a link between ras and metastatic behavior of tumor cells ... · a link between ras and metastatic...

[CANCER RESEARCH 53. 1516-1521. April I. 1W]

A Link between ras and Metastatic Behavior of Tumor Cells: ras Induces CD44Promoter Activity and Leads to Low-Level Expression of Metastasis-specificVariants of CD44 in CREF Cells1

Martin Hofmann,2 Wolfgang Rudy, Ursula Giinthert,3 Stephen G. Zimmer, Volker Zawadzki, Margot ZÃ¶ller,Rosemarie B. Lichtner,4 Peter Herrlich, and Helmut Ponta

Kernforschungszentrum Karlsruhe, Institut fÃ¼rGenetik und Toxikologie, P. O. Box 3640. D-7500 Karlsruhe I. Germany Â¡M.H.. W. R.. U. G., V. TL. P. H.. H. P.\: Department ofMicrobiolog\ and Immunologo University of Kentucky. Medical Center. Lexington, Kentucky 40536-OOH4 Â¡S.G. Z. Â¡;and Instituts fÃ¼rRadiologie und Puthoph\sioÃogie IM. Z. I andImmunologie. Â¡K.B. L Â¡,Deutsches Krehsforschungszenlrum Heidelberg, Im Neuenlieitner Feld 280. D-f)9()0 Heidelberg l. German\

ABSTRACT

The activated oncogene c-Ha-ras induces expression of the surface gly-

coprotein CD44 in cloned rat embryonic fibroblasts (CREF). Induction istranscriptional as shown by transient cotransfections of c-lla-ras expres

sion constructs and CD44 promoter reporter gene constructs and dependson the presence of an VI'-1 binding site at position -110. Increased tran

script levels for the standard isoform of CD44 (CD44s) are accompaniedby the appearance of alternatively .spliced RNAs and the synthesis ofvariants of CD44 (CD44v). These CD44v molecules differ from the standard type by the addition of sequences in the extracellular portion of themolecules. The occurrence of CD44v molecules in CREF cells upon induction of the (1)44 promoter is probably due to leakiness of the splicecontrol in these cells since stable transfection with c-Ha-ras does not alter

the CD44v/total CD44 ratio. Upon ras overexpression, however, using aninducible mouse mammary tumor virus-rax construct, a transient increaseof CD44v/total ( '1)44 ratio of 3â€”4has been determined suggesting that a

burst of ras expression, in the genetic background of CREF cells, influences both promoter activity and splice control or accuracy. The expression of CI)44v proteins is responsible for the metastatic potential in avariety of tumors (U. Gimthert et ai. Cell, 65: 13-24, 1991 ). Also in CREFcells expression of CD44v correlates with metastatic behavior, ros-trans-

fected CREF cells are not only fully transformed but also give rise tometastatic spread as measured in the spontaneous metastasis assay. The.nliiiuMi.il oncogene EIA counteracts rav-induced promoter function and,

consequently, inhibits metastalic behavior without extinguishing transformation.

INTRODUCTION

Activation and overexpression of ras are not only critical steps incarcinogenesis but have also been suggested to be responsible for theacquisition of invasive and metastatic properties (1,2). In most casesof in vivo carcinogenesis, ras activation is one of the first events incellular transformation (3, 4) suggesting that several other steps inaddition to ras activation must occur prior to the appearance of metastatic behavior. However, upon transfection into and expression ofras in either one of several cultured cell lines, invasiveness andmetastatic behavior were enhanced at a frequency that suggests directaction (5, 6). For instance, ra.v expressed in CREF5 cells moderately

increased metastatic behavior in the spontaneous metastasis assay (6).We have made use of this latter system and explored the expression ofa recently discovered metastasis marker, a variant of the surface

Received 9/W92; accepted 1/23/93.The costs of publication of this article were defrayed in part hy the payment of page

