Enhanced colonic tumorigenesis in alkaline sphingomyelinase (NPP7) knockout mice

Cancer Biology and Signal Transduction

Enhanced Colonic Tumorigenesis in AlkalineSphingomyelinase (NPP7) Knockout MiceYing Chen1,2, Ping Zhang1, Shu-Chang Xu2, Liping Yang3, Ulrikke Voss4, Eva Ekblad4,Yunjin Wu5, Yalan Min3, Erik Hertervig1,6, Åke Nilsson1,6, and Rui-Dong Duan1

Abstract

Intestinal alkaline sphingomyelinase (alk-SMase) generatesceramide and inactivates platelet-activating factor (PAF) and waspreviously suggested to have anticancer properties. The directevidence is still lacking. We studied colonic tumorigenesis inalk-SMase knockout (KO) mice. Formation of aberrant crypt foci(ACF) was examined after azoxymethane (AOM) injection.Tumor was induced by AOM alone, a conventional AOM/dextransulfate sodium (DSS) treatment, and an enhanced AOM/DSSmethod. b-Catenin was determined by immunohistochemistry,PAF levels by ELISA, and sphingomyelin metabolites by massspectrometry. Without treatment, spontaneous tumorigenesiswas not identified but the intestinal mucosa appeared thicker inKO than in wild-type (WT) littermates. AOM alone inducedmoreACF in KOmice but no tumors 28weeks after injection. However,combination of AOM/DSS treatments induced colonic tumors

and the incidence was significantly higher in KO than inWTmice.By the enhanced AOM/DSS method, tumor number per mouseincreased 4.5 times and tumor size 1.8 times inKOcomparedwithWT mice. Although all tumors were adenomas in WT mice, 32%were adenocarcinomas in KO mice. Compared with WT mice,cytosol expression of b-catenin was significantly increased andnuclear translocation in tumors was more pronounced in KOmice. Lipid analysis showed decreased ceramide in small intestineand increased sphingosine-1-phosphate (S1P) in both smallintestine and colon in nontreated KO mice. PAF levels in feceswere significantly higher in the KO mice after AOM/DSS treat-ment. In conclusion, lack of alk-SMase markedly increasesAOM/DSS–induced colonic tumorigenesis associated withdecreased ceramide and increased S1P and PAF levels. Mol CancerTher; 14(1); 259–67. �2014 AACR.

IntroductionMetabolism of sphingolipids has important implications in

many diseases, including inflammatory bowel disease and coloncancer (1). Sphingomyelin is a major type of sphingolipids exist-ing in all eukaryotic cell membranes and dietary products such asmilk, egg, andmeat (2, 3). Hydrolysis of sphingomyelin generatesceramide that can inhibit cell proliferation and induce differen-tiation and apoptosis via numerous signaling pathways (4).Ceramide can be degraded to sphingosine that can in turn formsphingosine-1-phosphate (S1P) by sphingosine kinase (SphK) orbe converted to ceramide-1-phosphate (C1P) by ceramide kinase.

In contrast with ceramide, S1P and C1P have stimulatory effectson cell proliferation, cell migration, and angiogenesis (5, 6). Thebalance of ceramide and S1P is a key factor determining cell fate,thus having important implications in tumorigenesis (7).

It has been shown that feeding animals with sphingomyelin orceramide analogues attenuated colonic tumorigenesis (8, 9). Thekey enzyme in the gut that hydrolyses sphingomyelin to ceramideis alkaline sphingomyelinase (alk-SMase; ref. 10), which wasdiscovered, purified, and cloned in our group (11, 12). Cloningstudies reveal that alk-SMase shares no structural similarities withother SMases, but belongs to the nucleotide pyrophosphatasephosphodiesterase (NPP) family and is now also called NPP7(11, 13, 14). However, NPP7 is inactive toward nucleotide phos-phate esters but has specific activities against phospholipidscontaining a phosphocholine moiety, including sphingomyelin,platelet-activating factor (PAF), and lysophosphatidylcholine(13, 15). Hydrolyses of these substrates result in the increasedceramide formation, inhibited proinflammatory effects of PAF,and reduced formation of lysophosphatidic acid. All have impli-cations in carcinogenesis (16).

In patients, alk-SMase activity is decreased by 25% in chroniccolitis and by 75% in colon cancer (17, 18). Low activity has alsobeen found in the feces from patients with colon cancer (19) andin bile of patients with hepatobiliary cancers (20). An alternativesplicing form of alk-SMase has been detected in both colonic andhepatic cancer cells (21, 22). The aberrant form has impairedsubstrate-binding site rendering the enzyme inactive. All thesefindings lead to a hypothesis that alk-SMase is an anticancer factorin the gut. Direct evidence is, however, still lacking. We hereexamined the tumorigenesis in alk-SMase knockout (KO) micegenerated inour groupby theCre-LoxP system (10) and show that

1Gastroenterology and Nutrition Laboratory, Department of ClinicalSciences Lund, University of Lund, Lund, Sweden. 2Gastroenterology,Tongji Hospital,Tongji University School of Medicine, Shanghai, China.3Cancer Research Center,Tumor Hospital of Nantong University, Nan-tong, China. 4Neurogastroenterology, Department of ExperimentalMedical Science, University of Lund, Lund, Sweden. 5Department ofPathology, Tongji Hospital, Tongji University School of Medicine,Shanghai, China. 6Gastroenterology, Ska

�ne University Hospital, Lund,

Sweden.

Current address for P. Zhang is Daqing Campus, Harbin Medical University,Daqing, China.

