in SM was associatedwith an increase in SM synthesis and a reduction in the activity of the(neutral) sphingomyelinase. Thus, it is possible that some forms ofneoplasia involve defect(s) in the sphingolipid signaling pathway(s) thatregulate growth arrest and apoptosis; if so, an exogenous source ofsphingolipid (i.e., via the diet) might be able to bypass the defect. Wrightet a!. (53) have recently described a cell line that is resistant to tumornecrosis factor a- and UV irradiation-induced cell death due to a defectin the activation of sphingomyelinase and the 24-Wa apoptotic protease,

and the addition of exogenous ceramide bypassed the defect to induceapoptosis in the cells. Therefore, the effects of dietary SM in our studiesmay involve a similar bypass of aberrant sphingolipid signaling. If this isthe case, this mechanism could be highly relevant both to understandingthe relationships between diet and cancer and to the development of newapproaches to cancer chemotherapy.

LysoSM?

1'Sphingomyein

+Ceramide •

(N-Acyl-Sphingosine)

adenomas rather than the more malignant adenocarcinomas (Fig. 5).Colorectal tumors are thought to begin as adenomas and progress toadenocarcinomas through further genetic mutations (such as mutations in DCC and p.53 tumor suppressor genes; Ref. 43); therefore, SMfeeding might prevent the progression of adenomas to adenocarcinomas, although we cannot exclude the possibility that it is capable ofpromoting reversion of adenocarcinomas to adenomas. Since the shiftin tumor type was statistically significant (P < 0.05) in the 0.05% SMgroup but not for the mice fed 0.1% SM, it is possible that high levelsof SM are less effective; however, a more extensive dose-responsestudy should be conducted before drawing this conclusion.5

Although there have been few analyses of the levels of sphingolipids in food, milk contains almost 0.1 mg of SM per ml (0.5 to 1mg/g of dry milk) (44), pork and beef contain approximately 0.25rng/g, and chicken has nearly 0.4 rng/g (45). Thus, these foods contain0.025 to 0. 1% SM, which compares well with the level (0.05% SM)that suppressed both the aberrant crypt foci and the development ofadenocarcinomas. It warrants mention that the basal AIN 76A diet isprobably “unnatural―with respect to its low sphingolipid content (i.e.,<0.005%), which is due to the components (casein, corn oil, sucrose,starch, and others) being essentially sphingolipid free (17).

Why did the first experiment indicate that SM decreased the tumornumber but the second showed a shift in tumor type? As alreadyemphasized, these results are probably not due to differences in theSM preparations; however, the two experiments differed in anotherway. In the first study, we were unable to isolate enough SM for theentire time course; therefore, SM was added to the diet for only thefirst 21 weeks. It is possible that some SM metabolites (such asceramide) inhibit the development of adenocarcinomas, but others(such as lyso-SM, which can be produced by bacterial enzymes and ismitogenic; Refs. 46 and 47) stimulate the growth of adenomas. Insupport of this possibility, SM feeding increased proliferation ofcolonic cells in the initial study (17) but only for DMH-treated mice.

Fig. 6 summarizes current information about sphingolipid metabolism and cell regulation that may be relevant to the findings of thisstudy. Some SM escapes digestion in the upper intestinal tract and is

5 These dose-response relationships might have been more clear if we had used a

protocol that increased the number of animals that developed tumors (for example, byusing higher amounts of DMH); however, because so little is known about the effects ofdietary sphingolipids, we were attempting to induce tumors in about one-half of the micein the control group so that increases as well as decreases in tumor incidence could bedetected.

+Sphingosine —

Fig. 6. Sphingolipid digestion in the lumen of the intestinal tract, uptake of metabolites,cellular metabolism, and possible effects on colonic cells. As described in the text, eachof the events illustrated in this diagram have been noted by various studies; however, theirintegration to explain the observations in this manuscript is still hypothetical.