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

1This work was supported by Grant He551/7-1 from the Deutsche Forschungsgemein

schaft.: Present address: Forschungsinstitut fÃ¼rMolekulare Pathologie (IMP). Dr. Bohr-Gasse

7. A-1030 Vienna. Austria.' Present address: Basel Institute for Immunology. Grenzacher Str. 487. CH-4058

Basel. Swit/erland.4 Present address: Schering AG. Postfach 650311. D- KKX)Berlin 65. Germany.s The abbreviations used are: CREF. cloned rat embryo fihroblasts; MMTV. mouse

mammary tumor virus; LTR, long terminal repeat; MAh. monoclonal antibody.

glycoprotein CD44 (7, 8). Metastatic behavior in CREF cells is indeedcorrelated with the expression of variants of CD44. Moreover, uponras synthesis under the control of a hormone-inducible promoter,

CD44 promoter activity is elevated and splice control is transientlydisturbed.

MATERIALS AND METHODS

Cell Culture. The rat pancreatic carcinoma cell lines BSp73AS andBSp73ASML (9) were grown in RPMI 1640, the 13762NF derivatives ( 10) andthe SV40 transformed human t'ihrnhlast line GM637 (obtained from National

Institute of General Medical Sciences. Camden. NJ) were grown in Dulbecco'smodified Eagle's medium, and the CREF cell line and its derivatives (11) were

grown in Eagle's minimal essential medium in 5 HIMsodium hydroxyethyl-

properazine-/V'-ethanesulfonic acid. pH 7.4. All media were supplemented

with 10% fetal calf serum, 2 HIMu-glutamine, and antibiotics.CREF/MMTV-I.TR-ra.v Cells. CREF cells were cotransfected with the

hygromycin-resistance plasmid phph (12) and a MMTV-LTR-ra.v construct(13). Transfectants were selected in Eagle's minimal essential medium. 10%

fÃªtaiscalf serum, and hygromycin (300 ug/ml). Individual clones were isolatedand analy/ed tor integration of MMTV-l.TR-ra.v DNA. Positive clones were

then screened for rtis expression upon induction with dexamethasone hy bothNorthern analyses and Western analyses using ras-specific polyclonal rahbit

antibodies kindly provided by J. Feramisco and A. SchÃ¶nthal. A clone wasselected that showed no detectable ra.v expression in the uninduced conditionand levels comparable to the c-Ha-ra.v-stably transfecled CREF cells after

treatment with dexamethasone.Metastasis Assays. For the experimental metastasis protocol. I x IO5cells

were injected into the lateral tail vein. All cell preparations used in animalexperiments were carefully prepared as single cell suspensions with 94-98%

viability determined hy trypan blue exclusion. The animals were monitoreddaily and sacrificed when becoming moribund. Lung colonies were determinedby serial sections and microscopic evaluation. Animals that had not developedsigns of disease within 180 days (nude mice) or 270 days (syngeneic rats),respectively, were considered negative. We consider that the experimentalassay measures the ability of cells to survive transport in the blood stream, toextravasate, and to develop tumor foci. It differs from a simple tumorigenicity

assay.For the spontaneous metastasis protocol, 1 x IO6cells were injected into the

right flank. Alternatively, 5 x IO4 cells were enmeshed in alginate beads and

implanted s.c. in the tail (14). The development of tumors was monitored.Animals were sacrificed when becoming moribund and overt lung mÃ©tastaseswere scored. In those cases where overt mÃ©tastaseswere not detected, at leastthree animals were carefully examined for micrometastases following paraffin-

embedded sectioning and hematoxylin and eosin staining. No metastatic fociwere detected. We avoided excision of primary tumors in metastasis assayssince this procedure has the potential of artificially allowing metastasis to takeplace. The observation period was as described above.

Other Methodology. The monoclonal antibody 1.1ASML and the complementary DNA probes used have been described previously (7). RNA andprotein blot analyses, polymerase chain reaction amplification. DNA transfec-

tions, and DNA sequencing were performed according to standard protocolsand have been described (7).