Corresponding Authors: Rui-Dong Duan, Gastroenterology and Nutrition Lab-oratory, BMC, B11, Department of Clinical Sciences, Lund University, Lund, Skane22184, Sweden. Phone: 46-46-222-0709; E-mail: [email protected];and Shu-Chang Xu, Gastroenterology, Tongji Hospital, Tongji University Schoolof Medicine, 200065 Shanghai, China. Phone: 86-21-66111287; Fax: 86-21-56050191; Email: [email protected]

doi: 10.1158/1535-7163.MCT-14-0468-T

�2014 American Association for Cancer Research.

MolecularCancerTherapeutics

www.aacrjournals.org 259

on March 14, 2016. © 2015 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst November 7, 2014; DOI: 10.1158/1535-7163.MCT-14-0468-T

http://mct.aacrjournals.org/

lack of alk-SMase significantly enhanced colon cancer suscepti-bility induced by carcinogens and inflammation.

Materials and MethodsMaterials

Azoxymethane (AOM), anti–b-catenin antibody, and secondantibodies used were purchased from Sigma-Aldrich, anti-mNPP7 antibody from R&D Systems, and dextran sulfate sodium(DSS) with M.W. 36,000–50,000 from MP Biomedicals. TheGenomic DNA Extraction Kit was from Fermentas and the PAFAssayKit fromAntibodies-Online, AachenGermany. Standards ofsphingomyelin (d18:1/16:0 and d18:0/16:0), ceramide (d18:1/17:0, d18:1/16:0 andd18:0/16:0), andS1P (18:1)were purchasedfrom Avanti Polar Lipids.

AnimalsAlk-SMase KO mice were generated by the Cre-LoxP system as

described previously (10). In brief, two loxP sites are inserted toflank the exon 2 of ENPP7 located on chromosome 11 (EnsemblGene ID:ENSMUSG00000046697).Deletion of the floxed regionuponCre recombinase reaction causes a frame-shift, which createsa novel stop codon that stops the translation at early stage. TheKOmiceweremaintained onC57BL/6 background and themicewerebackcrossed eight times to get an inbred strain. Unless specifiedelsewhere age and sex matched C57BL/6 WT (wild-type) micewere purchased fromMollegaard; same source for backcross of themice as described previously (10). The animals were fed standardpellets with free access to water. The ethics permit for generatingand usingWT and KOmice in the studywas issued by theMalm€o/Lund Ethics Committee.

Histology and histopathology examinations of the intestinaltract

The tissue samples of colon and small intestine were fixed in4% paraformaldehyde in 0.1 mol/L PBS pH 7.2 overnight andthen stored in 70% ethanol. The samples were then dehydratedand embedded in paraffin. Three sections from each samplewere cut, dewaxed, and stained in hematoxylin and eosin. Themucosal thickness of small intestine and colon was measuredfrom the longitudinal sections by Image Scope software, whichwas performed in four littermates (2 WT and 2 KO). Only theintact villi or crypt was taken into account. In tumorigenesisstudies, the tumors identified were excised together with thesurrounding tissues and handled similarly as above for histo-pathology examination.

Study protocols in tumorigenesisThe studieswere started inmice at the age of 8 to 10weeks. Four

study protocols were used to examine the tumorigenesis.Study A examines aberrant crypt foci (ACF) formation accord-

ing to Bird and colleagues (23). In brief, 3WT and 3 KOmiceweres.c. injected with AOM (12mg/kg), once a week for 2 weeks. Fourweeks after the last injection, the colon was stained with meth-ylene blue and the number and the size of ACF were countedunder a dissecting microscope by a person unaware of the samplesource.

Study B examines tumorigenesis induced by AOM alone. TenWT and 10 KO mice were injected with AOM as in Study A. Thetumorigenesis was examined 28 weeks after the last injection.

Because in Study B, tumorigenesis was not identified, a con-ventional combination of AOM and DSS treatments was used

(Study C; refs. 24, 25). TenWT and 10 KOmice were injectedwithAOM(10mg/kg) once.Oneweek later, the animalswere given1%DSS in the drinking water for 1 week followed by 2 weeks of tapwater. The DSS treatment was repeated for additional 2 cycles.Two weeks after the last cycle, the tumorigenesis was examined.

Study D examines tumorigenesis with a long-term enhancedmethod that combined the treatments in Study B and C. Ten WTand 10 KO mice were injected with AOM twice as in Study B.Twenty-one weeks later, the mice received an additional AOMinjection followed by three cycles of DSS treatment as in Study C.The tumorigenesis was examined 2 weeks after the last DSS cycle.In Study B, C, andD, the body weights of themice weremeasuredat least once a week.

Figure 1.Alk-SMase expression and morphologic changes of the colon in the WT andKO mice. A, cellular location of alk-SMase on the apical surface of the villi inWT mice, and disappearance in the KO mice (B). C and D, representativemicrographs showing the increased colonic thickness in KO mice comparedwithWTmice. F, aberrant crypt with hyperchromatism and nuclear crowding(pointed by arrows) in KO mice compared with the WT mice (E), after AOMtreatment in Study B; bar, 20 mm.

Chen et al.

Mol Cancer Ther; 14(1) January 2015 Molecular Cancer Therapeutics260




Upon examination, the animals were anesthetized withxylazine/ketamine. The small intestine and colon wereremoved and cut opened. The number of tumors appearingwas counted by the naked eyes and rechecked under a dissect-ing microscope (�40). The length and the width of each tumorwere measured by a digital micro ruler. The tumor area wascalculated by multiplying length and width. The tumors andthe surrounding tissues were then excised, fixed, and stained forpathologic examination. The signs of dysplasia and tumori-genesis were evaluated regarding cellular morphology, nucle-oplasm ratio, and mitoses.