6 E. M. Schmelz and A. H. Merrill, Jr., unpublished observations.

4940

Biological activities:

{Growthstimulation.Sphingomyelin(&Glycolipids)

IDefectiveinneoplasia?. I Inhibition of growth,

Ceranude@ induction ofL differentiation,apoptosis

4+@ Inhibition/stimulationofSphingosine@ growth, cytotoxic,

t inductionofapoptosis

REFERENCES

1. Vargas, P. A., and Alberts, D. S. Primary prevention of colorectal cancer. Cancer(Phila.),70:1229—1235,1992.

2. Herbsman, N., and Homsby-Lewis, L. The effects of diet on colon cancer. Gastroenterology, 105: 604—605, 1993.

3. Giovannucci, E., and Willeti, W. C. Dietary factors and risk of colon cancer. Ann.Med., 26: 443—452,1993.

4. Rogers, A., Zeisel, S. H., and Groopman, J. Diet and carcinogenesis. Carcinogenesis(Lond.),/4:2205—2217,1993.

5. Nixon, D. W. Diet and chemoprevention of colon polyps and colorectal cancer.Semin. Surg. Oncol., 11: 41 1—415,1995.

6. Reddy, B. S. Nutritional factors and risk of colon cancer. Crit. Rev. Food Sci. Nutr.,35: 175—190,1995.

7. Hannun, Y. A., and Linardic, C. M. Sphingolipid breakdown products: anti-proliferative and tumor-suppressor lipids. Biochim. Biophys. Acta, 1154: 223—236,1993.

8. Spiegel, S., and Merrill, A. H., Jr. Sphingolipid metabolism and growth regulation.FASEB J., in press, 1996.

9. Zhang, H., Desai, N. N., Olivera, A., Seki, T., Brooker, 0., and Spiegel, S. Sphingosine-l-phosphate. a novel lipid, involved in cellular proliferation. J. Cell Biol.. 114:155—167,1991.

10. Stevens, V. L., Nimkar, S., Jamison. W. C., Liotta, D. C., and Merrill, A. H., Jr.Characteristics of the growth inhibition and cytotoxicity of long-chain (sphingoid)bases for Chinese hamster ovary cells: evidence for an involvement of protein kinaseC. Biochim. Biophys. Acta, 1051: 37—45,1990.

INHIBITION OF COLON CARCINOGENESIS BY SPHJNOOMYELIN

11. Jayadev, S., Liu, B., Bielawska, A. E., Lee, J. Y., Nazaire, F., Pushkareva, M. Y.,Obeid, L. M., and Hannun, Y. A. Role for ceramide in cell cycle arrest. J. Biol.Chem., 270: 2047—2052,1995.

12. Stevens, V. L., Owens, N. E., Winton, E. F., Kinkade, J. M., Jr., and Merrill, A. H.,Jr. Modulation of retinoic acid-induced differentiation of human leukemia (HL-60)cellsby serumfactorsandsphinganine.CancerRes.,50: 222—226,1990.

13. Merrill. A. H., Jr., Sereni, A. M., Stevens, V. L., Hannun, Y. A., Bell, R. M., andKinkade, J. M., Jr. Inhibition of phorbol ester-dependent differentiation of humanpromyelocytic leukemic (HL-60) cells by sphinganine and other long-chain bases.J.Biol.Chem.,261:12610—12615,1986.

14. Hannun, Y. A., and Obeid, L. M. Ceramide: an intracellular signal for apoptosis.Trends Biochem. Sci., 20: 73—77,1995.

15. Merrill, A. H., Jr., and Sweeley, C. C. Sphingolipid metabolism and cell signalling.In: D. E. Vance and J. Vance (eds.), Biochemistry of Lipids, Lipoproteins, andMembranes, Ed. 2, Vol. 31, pp. 309-338. Amsterdam, the Netherlands: ElsevierScience Publishers, 1996.

16. Schmelz, E. M., Crall, K. J., Larocque, R., Dillehay, D. L., and Merrill, A. H., Jr.Uptake and metabolism of sphingolipids in isolated intestinal loops. J. Nutr., 124:702-712, 1994.

17. Dillehay, D. L., Webb, S. K., Schmelz, E. M., and Merrill, A. H., Jr. Dietarysphingomyelin inhibits l,2-dimethylhydrazine-induced colon cancer in CF1 mice. J.Nutr., 124: 615—620, 1994.

18. National Research Council. Guide for the Care and Use of Laboratory Animals.Publication No. 88—23(rev.). Bethesda, MD: National Institutes of Health, 1985.

19. Ann, Q., and Adams, J. Collision-induced decompositions of sphingomyelins forstructural elucidation. Biol. Mass Spectrom., 2: 285—294,1993.

20. Strobel, F. H., and Adams, J. How to measure the kinetic energy distribution of theprecursor ion main beam in MIKES experiments. J. Am. Soc. Mass Spectrom.. 6:1232—1242,1995.