1516

Research. on February 18, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

ras INDUCTION OF METASTASIS-SPECIFIC VARIANTS OF CD44

Table I Influence of ras and adeno" EIA <EIB) on lumorigenic properties of CREF cellsAnchorange-independent growth in soft agar was determined as described (II). Briefly. 5 x IO4 cells in Eagle's minimal essential medium plus 12.5 imi sodium hydroxyethyl-

properazine-W-ethanesulfonic acid (pH 7.4), supplemented with 10% fetal calf serum, in 0.4% agarose were seeded on top of a base layer prepared in the same medium with 0.8%agarose. Cells were fed once per week with medium containing 0.4% agarose. The number of colonies (>0.1 mm diameter) was counted after 3-4 weeks. +, more than 100. Tumorigenicpotential was determined in athymic nude mice and syngeneic Fischer rat strain 344. One x 10ft cells in 0.3 ml phosphate-buffered saline were injected supraclavicularly into

3-4-week-oId female nn/nu athymic mice or s.c. into the right flank of 4-6-week-old female Fischer rats. Animals were monitored 3 times per week tor the presence of palpable tumors.All cell preparations were carefully prepared as single cell suspensions with 94â€”98%viability determined by trypan blue exclusion. For experimental metastasis assays, I x K)-**cells

were injected into the lateral tail vain. The animals were monitored daily for signs of disease and lung colonies determined by serial sections. For spontaneous metastasis assays, 1 xIO6 cells were injected into the right flank (SCF). Alternatively 5 x IO4 cells were enmeshed in alginate beads and implanted s.c. in the tail (SCT) (14). The development of tumors

was monitored and overt lung mÃ©tastaseswere scored at sacrifice.Statistical evaluation by the Fisher-Yates test (29) of the differences in numbers of animals with lung mÃ©tastasesin syngeneic rats upon injection of CREF cells, CREF/v-Ha-nw cells

or CREF/c-Ha-ra.v* cells, SCF or SCF, revealed that they are at least 95% significant. Comparison of adenoviral EIA transfectants revealed that only the difference betweenCREF/v-Ha-ra.v (2 of 6) and CREF, Adeno El A/EIB/v-Ha-ra.v (0 of 6) is noi significant (<95%); all other differences are at least 95% significant.

Tumori genie ity Lung mÃ©tastases

Nude mice

CelllineCREFCREF/v-Ha-ra.vCREF/c-Ha-ra.s*CREF.

AdenoEIA/EIBCREF.AdenoEIA/EIB/\-Ha-rasCREF.Adeno ElA/EIB/c-Ha-ras*Growth

insoft agar Nudemice0/18+

6/6+12/12+0/18+6/6+

12/12Syngeneic

rats0/189/912/120/189/912/12Tailvein0/126/612/120/126/612/12SCF0/183/312/120/180/60/12TTailvein0/126/612/120/126/612/12frSCF0/182/65/120/180/60/12SCT0/64/58/80/60/60/12

" Adeno EIA, adenoviral EIA', " c-Ha-ra.v*. T24 activated e-Ha-ras.

RESULTS

Metastatic Properties of /-Â¿Â»-transformedCREF Cells. CREF

are not tumorigenic as defined by the standard assay system [180 daysin nude mice and 270 days in syngeneics (14)]. Transfection with theactivated form of c-Ha-ras (T24) (15, 16) or with viral Ha-ras (17)

gives rise to transformed clones that are tumorigenic and, withoutprior I'Mvivo selection, metastatic in the tail vein assay (experimental

metastasis protocol) (6) (Table 1). For instance, 1 x 10s injected cells

(activated c-Ha-ras)/animal suffice to colonize the lungs in 12 of 12

animals. The ras transformants show, in addition, a moderate degreeof metastatic behavior by spreading from a s.c. site of injection orimplantation (spontaneous metastasis assay) to lymph nodes (notshown) and lung (Table 1). Upon s.c. injection of 1 X IO6 viable