Immunohistochemistry studies for expression of b-cateninIHC analysis of b-catenin was performed in both nontreated

WT (n ¼ 3) and KO (n¼ 3) mice as well as in AOM/DSS-treatedWT (n ¼ 3) and KO (n ¼ 5) mice in Study D. Longitudinalsections of colon were used. Paraffin sections (5 mm) weredeparaffinized, hydrated, and subjected to antigen retrieval byboiling in citrate buffer (0.01 mol/L pH 6) for 2 � 8 minutes.

Sections were cooled and washed in PBS containing 0.25%Triton X100 (PBST) before being incubated overnight at 4�Cwith a polyclonal b-catenin antibody diluted (1:8,000) in PBSTcontaining 0.25% BSA. b-Catenin was visualized either withfluorescein (Z0196 Dako) or Alexa Flour 594 conjugated IgGantibodies (Jackson ImmunoResearch). Secondary antibodieswere mixed and incubated 1 hour at room temperature.Hoechst (Life Technologies) cell nuclei counter staining wasperformed according to the manufacturer's protocol. Mountingwas in PBS:glycerol 1:1 followed by fluorescence microscopy(Olympus BX43, LRI) with appropriate filter setting. IHC of alk-SMase in the intestinal mucosa of WT and KO mice wasperformed as above using polyclonal anti-NPP7 as first anti-body and anti-goat IgG as second antibody.

Assay of sphingolipids in the intestinal mucosa by MSSphingomyelinmetaboliteswere determined in6WTand6KO

nontreatedmice. After sacrificing, the intestinal tract was removedand intestinal content and feces were washed out with ice-cold

0 2 4 6 8 1060

80

100

120

140WTKO

A

Time (wk)

Ch

ang

e o

f bo

dy

wei

gh

t (%

)

WT KO0

10

20

30

40

50

60 *B

Co

lon

len

gth

(mm

)

WT KO

0

5

10

15 P = 0.0142

*

C

Tu

mo

r n

um

be

r/m

ou

se

WT KO0.0

2.5

5.0

7.5

10.0

12.5

D

Tu

mo

r si

ze (

mm

2 )

1 2 5 8 10 wk

Figure 2.Tumorigenesis induced by AOM plusDSS treatment (StudyC). As illustratedat the top, the animals were treatedwith a single AOM injection (arrow),followed by three cycles of DSS indrinking water (dark boxes). The timein weeks was indicated (detailed inMaterials and Methods section). Aftertreatment, the gain of body weight isshown in A, colon length in B, tumornumber per mouse in C, and size oftumor area in D; �, P < 0.05, comparedwith WT mice; n ¼ 10 for WT and 7 forKO mice.

Lack of NPP7 Enhances Colonic Tumorigenesis

www.aacrjournals.org Mol Cancer Ther; 14(1) January 2015 261




PBS buffer. The tissues were frozen immediately in liquid nitro-gen. The phospholipids and sphingolipids were extracted asdescribed previously (26). In brief, the frozen tissueswere homog-enized and 50 mL of homogenate was mixed with 950 mL meth-anol containing internal standards. After lipid extraction, 800 mLof supernatant was directly used for S1P analysis in negative ionMRM mode, and sphingomyelin and ceramide, in positive ionMRM mode. The analyses were performed using the API 4500QTRAP mass spectrometer (Applied Biosystems/MDS SCIEX).Both nebulizer and desolvation gases were nitrogen. Sampleswere loaded through an LC system with an auto sampler. Typicaloperating parameters were as follows: curtain gas 25, collision gasmedium, ion source gas 1(GS1) 45, and ion source gas 2 (GS2) 50.Electrospray voltage with positive ion MRMmode was 5500 andthat with negative ion mode, �4500. The temperature of theheater was 500�C.

Measurement of PAFAfter the AOM and DSS treatments in Study C, animals were

sacrificed and tumorigenesis examined. Fecal samples were col-lected from the colon, dried under nitrogen gas, andweighed. The

dried feces were suspended in saline, centrifuged at 10,000 rcf at4�C for 10 minutes and the supernatant was collected. The wholecolonic mucosa was scraped and lysed as described previously(10), followed by centrifugation at 10,000 rcf at 4�C for 10minutes. The levels of PAF in feces and mucosal lysates weredetermined by an ELISA kit according to the manufacturer'sprotocol and the values were adjusted with the fecal weight orprotein levels of the samples.

Statistical analysisThe results are expressed asmean� SE. Statistical analyses were

performed by the nonpaired Student t test and a P value of <0.05was considered statistical significant.

ResultsObservation of intestinal morphology and spontaneoustumorigenesis in the KO mice

The main genotypic and phenotypic changes in the KO micehave been reported previously (10). Alk-SMase activity was notdetectable in intestinal mucosa, content, and feces in these KO

0 10 20 3090

100

110

120

130

140WT

KO

WT KO0

50

100

150

*P = 0.0070

% B

od

y w

eig

ht

A

Time (wk)

Ch

ang

e o

f b

od

y w

eig

ht

(%)

WT

KO

0

50

100

150

Inci

den

ce o

f tu

mo

r (%

)

B

WT

KO0

2

4

6

8

***

Tu

mo

r n

um

ber

/mo

use

C

WT

KO0

5

10

15

20

25*

Siz

e o

f ea

ch t

um

or

(mm

2 )

D

1 2 21 22 25 28 30 wk

Figure 3.Tumorigenesis induced by long-termenhanced AOM and DSS treatment(Study D). As illustrated at the top, themice were treated with three AOMinjections as indicated by arrows andfollowed by three cycles of DSS indrinking water (dark boxes; detailed inMaterials and Methods). Dynamicchanges in body weight are shown inA and final body weight in the inset.Tumor incidence is shown in B, numberof tumors per mouse in C, and size ofeach tumor in D; � , P < 0.05 and�� , P <0.005 comparedwithWTmice;n ¼ 10 for both WT and the KO mice.