21. Rogers. A., and Nauss, K. M. Rodent models for carcinoma of the colon. Dig. Dis.Sci.,30:875-1025,1985.

22. Weisburger, J. H., and Fiala, E. S. Experimental colon carcinogenesis and their modeof action. In: H. Autrop and 0. M. Williams (eds.). Experimental Colon Carcino

genesis, pp. 27—50.Boca Raton, FL: CRC Press, 1982.23. Merrill. A. H., Jr., Wang, E., Mullins, R. E., Jamison, W. C. L., Nimkar, S., and

Liotta, D. C. Quantitation of free sphingosine in liver by high-performance liquidchromatography. Analyt. Biochem., 171: 373-38 1, 1988.

24. Bird, R. P. Observations and quantification of aberrant ciypts in the murine colon treatedwith a colon carcinogen: preliminary findings. Cancer Leu., 37: 147—151,1987.

25. Elsayed,A. M.,andShamsuddin,A. M.Neoplasmsof thecolon.In: S. F. Stinson,H. M. Schuller, and 0. Reynik (eds.), Atlas of Tumor Pathology of the Fisher Rat,Vol. 1, pp. 133—192.Boca Rota, FL: CRC Press, 1990.

26. Morrison. W. R. Polar lipids in bovine milk. I. Long-chain bases in sphingomyelin.Biochim. Biophys. Acts, 176: 537—546,1969.

27. Hannun, Y. A., Loomis, C. R., Merrill, A. H., Jr., and Bell, R. M. Sphingosineinhibition of protein kinase C activity and of phorbol dibutyrate binding in vitro andin human platelets. J. Biol. Chem., 261: 12604—12609, 1986.

28. Borek, C., Ong, A., Stevens. V. L., Wang, E., and Merrill, A. H., Jr. Long-chain(sphingoid) bases inhibit multistage carcinogenesis in mouse C3H/lOT½ cells treatedwith radiation and phorbol 12-myristate 13-acetate. Proc. NatI. Acad. Sci. USA, 88:1953—1957,1991.

29. Borek, C., and Merrill, A. H., Jr. Sphingolipids inhibit multistage carcinogenesis andprotein kinase C. In: 0. Bronzetti, M. Hayatsu, S. De Flora, M. D. Waters, D. M.Shankel(eds.),Antimutagenesisand AnticarcinogenesisMechanismsIII, pp. 367—371. New York: Plenum Publishing Corp., 1993.

30. Merrill, A. H., Jr., Liotta, D. C., and Riley, R. E. Bioactive properties of sphingosineand strucwrally related compounds. In: R. M. Bell, J. H. Exton, and S. M. Prescott,(eds.), Handbook of Lipid Research, Vol. 8, pp. 205—237. New York: PlenumPublishing Corp., 1996.

31. Gupta, A. K., Fischer, 0. J., Elder, J. T., Nickoloff, B. J., and Voorhees, J. J.Sphingosine inhibits phorbol ester-induced inflammation, ornithine decarboxylaseactivity, and activation of protein kinase C in mouse skin. J. Invest. Dermatol., 91:486—491,1988.

32. Enkvetchakul, B., Mcmli, A. H., Jr., and Birt, D. F. Inhibition of ornithine decarboxylase activity in mouse epidermis by sphingosine sulfate. Carcinogenesis (Lond.),10:379—381,1989.

33. Enkvetchakul, B., Barnett, 1., Liotta, D. C., Geisler, V., Menaldino, D., Merrill,A. H., Jr., and Birt, D. F. Influences of sphingosine on two-stage skin tumorigenesisin SENCAR mice. Cancer Len., 62: 35—42,1992.

34. Wertz, P. W., and Downing, D. T. Free sphingosines in porcine epidermis. Biochim.Biophys. Acts, 1002: 213—217,1989.

35. Endo, K., Igarashi, Y., Nisar, M., Thou, Q. H., and Hakomon, S-I. Cell membranesignaling as target in cancer therapy: inhibitory effect of N,N-dimethyl and N,N.Ntrimethyl sphingosine derivatives on in vitro and in vis'o growth of human tumor cellsinnudemice.CancerRes..51: 1613—1618,1991.

36. Okoshi, H., Hakomori, S-I., Nisar, M., Thou, Q. H., Kimura, S., Tashiro, K., andlgarashi, Y. Cell membrane signaling as target in cancer therapy. II. Inhibitory effectof N,N.N-trimethylsphingosine on metastatic potential of murine B 16 melanoma cellline through blocking of tumor cell-dependent platelet aggregation. Cancer Res., 51:6019—6024, 1991.