tumor cells into nude mice, the c-Ha-ras transformants metastasized

in all animals [12 of 12 (Table 1)]. Upon s.c. injection into syngeneicrats, only a fraction of the animals developed lung mÃ©tastases[5 of 12(Table 1)]. The efficiency of metastasis formation was higher upon s.c.implantation into the tail than after injection into the flank (Table 1).The s.c. tail assay allows the animals to carry a tumor burden for anextended period and increases the chances of overt mÃ©tastasesto form(20-30 days, s.c. flank, versus 50-100 days, s.c. tail). The more

efficient spontaneous metastasis formation in nude mice probablyreflects the lower growth rate (6-8 days doubling time in nude miceversus 3-4 days in syngeneic animals). An immunological component

cannot, however, be ruled out.Transfection of the adenoviral gene E1AIEÃŒBled to cellular trans

formation of CREF cells [as measured by growth in soft agar (Table1)], but the transformants were not tumorigenic in animals, neither innude mice nor in syngeneic rats, and were not metastatic. Introductionof EIA/EIB into fro-transformed CREF cells abolished spontaneous

metastasis formation into the lung. The development of mÃ©tastasesupon i.v. injection was not influenced (Table 1). EIA/EIB obviouslyinhibits a process that is not limiting in hematogenic lung colonizations (i.v. injection protocol). Some parasacral lymph node mÃ©tastaseswere found after s.c. tail implantation (not shown). Since we have nodetailed kinetic study of their formation, we cannot define at whichstage EÃŒAIEÃŒBinterferes.

ras-transformed CREF Cells Express CD44s and CD44v. Sev

eral biochemical functions are probably required to convert locallygrowing tumor cells into metastatic tumor cells. The most recentlydiscovered property concerns the surface expression of CD44 and its

A B C D E

kb

4.3 â€”¿�

2.9â€”

2.0-

4,

2.

3 â€”¿�

9â€”

2.0^ lÃ¬ le

CD44v

v-Ha-ras

CD44S

GAPDH

Fig. 1. ras induces CD44 expression. Various cell lines were examined for CD44 RNAexpression. Polyadenylated RNA was prepared and 3 ug were loaded per lane. After sizefractionation on an agarose gel the RNAs were blotted onto Hybond N membrane andhybridized to either a CD44v-specific probe (sequences of pMetal from positions 941 to1108 (7)] or. after stripping the filter of radioactivity, to a CD44s probe (sequences ofpMetal from positions 1403 to 1765 (7)). Further rehybridizations were done with probesspecific for Ha-raj |2.9-kilobase (kh) Sad fragment of pEJ (16)1 and glyceraldehyde-phosphate dehydrogenase (GAPDH) ( I..Vkiloba.se Pst\ fragment from pGAPDH-13 (27)].polyadenylated RNA was prepared from: Lane A, BSp73AS cells (9); faine B,BSp73ASML cells (9); Lane C. cloned rat embryo fibroblasts (CREF. 11); Lane D, cloneof CREF cells stably transfected with v-Ha-ras (28); Lane E. clone of CREF cells stablytransfected with activated c-Ha-rav (T24) (28).

1517




kD A B C D E

205H

F G H

116-

80-

'4 Â«M

Fig. 2. CD44v proteins are synthesized upon ras expression. Protein extracts wereprepared from various cell lines and 50-pg aliquots were applied to sodium dodecylsulfate-polyacrylamide gel electrophore.sis. The proteins were blotted to a polyvinyl

difluoride membrane (Millipore, Eschborn, Germany). CD44v proteins were detected bythe MAbl.lASML followed by a second incubation with horseradish peroxidase coupledto sheep anti-mouse polyclonal antibodies. The enhanced chemiluminescence system ofAmersham was used for detection of antibody bound. kD, molecular weight is thousands.Protein extracts were prepared from: iMtie A, BSp73AS cells (9); Lane B, BSp73ASMLcells (9); Lane C, CREF cells (11); Lane D, CREF v-Ha-nu (28); Lane E, CREF c-Ha-ras(T24) (28); Lane f, CREF adenoviral E/A/EIB (II); Lane G, CREF adenoviral El AIEIB/v-Ha-ras (28); Lane H, CREF adenoviral ElA/ElB/c-Hz-ras (T24) (28).