Chen et al.





mice. The animals look normal and no significant changes werefound for body weight, food intake, intestinal movement, smallintestine and colon length, and fertility. In the present study, wefurther assessed alk-SMase expression in WT and the KO mice byIHC. As shown in the top in Fig. 1, intense alk-SMase immuno-reactivity is located at the apical surface of the columnar epithe-lium covering the villi in WT mice. No alk-SMase labeling isdetected in the KO mice. Intestinal histology was compared in 4littermates (2 WT and 2 KO) and the mucosal architecture of theKOmice was largely normal. However, the colonic mucosa in KOmice increased by 22% (260.5 � 3.6 vs. 214.0 � 4.6 mm, P <0.0001, n ¼ 62 for KO and 56 for WT), and the small intestinalmucosa increased by 9% (463.6 � 3.9 vs. 424.5 � 7.0 mm, P <0.0001, n ¼ 36 for both groups) compared with WT mice.Representative micrographs of colonic thickness are shown inthe middle (Fig. 1C and D). Spontaneous tumorigenesis wasexamined in 4, 5, and 3 KO mice at the age of 10, 12, and 17months, respectively. No tumors were identified either in thesmall intestine or colon.

Study A and B: effects of AOM alone on ACF formation andcolonic tumorigenesis

Four weeks after injection of AOM the ACF number permouse in 3 WT mice were 15.3 � 2.23 and that in KO mice,24.0 � 1.5 (P < 0.05). However, when examined 28 weeks afterAOM injection in Study B, no tumors were identified either inthe colon or in the small intestine in both WT and KO mice.Pathologic studies only showed more aberrant crypt withhyperchromatism and nuclear crowding in the KO than in theWT mice (Fig. 1E and F).

Study C and D: increased tumorigenesis by combination ofAOM and DSS treatment

Because of the failure to induce colonic tumors by AOM alonein Study B, animals were treated with a combination of AOM andDSS by a short-term conventional method (Study C) and a long-term enhanced method (Study D).

In Study C, the gain in body weight was halted in KO but not intheWTmice (Fig. 2A). The length of the colon upon examinationwas 13% shorter (P < 0.05) in KO than in WT mice (Fig. 2B). Thetumors were clearly identified in the distal part of the colon. Mostof them were round with smooth surface. The number of tumorspermouse inKOmicewas 3.5 times that inWTmice (Fig. 2C). Themean size of tumors in termof tumor surfacewas slightly higher inKO mice without statistical significance (Fig. 2D). During DSStreatment, 3 KO mice were dying due to severe diarrhea anddehydration. These mice were terminated and excluded from thestudy due to the ethical considerations. All WT mice survived.

In Study Dwith a long term enhanced induction of tumors, thegain in bodyweightwas lower in theKOgroup comparedwith the

Figure 4.Representative macrographs showing the tumorigenesis in Study D. A, colonfrom a WT mouse with tumors. B, colon from a KO mouse with tumors. Theunit of the rulers in A and B are in cm. C to E, photos under a dissectingmicroscope (�40). C, tumors in the WT group. D, tumors in theKO group. E, the largest tumor in the KO group that almost inducedobstruction of the colon.

Figure 5.Representative micrographs showing pathologic changes in Study D. A1 andA2, adenomaswith high-grade dysplasia fromWT (A1) and KO (A2)mice. A3,an adenoma with cancerous transformation in KO mice. B1, the largestadenoma with high-grade dysplasia found in the WT mice group. B2, arepresentative larger adenomawith high-grade dysplasia in a KOmouse, andB3 is a large adenocarcinoma protruding into the lumen in KO mice. Similarmoderate inflammations are also visible in A1 to B3. C1 and C2, deepinflammatory lesions in WT and KO mice; and C3 shows severe ulcerativecolitis in the KO group. Bars, 50 mm (A and C) and 500 mm (B).






WT group (Fig. 3A). Reduced body weight was already observedbefore DSS treatment and became more pronounced thereafter.Ten WT and 10 KO mice were involved in the study. Uponsacrifice, tumors were found in all 10 KO mice but only in 5 ofthe WT mice (Fig. 3B). In addition, KO mice have more tumorsand the tumor number per mouse was 4.5 times higher in the KOmice compared with the 5WTmice (Fig. 3C). The average surfacesize of tumors was 8.3 mm2 in the KO group and 4.6 mm2 in theWT group (Fig. 3D). As can also be seen in Fig. 3D, about 75% ofthe tumors in the KOgroupwere larger than the average size of thetumors inWTmice. Figure 4 shows representativemacrographs ofthe differences in tumor number and size between WT and KO

mice. In general, only 1 to 3 small tumors were found in the WTmice, but both number and the size of the tumors in KOmice aresignificantly greater than in WT mice (Fig. 4A vs. B and Fig. 4C vs.D). In 2 KO mice, the tumors were remarkably large that almostcause obstruction of the colon (Fig. 4E).

Histopathology examinations found that the tumors in WTmice were all adenomas, of which 50% exhibited low- and 50%high-grade dysplasia. Of the tumors in KO mice, 32% wereadenomas with low-grade dysplasia, 36% were adenomas withhigh-grade dysplasia, and 32% were adenocarcinomas. Figure 5shows representative high-grade dysplasia in WT (Fig. 5A1) andKO mice (Fig. 5A2), and cancerous transformation in KO mice(Fig. 5A3). Figure 5B1 is the largest adenoma found in the WTgroup, Fig. 5B2 is an adenoma that is larger than Fig. 5B1 andwithhigh-grade dysplasia in KOmice. Figure 5B3 shows an ever largeradenocarcinoma in KO mice. Inflammatory signs such as infil-tration of inflammatory cells and crypt erosion were identified inbothWTandKOmice (Fig. 5A1–3 andB1–3).Most inflammationsigns are consideredmoderate, and essentially no great differencesbetween these two groups were identified at the final stage of thetreatment. Deep lesions reflecting chronic inflammation werefound in both WT and the KO mice as shown in Fig. 5C1–3.