37. Park, Y. S., Hakomori, S-I., Kawa, S., Ruan, F., and Igarashi, Y. Liposomal N.N,Ntrimethylsphingosine (TMS) as an inhibitor of Bl6 melanoma cell growth andmetastasis with reduced toxicity and enhanced drug efficacy compared to free TMS:cell membrane signaling as a target for cancer therapy III. Cancer Res., 54: 2213—2217,1994.

38. McLellan, E. A., Medline, A., and Bin, R. P. Sequential analyses of the growth andmorphological characteristics of aberrant crypt foci: putative neoplastic lesions.Cancer Res., 51: 5270—5274,1991.

39. Pretlow, T. P., Barrow, B. J., Ashton, W. S., O'Riordan, M. A., Pretlow, 1. 0.,Jurcisek, J. A., and Stellato, 1. A. Aberrant crypts: putative neoplastic foci in humancolonic mucosa. Cancer Res., 51: 1564—1567, 1991.

40. Bruce, W. R., Archer, M. C., Corpet, D. E., Medline, A., Minkin, S., Stamp, D., Yin,Y., and Thang, X. Diet, aberrant crypt foci and colorectal cancer. Mutat. Res., 290:111-118,1993.

41 . Kristiansen, E., Thorup, I., and Meyer, 0. Influence of different diets on developmentof DMH-induced aberrant crypt foci and colon tumor incidence in Wistar rats. Nutr.Cancer, 23: 151—159,1995.

42. Magnuson. B. A., Carr, I., and Bird, R. P. Ability of aberrant crypt foci characteristicsto predictcolonic tumor incidencein rats fed cholic acid. CancerRes.,53: 4499—4504, 1993.

43. Vogelstein. B., and Kinzler, K. W. The multistep nature of cancer. Trends Genet.. 9:138—141,1993.

44. Zeisel, S. H., Char, D., and Shear, N. F. Choline, phosphatidylcholine and sphingomyelin in human and bovine milk and infant formulas. J. Nuts., 116: 50—58,1986.

45. Blank, M. L., Cress, E. A., Smith, Z. L., and Snyder, F. Meats and fish consumed inthe American diet contain substantial amounts of ether-linked phospholipids. J. Nutr.,122: 1656—1661,1992.

46. Ito, M., Kurita, 1., and Rita, K. A novel enzyme that cleaves the N-acyl linkage ofceramides in various glycosphingolipids as well as sphingomyelin to produce theirlyso forms. J. Biol. Chem., 270: 24370—24374, 1995.

47. Desai, N. N., Carlson, R. 0., Mattie, M. E., Olivera, A., Buckley, N. E., Seki, T.,Brooker, 0., and Spiegel, S. Signaling pathways for sphingosyl-phosphorylcholinemediated mitogenesis in Swiss 3T3 fibroblasts. J. Cell Biol., 121: 1385—1395,1993.

48. Pushkareva, M., Chao, R., Bielawska, A., Merrill, A. H., Jr.. Crane, H. M., Lagu, B.,Liotta, D. C., and Hannun, Y. A. Stereoselectivity of induction of the retinoblastomagene product (pRb) dephosphorylation by o-erythro-sphingosine supports a role forpRb in growth suppression by sphingosine. Biochemistry. 34: 1885—1892,1995.

49. Wolff, R. A., Dobrowsky, R. T., Bielawska, A., Obeid, L. A., and Hannun, Y. A. Roleof ceramide-activatedproteinphosphatasein ceramide-mediatedsignaltransduction.J. Biol. Chem.. 269: 19605—19609,1994.

50. Kolesnick, R N., and Golde, D. W. The sphingomyelin pathway in tumor necrosisfactor and interleukin-l signalling. Cell, 77: 325—328,1994.

51 . Mattie, M., Brooker, 0., and Spiegel, S. Sphingosine-l-phosphate, a putative secondmessenger, mobilizes calcium from intemal stores via an inositol trisphosphateindependent pathway. J. Biol. Chem., 269: 3 18 1—3188, 1994.

52. Dudeja, P. K.. Dahiya, R., and Brasitus, T. A. The role of sphingomyelin synthase andsphingomyelinase in 1,2-dimethylhydrazine-induced lipid alterations of rat colonicplasma membranes. Biochim. Biophys. Acta, 863: 309—312, 1986.

53. Wright, S. C., Theng, H., and Zhong, J. Tumor cell resistance to apoptosis due to adefect in the activation of sphingomyelinase and the 24 kDa apoptotic protease(AP24). FASEB J.. 10: 325—332,1996.

4941