ABCDEFGHI

Â«*1

-170

-1 1 6- 97

Fig. 3. Variant CD44 proteins are expressed in 13762NF rat mammary carcinoma cells.Western blot analysis of protein extracts of various cell lines derived from the I3762NFrat mammary carcinoma (10). Compared are cell lines derived from the parental tumor:MTPa (Lane Al, MTA (Lane fi). MTC (Lane C), MTF7 (Lane D) with cell linesestablished from lymph node and lung metastasis MTLy (Lane E). MTLn2 (iMne F).MTLn3 (LaneG). Extracts derived from BSp73AS cells (Lane Hi and BSp73ASML cells(Lane /) have also been loaded. The variant CD44-specific MAbl.lASML was used tostain the proteins as described in Fig. 2. Ordinate, positions of molecular weight markerstimes thousands.

variants. The smallest isoform of CD44 (CD44s) seems to promotehematogenic lung colonization (18) while expression of either one oftwo specific variants containing 162 or 85 amino acids, respectively,inserted into the extracellular portion of CD44s confers metastaticbehavior in the spontaneous assay (7, 19). CD44 expression was,therefore, examined in the CREF system. By Northern blot analysis,the ras transformants contained much more CD44s than non-trans

formed CREF cells (Fig. 1; compare Lanes D and E with Lane C).Ras-transformed CREF cells also express variants of CD44 detected

as RNA species hybridizing to the variant portion of the CD44v clonepMeta-1 (Fig. 1), and as protein bands in Western blots using the

MAb 1.1ASML (Fig. 2). CD44v expression as a metastatic marker wasfirst detected in the highly metastatic pancreatic carcinoma cell lineBSp73ASML (7). BSp73ASML cells and other also highly metastaticrat cell lines express high levels of several CD44v RNAs (Ref. 7; Fig.1). The sizes of the RNA species in ra.c-transformed CREF cells

correspond roughly to those in BSp73ASML. The RNA levels, however, were by far lower (Fig. 1, compare Lanes B, D, and E). Judgedby epitope staining and fluorescence-activated cell sorter analysis, thetransformants seem to carry in the order of IO4 molecules/cell surface,at least one order of magnitude less than BSp73ASML cells.6 Inter

' Unpublished observations.

estingly, there is a difference in posttranslational modification, inasmuch as BSp73ASML produce several high molecular weight speciesof CD44v (Mr 120,000-200,000 k) while the rai-transformed CREFcells carry predominantly epitope-positive proteins with molecularweights of about 100,000-120,000 (Fig. 2). Mr 100,000-120,000

forms are also found in variants of the rat mammary carcinoma13762NF that have low metastatic capacity (10), namely MTPa,MTC, MTF7, and MTA (Fig. 3). These cell lines express low levels ofsome CD44 variant specific RNA (Ref. 7, Fig. 6) Note, however, theextremely low expression of CD44v in the nonmetastatic cell linesBSp73AS and CREF. Metastases-derived cell lines of 13762 NF ex

press CD44v proteins with apparent molecular weights similar tothose observed in BSp73ASML cells (Fig. 3).

E1AIE1B transfectants qualitatively share the pattern of CD44RNAs and proteins (not shown and Fig. 2). The levels of both basalexpression and ra.v-induced expression are, however, reduced.