Increased b-catenin expression and nucleus translocation inboth tumor and surrounding nontumor tissues of the mice

The expression and cellular localization of b-catenin isshown in Fig. 6. b-Catenin was localized peripherally in theenterocytes in WT mice. Its expression was mildly increased inthe cytosol in the nontreated KO mice (Fig. 6A–C vs. D–F).After AOM/DSS treatment, in the nontumor tissues, moreobviously increased expression of b-catenin throughout theentire cytoplasm was identified in the KO mice compared withthe WT mice (Fig. 6G–L). In the tumor tissues, both WT and KOmice displayed increased translocation of b-catenin to theentire cytoplasm as well as into the nucleus (Fig. 6M–R). Thenuclear redistribution and staining intensity were more pro-nounced in tumor tissues of the KO mice as compared with theWT mice (Fig. 6Q vs. R).

Increased S1P, reduced ceramide formation in themucosa, andincreased PAF in the feces of the KO mice

To get insight into the underlying biochemical changes, wedetermined the levels of sphingomyelin, ceramide, and S1P in themucosa of 6WT and6KOnontreatedmice. As shown in Fig. 7, thesphingomyelin levels in small intestinal and colonic mucosa didnot differ (Fig. 7A); a significant decrease in ceramide (Fig. 7B)wasidentified in small intestine but not in colon in KO mice. Inter-estingly, S1P levels (Fig. 7C) in the KO mice were significantlyincreased in both small intestine and colon, the increase beinggreater in colon than in the small intestine (53% vs. 30%). PAFlevels in colonic mucosa (Fig. 7D) and feces (Fig. 7E) from 10WTand 7 KO mice after treatment with AOM/DSS in Study C weremeasured. The results revealedmore than200% increase of PAF inthe feces of KO mice (P < 0.01) compared with that in the WTmice, whereas no significant changes in colonic mucosa wereidentified.

DiscussionThat intestinal alk-SMase may protect colon from tumorigen-

esis has been suggested for about two decades based on indirect

Figure 6.Representative micrographs showing expression of b-catenin in the colon ofnontreated and treated mice in Study D. Hoechst nuclear staining (blue) isshown inA,D, G, J,M, andP, andb-catenin immunoreactivity (green) in B, E, H,K, N, and Q. Hoechst and b-catenin labelings are merged in C, F, I, L, O, and R,rendering a faint bluish colorwhen colocalized. The rowsA to F display resultsfrom nontreated WT and KO mice. The rows G to L display the changes innontumor tissues, and M to R the tumor tissues from WT and KO aftertreatment in Study D. Increased b-catenin cytoplasmic staining intensity isobserved in nontreated KO mice and in nontumor tissues of the treated KOmice compared with the corresponding WT mice. The rows M to R show thechanges in tumor tissues, and a nuclear localization ofb-catenin is revealed bythe costaining of Hoechst and b-catenin (panels under merged). Likein nontumor tissues, tumor tissues in KO mice have an increased intensity ofcytoplasmic and nuclear b-catenin compared with WT tumor tissue; bars,20 mm.

Chen et al.





evidences such as generationof ceramide, inactivation of PAF, andthe activity reduction inpatientswith colon cancer (1, 16, 27). Thepresent study shows markedly increased tumor incidence, tumornumber, tumor size, malignant transformation, and increasedb-catenin in alk-SMase KOmice. It thus for the first time providesconvincing evidence that alk-SMase is a physiologic factor thatcounteracts both initiation and malignant transformation oftumorigenesis in the colon.

The cancer-preventing effect of alk-SMase is not manifesteduntil the organism encounters carcinogenic factors, because nospontaneous tumors were identified up to the age of 17 monthsin the KOmice. Without exposure to carcinogen, the phenotypeof the KO mice is largely normal, although the mucosa of bothcolon and small intestine show signs of hypertrophy. This signwas reported before (10) and confirmed in the present study.The hypertrophy may be linked to the increased S1P andb-catenin as found in the nontreated KO mice. These changesare themselves probably benign and may not lead to sponta-neous tumorigenesis. Upon exposure to carcinogens, the risk ofdysplasia increased in alk-SMase KO mice, as shown by theincreased formation of ACF, an early marker of dysplasia(23, 28). However, the ACF formed did not eventually resultin visible tumorigenesis after exposure of AOM. The failure maybe related to the genetic predisposition of the mice. Previousstudies have shown that C57BL/6J mice are refractory to AOM,and a combination of AOM and DSS is required to inducecolon cancer in these mice (29, 30). The genetic background ofalk-SMase KO mice is C57BL/6, which is a founder line ofC57BL/6J.

Using the combined treatment of AOM and DSS, we inducedcolonic tumors in both WT and KOmice. Both the incidence andtumor number were significantly higher in KO than in WT mice.The tumorigenesis was positively correlated with the strength of

carcinogenic treatment. Although the conventional method withone exposure to AOM and three cycles of DSS increased numberand incidence but not size of the tumors inKOmice, the enhancedmethod with three exposures of AOM and three cycles of DSSincreased all incidence, number and size of the tumors. The 3 KOmice sacrificed in Study C were due to severe diarrhea anddehydration after DSS treatment, indicating that the young KOmice aremore sensitive toDSS treatment comparedwithWTmice.Furthermore, all tumors induced in the WT mice were character-ized as adenomas, whereas about one third of the tumors in KOmice were adenocarcinomas, indicating that lack of alk-SMasefacilitates both tumorigenesis andmalignant transformation. Theresults are reminiscent of our previous findings in humans thatmalignant transformation is associated with a progressive reduc-tion of alk-SMase from 25% in chronic colitis to 75% in coloncancer (16–18).

Biochemical analysis found several changes that provideinsight of the mechanism related to the enhanced susceptibilityof carcinogenesis in the KO mice. Significant reduction of cer-amide was found in small intestinal mucosa not the colon. This isin agreement with the fact that alk-SMase level is high in themucosa of small intestine and low in colon (16). However, it isworthwhile noting that the ceramide generation from diet wassharply decreased both in small intestinal content and feces by upto 90% in the KO mice (10). The second finding related to themechanism is the increased PAF level in the intestinal lumen afterAOM/DSS treatment. PAF is a proinflammatory and proliferativefactor that can be synthesized rapidly in inflammatory tissues andreleased into the extracellular environment to trigger biologiceffects via specific receptors on the cell membrane (31–33).Increased PAF levels have been found in ulcerative colitis, nec-rotizing enterocolitis, and colorectal cancer (31, 34). Alk-SMase,by cleaving phosphocholine from PAF, induces inactivation of

Figure 7.Changes of sphingomyelin, ceramide,S1P, and PAF in the WT and KO mice.The mucosa of small intestine andcolon in nontreated WT and KO micewas scraped and lysed. Thesphingolipids in the lysates wereextracted and subjected to massspectrometry analysis and adjustedwith the protein levels in the samples.The levels of sphingomyelin, ceramide,and S1P are shown in A to C,respectively. For PAF analysis, thecolonic mucosa from the AOM/DSS-treatedmice was scraped and lysed asabove. The fecal samples werecollected, dried, and suspended insaline. The PAF levels in the lysate andsupernatant of fecal suspensions weredetermined by ELISA. The values wereadjusted with protein levels in thelysate or dried fecal weight. Changesof PAF in mucosa (D) and in feces (E);� , P < 0.05 and �� , P < 0.01, comparedwith the WT mice.






PAF (15), thereby evoking anti-inflammatory and anticancereffects.

Of interest, but not expected is the finding of increased S1P inthe KOmice in both small intestine and colonwith the increase incolon predominant. Among sphingolipid metabolites, S1P is akey molecule with cancer-promoting effects (35). Inhibitors ofSphK suppressed pathogenesis of colorectal tumors and colitis(36). The question how lack of alk-SMase results in an elevatedS1P level cannot be simply answered because of the complexity ofsphingomyelin metabolism pathways. The S1P levels are affectedby SphK to generate S1P, S1P phosphatase and S1P lyase toeliminate S1P, ceramidase to provide sphingosine for S1P for-mation, and also NPP2 and NPP6, which hydrolyze sphingosyl-phosphocholine to generate S1P and sphingosine, respectively(37). To reveal the biologic mechanism leading to elevated S1Prequires detailed lipidomic and proteomic analysis in futurestudies.

The increased S1P and PAF levels in the KO mice explain notonly the tumorigenesis but also the increased expression andnuclear translocation of b-catenin, a key transcriptional factor forcell survival (38), because both S1P and PAF are known tostimulate synthesis and translocation of b-catenin in cancer cells,including intestinal epithelial cells (39–41). Overexpression ofPAF acetylhydrolase reduces the amount of nuclear b-catenin andinduces a shift of the protein from the nucleus to the cytoplasm(42). It is interesting that an increased b-catenin is already dem-onstrated in nontreated KO mice, which may be responsible forthe hypertrophy of the gut and also render the animal at apredisposition in response to carcinogens.

Finally our results could have implications in cancer preven-tion and treatment. As we showed before, tumorigenesis isassociated with a progressive reduction of alk-SMase activity,and the activity can be increased by factors known to counteractcolon cancer such as soluble fiber, 5-ASA, and ursodeoxycholicacid, and decreased by AOM and high-fat diet supposed toenhance colon cancer development (16, 43). Providing recom-binant alk-SMase may reduce susceptibility to colon cancer

induced by carcinogens, and benefit patients with high risk ofcolon cancer. At least in animal studies instillation of recom-binant alk-SMase has been shown to improve the colitisinduced by DSS (44).

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

Authors' ContributionsConception and design: Y. Chen, S.-C. Xu, Å. Nilsson, R.-D. DuanDevelopment of methodology: Y. Chen, E. Ekblad, Y. Min, R.-D. DuanAcquisition of data (provided animals, acquired and managedpatients, provided facilities, etc.): P. Zhang, L. Yang, U. Voss, E. Ekblad,R.-D. DuanAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): Y. Chen, P. Zhang, L. Yang, U. Voss, E. Ekblad, Y. Wu,Å. Nilsson, R.-D. DuanWriting, review, and/or revision of the manuscript: S.-C. Xu, L. Yang, U. Voss,E. Ekblad, E. Hertervig, Å. Nilsson, R.-D. DuanAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases):Study supervision: S.-C. Xu, E. Hertervig, Å. Nilsson, R.-D. Duan

AcknowledgmentsThe authors thank Anna Themner-Persson for excellent assistance in path-

ologic studies and Zhengwen Zhao at Institute of Chemistry, Chinese Academyof Science, Beijing China for MS analysis.