From Northern analyses (and from a series of reverse transcriptionpolymerase chain reaction analyses using oligonucleotides flankingthe variant region (not shown)), CD44v/total CD44 RNA ratios can becalculated. Because CD44s sequences are contained in CD44s as wellas in CD44v molecules and resolution of RNA is poor due to thegreater abundance of CD44s and heterogeneity of RNA sizes, aCD44s probe measures total CD44s and CD44v RNA and indicatespromoter activity. CD44v expression would depend on promoter activity and alternative splicing. Proportional increases would suggestleakiness of the splice control while ratio increases would indicatespecific effects on the splice control. The data of Fig. 4 demonstratera.s-dependent promoter activity. The most dramatic increase in totalCD44 transcripts was seen with \-ras transformation. Variant RNAs

were elevated by ras proportionately to the increase of promoteractivity, but the level of CD44v came never close to that detected inBSp73ASML (Fig. I ). Inhibition of promoter activity by ElA/ElBwas observed with nontransfected control cells and all ras transformants (Fig. 4). No difference in the ratio CD44v/total CD44 could bedetected with certainty between stable raÃtransfectants and control

relative IOinduction

8â€”

6â€”

2â€”

Fig. 4. Quantitation of CD44 and CD44v induction in stably raj-transfected CREFcells. Quantitation by densitometric scanning of the Northern blotting analysis shown inFig. 1 for CREF cells (A). CREF v-Ha-nu (B). CREF c-Ha-ra.s (T24) (C). A similaranalysis (not shown ) was performed for CREF adenoviral El A/El B (D ), CREF adenoviralElA/ElB/v-Ha-ras (Â£). and CREF adenoviral EIA/ElB/c-Ha-ras (T24) (F). The values

for CD44 and CD44v expression in CREF cells are arbitrarily set as 1. Note that promoteractivity has been determined by use of a probe that recognizes all forms of CD44. whereasCD44 variant expression is determined with variant exon specific probe. H. promoteractivity; â€¢¿�CD44 variant expression.

1518



rus INDUCTION OF METASTASIS-SPECIFIC VARIANTS OF CD44

cells, indicating nxs-induced promoter activation but no significantsteady-state influence on splicing. The synthesis of variants could be

explained by leakiness of the splice control.Immediate Action of ras on CD44 Expression. To explore

whether CD44s and CD44v expression were immediate consequences

DCREFâ€¢¿�CREF+dexg CREF-LTRiasâ€¢¿�CREF-LTRrastdex

24 36 48 hours

Fig. 5. Kinetics of induction of CD44 by ras. CREF cells stably transfected withMMTV-LTR-ra.v were treated and harvested as described in Fig. 5. Total RNA wasprepared. Twenty pg of total RNA each were subjected to Northern blotting analysis usingeither a CD44s-specific probe (a) or a CD44v-specific probe (b\ The two probes havebeen described in Fig. 1. Values show a quantitation of the Northern blotting analysis (asdescribed in Fig. 4). dex, dexamethasone.

of ras expression or activation or, alternatively, long-term indirect

effects of ras transformation, we stably introduced into CREF cells ahormone-inducible ras expression construct (MMTV-LTR-ra.v). The

induction by hormone of ras expression caused the appearance of theMAbl.lASML epitope on the cell surface (not shown). Northern blothybridizations revealed that ras induced both CD44s and CD44v RNAaccumulation. A quantitation is shown in Fig. 5. The increase inpromoter activity is only 3-fold and the time course is slow, with

significant elevations only after 12 h. Larger RNA splice variantscarrying epitope-encoding sequences were enhanced about 8-fold,

suggesting that under these more dramatic changes of ras expressionthere may be a ras-induced change in the splice pattern. The appear

ance of CD44v RNA was followed by a correspondingly late increasein CD44 variant protein as detected by Western blot analysis usingMAbl.lASML (Fig. 6). The accumulation did not start before 12 hafter ras induction and exceeded that in the stably transfected CREF/c-tÃ¬a-ras(T24) cell line. The transient ras increase seems also to giverise to the larger modification products around Mr 150,000-180,000.