Grant SupportThe work was supported by grants from Region Ska

�ne, The Crafoord

Foundation and Albert Pa�hlsson foundation in Sweden, and Tongji Medical

School grant and Jiangsu Shuangchuang Foundation in China.The costs of publication of this articlewere defrayed inpart by the payment of

page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received June 11, 2014; revised October 2, 2014; accepted October 28, 2014;published OnlineFirst November 7, 2014.

References1. Duan RD, Nilsson A. Metabolism of sphingolipids in the gut and its

relation to inflammation and cancer development. Prog Lipid Res2009;48: 62–72.

2. Zeisel SH, Char D, Sheard NF. Choline, phosphatidylcholine, and sphin-gomyelin in human andbovinemilk and infant formulas. J Nutr 1986;116:50–8.

3. Blank M, Cress EA, Smith ZL, Snyder F. Meats and fish consumed in theAmerican diet contain substantial amounts of ether-linked phospholipids.J Nutr 1992;122: 1656–61.

4. Hannun YA, Obeid LM. Principles of bioactive lipid signalling: lessonsfrom sphingolipids. Nat Rev Mol Cell Biol 2008;9: 139–50.

5. Spiegel S, Milstien S. Sphingosine-1-phosphate: an enigmatic signallinglipid. Nat Rev Mol Cell Biol 2003;4: 397–407.

6. Gomez-Munoz A. Ceramide-1-phosphate: a novel regulator of cell activa-tion. FEBS Lett 2004;562: 5–10.

7. Ogretmen B, Hannun YA. Biologically active sphingolipids in cancerpathogenesis and treatment. Nat Rev Cancer 2004;4: 604–16.

8. Dillehay DL, Webb SK, Schmelz EM, Merrill AHJr. Dietary sphingomyelininhibits 1,2-dimethylhydrazine-induced colon cancer in CF1 mice. J Nutr1994;124: 615–20.

9. Merrill AH Jr, Schmelz EM, Wang E, Schroeder JJ, Dillehay DL, Riley RT.Role of dietary sphingolipids and inhibitors of sphingolipidmetabolism incancer and other diseases. J Nutr 1995;125: 1677S–82S.

10. ZhangY,ChengY,HansenGH,Niels-Christiansen LL, Koentgen F,OhlssonL, et al. Crucial role of alkaline sphingomyelinase in sphingomyelindigestion: a study on enzyme knockoutmice. J Lipid Res 2011;52: 771–81.

11. Nilsson Å. The presence of sphingomyelin- and ceramide-cleavingenzymes in the small intestinal tract. Biochim Biophys Acta 1969;176:339–47.

12. Duan RD, Cheng Y, Hansen G, Hertervig E, Liu JJ, Syk I, et al. Purification,localization, and expression of human intestinal alkaline sphingomyeli-nase. J Lipid Res 2003;44: 1241–50.

13. Duan RD, Bergman T, Xu N, Wu J, Cheng Y, Duan J, et al. Identification ofhuman intestinal alkaline sphingomyelinase as a novel ecto-enzymerelated to the nucleotide phosphodiesterase family. J Biol Chem2003;278: 38528–36.

14. Stefan C, Jansen S, Bollen M. NPP-type ectophosphodiesterases: unity indiversity. Trends Biochem Sci 2005;30: 542–50.

15. Wu J, Nilsson A, Jonsson BA, StenstadH, AgaceW, Cheng Y, et al. Intestinalalkaline sphingomyelinase hydrolyses and inactivates platelet-activatingfactor by a phospholipase C activity. Biochem J 2006;394: 299–308.

16. Duan RD. Alkaline sphingomyelinase: an old enzyme with novel implica-tions. Biochim Biophys Acta 2006;1761: 281–91.

17. Hertervig E, Nilsson A, Nyberg L, Duan RD. Alkaline sphingomyelinaseactivity is decreased in human colorectal carcinoma. Cancer 1997;79:448–53.


Chen et al.




18. Sj€oqvist U, Hertervig E, Nilsson Å, Duan RD, Ost A, Tribukait B, et al.Chronic colitis is associated with a reduction of mucosal alkaline sphin-gomyelinase activity. Inflamm Bowel Dis 2002;8: 258–63.

19. Di Marzio L, Di Leo A, Cinque B, Fanini D, Agnifili A, Berloco D, et al.Detection of alkaline sphingomyelinase activity in human stool: proposedrole as a new diagnostic and prognostic marker of colorectal cancer. CancerEpidemiol Biomarkers Prev 2005;14: 856–62.

20. Duan RD, Hindorf U, Cheng Y, Bergenzaun P, Hall M, Hertervig E, et al.Changes of activity and isoforms of alkaline sphingomyelinase (nucleotidepyrophosphatase phosphodiesterase 7) in bile from patients undergoingendoscopic retrograde cholangiopancreatography. BMC Gastroenterol2014;14: 138.

21. Cheng Y, Wu J, Hertervig E, Lindgren S, Duan D, Nilsson Å, et al. Iden-tification of aberrant forms of alkaline sphingomyelinase (NPP7) associ-ated with human liver tumorigenesis. Br J Cancer 2007;97: 1441–8.

22. Wu J, Cheng Y, Nilsson A, Duan RD. Identification of one exon deletion ofintestinal alkaline sphingomyelinase in colon cancer HT-29 cells and adifferentiation-related expression of the wild-type enzyme in Caco-2 cells.Carcinogenesis 2004;25: 1327–33.

23. Bird RP. Observation and quantification of aberrant crypts in the murinecolon treated with a colon carcinogen: preliminary findings. Cancer Lett1987;37: 147–51.

24. Chumanevich AA, Poudyal D, Cui X, Davis T, Wood PA, Smith CD, et al.Suppression of colitis-driven colon cancer in mice by a novel smallmolecule inhibitor of sphingosine kinase. Carcinogenesis 2010;31:1787–93.