Direct Activation of the CD44 Promoter by ras. To distinguishwhether ras increases CD44 mRNA accumulation by activating theCD44 promoter or by increasing the stability of the mRNA half-life,

we cloned CD44 promoter sequences, constructed chimeric geneswith the chloramphenicol acetyltransferase reporter, and determinedthe effect of ras overexpression on the transfected reporter constructs(Table 2). raÃoverexpression was achieved either by cotransfectingp-Ha-ras [EJ (16)] or by induction of a stably transfected MMTV-LTR-ra.5 construct. The cotransfection experiments yielded activation

of the CD44 promoter, but only low induction levels. Activated raÃcaused 1.7-fold induction (Table 2). The presence of adenovirus EIA

obliterated induction. Transfection of the reporter gene into CREFcells stably transfected with MMTV-LTR-ras gave slightly higherinduction levels. The CD44 promoter was up-regulated 2.9-fold by ras

(Table 2). A collagenase promoter chloramphenicol acetyltransferaseconstruct that had been shown to be strongly induced following rasinduction in mouse NIH3T3 (20) was also only poorly up-regulated by

ras in CREF cells (Table 2). In the immortalized human fibroblast lineGM637, ras-dependent regulation of the CD44 promoter could beconfirmed, with induction factors of 32-fold (Table 2). The collagen-

CREF/MMTV-LTR-ras

-Dex +Dex

0 6 12 36 48 0 6 12 3648 O O kDâ€”¿�205

80Fig. 6. Induction of CD44v protein synthesis by raj. CREF cells stably transfected with MMTV-LTR-ra.s ("Materials and Methods") were grown to 70% confluency. They were

then treated with dexamethasone (Dex) ( 10~7M)where indicated. At the times indicated, cells were harvested and protein extracts were prepared. They were subjected to Western blotting

analysis as described in Fig. 2. kD, molecular weight in thousands.

1519




Table 2 Induction of the CD44 promoter by rasA human CD44 promoter-chloramphenicol acetyltransterase (CAT) construct was transiently transfecled either alone or together with the ras expression plasmid pEJ ( 16) or together

with pEJ and an adenovirus E/A (Ad.EM ) expression plasmid (26) into the cell lines indicated. CAT activity was determined 24 h after transfection. The CD44 promoter CAT constructcontains the region between positions -457 and +125 (start of transcription, +1) inserted into the vector pBLCAT4 (30). Its isolation will be described elsewhere.7 The sequence isidentical to the one published (31 ). The mutant hCD44 promoter construct contains an exchange of a thymidine at position -110 to a guanine, and of a thymidine position -114 to acytosine. These two exchanges destroy a presumptive AP-1 binding site that covers the sequences from -108 to -114. Transfections were performed according to standard protocols(30) using the calcium phosphate method. C AT activity determinations were performed with appropriate amounts of cell extracts (usually about 70 p.g protein). The collagenase promoterconstruct has been described previously (31).