25. Thaker AI, Shaker A, Rao MS, Ciorba MA. Modeling colitis-associatedcancer with azoxymethane (AOM) and dextran sulfate sodium (DSS). JVis Exp 2012;pii: 4100:doi:10.3791/4100.

26. Zhao Z, Xu Y. An extremely simple method for extraction of lysopho-spholipids and phospholipids from blood samples. J Lipid Res 2010;51:652–9.

27. Garcia-Barros M, Coant N, Truman JP, Snider AJ, Hannun YA. Sphingo-lipids in colon cancer. Biochim Biophys Acta 2014;1841: 773–82.

28. Takayama T, Katsuki S, Takahashi Y, Ohi M, Mojiri S, Sakamoki S, et al.Aberrant crypt foci of the colon as precursors of adenoma and cancer. NEngl J Med 1998;339: 1277–84.

29. Suzuki R, Kohno H, Sugie S, Nakagama H, Tanaka T. Strain differences inthe susceptibility to azoxymethane and dextran sodium sulfate-inducedcolon carcinogenesis in mice. Carcinogenesis 2006;27: 162–9.

30. Clapper ML, Cooper HS, Chang WC. Dextran sulfate sodium-inducedcolitis-associated neoplasia: a promising model for the development ofchemopreventive interventions. Acta Pharmacol Sin 2007;28: 1450–9.

31. Izzo AA. PAF and the digestive tract. A review. J Pharm Pharmacol 1996;48:1103–11.

32. Marshall JC. Such stuff as dreams are made on: mediator-directed therapyin sepsis. Nat Rev Drug Discov 2003;2: 391–405.

33. Shimizu T. Lipidmediators in health anddisease: enzymes and receptors astherapeutic targets for the regulationof immunity and inflammation. AnnuRev Pharmacol Toxicol 2009;49: 123–50.

34. Guerrant RL, Fang GD, Thielman NM, Fonteles MC. Role of plateletactivating factor in the intestinal epithelial secretory and Chinese hamsterovary cell cytoskeletal responses to cholera toxin. Proc Natl Acad Sci U S A1994;91: 9655–8.

35. Spiegel S, Milstien S. Sphingosine 1-phosphate, a key cell signaling mol-ecule. J Biol Chem 2002;277: 25851–4.

36. Maines LW, Fitzpatrick LR, French KJ, Zhuang Y, Xia A, Keller SN, et al.Suppression of ulcerative colitis in mice by orally available inhibitors ofsphingosine kinase. Dig Dis Sci 2008;53: 997–1012.

37. Zimmermann H. Two novel families of ectonucleotidases: molecularstructures, catalytic properties and a search for function. Trends PharmacolSci 1999;20: 231–6.

38. Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, Vogelstein B, et al.Activation of beta-catenin-Tcf signaling in colon cancer by mutations inbeta-catenin or APC. Science 1997;275: 1787–90.

39. Matsuzaki E, Hiratsuka S, Hamachi T, Takahashi-Yanaga F, Hashimoto Y,Higashi K, et al. Sphingosine-1-phosphate promotes the nuclear translo-cation of beta-catenin and thereby induces osteoprotegerin gene expres-sion in osteoblast-like cell lines. Bone 2014;55: 315–24.

40. Greenspon J, Li R, Xiao L, Rao JN, Sun R, Strauch ED, et al. Sphingosine-1-phosphate regulates the expression of adherens junction protein E-cad-herin and enhances intestinal epithelial cell barrier function. Dig Dis Sci2010;56: 1342–53.

41. Jung HY, Jun S, Lee M, Kim HC, Wang X, Ji H, et al. PAF and EZH2induce Wnt/beta-catenin signaling hyperactivation. Mol Cell 2013;52:193–205.

42. Livnat I, Finkelshtein D, Ghosh I, Arai H, Reiner O. PAF-AH catalyticsubunits modulate the Wnt pathway in developing GABAergic neurons.Front Cell Neurosci 2010;4:pii: 19 doi 103389.

43. Andersson D, Cheng Y, Duan RD. Ursolic acid inhibits the formation ofaberrant crypt foci and affects colonic sphingomyelin hydrolyzingenzymes in azoxymethane-treated rats. J Cancer Res Clin Oncol2008;134: 101–7.

44. Andersson D, Kotarsky K, Wu J, Agace W, Duan RD. Expression of alkalinesphingomyelinase in yeast cells and anti-inflammatory effects of theexpressed enzyme in a rat colitis model. Dig Dis Sci 2009;54: 1440–8.






2015;14:259-267. Published OnlineFirst November 7, 2014.Mol Cancer Ther Ying Chen, Ping Zhang, Shu-Chang Xu, et al. (NPP7) Knockout MiceEnhanced Colonic Tumorigenesis in Alkaline Sphingomyelinase

Updated version

10.1158/1535-7163.MCT-14-0468-Tdoi:

Access the most recent version of this article at:

Cited articles

http://mct.aacrjournals.org/content/14/1/259.full.html#ref-list-1

This article cites 44 articles, 15 of which you can access for free at:

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

Subscriptions

Reprints and

[email protected]

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

Permissions

[email protected]

To request permission to re-use all or part of this article, contact the AACR Publications Department at



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-14-0468-T

http://mct.aacrjournals.org/content/14/1/259.full.html#ref-list-1

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

mailto:[email protected]

mailto:[email protected]


Enhanced colonic tumorigenesis in alkaline sphingomyelinase (NPP7) knockout mice

Documents

Transcript of Enhanced colonic tumorigenesis in alkaline sphingomyelinase (NPP7) knockout mice