M57/+125)hCD44CATCREF

CREF + pEJCREF + pEJ +Ad.EMCREF

LTR rasCREF LTR ras +Dex'GM637

GM 637 +pEJGM637 + pEJ + Ad.Â£7/lCAT

activity"0.77

Â±0.4*

1.4 Â±0.70.45 Â±0.20.46

Â±0.31.36Â±0.70.15

Â±0.034.8 Â±0.45

0.26 Â±0.05Foldinduction1.7

+ 0.30.7 Â±0.22.9

Â±0.431.9

Â±4.41.6Â±0.15AP-1

mutantCATactivity"0.56

Â±0.20.57 Â±0.220.54 Â±0.260.38

Â±0.20.38Â±0.210.38

Â±0.20.38 Â±0.20.37 Â±0.2HCD44CATFold

induction1.0Â±0.1

1.0Â±0.21.0

Â±0.051.0Â±0.l1

.0 Â±0.08(-737+63

Â»coIICATCATactivity"4.26

Â±1.36.6 Â±1.22.8Â±0.60.69

Â±0.21.07Â±0.40.4

Â±0.2115.6Â±2.61.36

Â±0.4Fold

induction1.5

Â±1.20.7 Â±0.051.57Â±0.136.8

Â±3.73.1Â±0.4

" pmol of acetylated chloramphenicol/ug of protein/h.'' Mean Â±SE calculated from three independent experiments.' Dex, H)"6 M dexamethasone.

ase promoter was also efficiently up-regulated in these cells. In

GM637 cells, adenovirus EIA inhibited promoter activation as it didin CREF cells (Table 2).

Sequence analysis of the CD44 promoter revealed a presumptivebinding site for the transcription factor AP-1 at position -HO.7 In

several other genes the transcription factor AP-1 is responsible for

gene activation by the ras oncogene (20). The functional significanceof the AP-1 binding site in the CD44 promoter was examined by theintroduction of point mutations into the AP-1 recognition site. This

mutant was no longer inducible by the ras protein, neither in CREFcells nor in GM637 cells (Table 2), suggesting that AP-1 (consistingof the oncogene products fos and jun or another member of the AP-1family) mediates ras-induced activation of the CD44 promoter.

DISCUSSION

Invasive and metastatic properties are probably conferred on tumorcells by a whole set of cellular functions. Even the first step in themetastatic process, migration of tumor cells from the site of implantation to the lymph nodes (as measured in the spontaneous metastasisassay), is likely to depend on several properties, one of which is theexpression of a variant of CD44 on the cell surface. We have shownhere that in the CREF cell system expression of CD44s and CD44vcorrelates with metastatic behavior. The effect of ras-induced CD44

expression on spontaneous metastasis formation in isogeneic animalswas moderate as compared to that seen previously for cells of thepancreatic carcinoma cell line BSp73ASML spontaneously expressingCD44 variants (7). BSp73ASML cells, however, express 10 timeshigher levels of several CD44 variants than do ras-transformed CREF

cells. In immunocompromised animals, however, even cells expressing low levels of CD44v sufficed to give mÃ©tastasesin all animalsgiven injections (Table 1). This perhaps documents better survival ofcells reaching the circulation in nude mice than in isogeneic rats ormight be explained by differences in tumor growth rates (see arguments in "Results").

The putative multitude of individual properties needed for metastatic behavior could be acquired step by step through mutation andselection or be induced by pleiotropic control proteins, e.g., as theimmediate consequence of a transforming gene. Several transformingoncogenes indeed act as pleiotropic control proteins for genes relevant

7 V. Zawadzki et ai, manuscript in preparation.

to invasiveness. For instance ras, jun, or fos and other cellular oncogenes induce the expression of collagenase type I and IV (21), ofstromelysin/transin (22), and of cathepsin (23). Here we show that rasinfluences the expression of CD44 by increasing its promoter activity.The promoter resembles those of other rai-responsive genes, ras is

probably not the only determinant, and its effect on the CD44 promoter may depend on components of signal transduction not equallyabundant in all cells. Therefore even proliferating tissues could be lowor negative for CD44s or CD44v [e.g., intestinal epithelium (24) andvarious cell lines (8)).

Adenovirus is known to cause tumors of low invasiveness. Moreover, expression of adenoviral genes reduces tumor progression (25).Interestingly, the adenoviral oncogene EIA interferes with the expression of proteases and CD44 (Table 2; Figs. 2 and 4; Ref. 26). EIA isalso the likely regulatory component in our E1A/E1B transfectantssince in cotransfection experiments EIA alone suffices to reduceCD44 promoter activity (Table 2). The reduction of promoter activityfor CD44 (and perhaps other relevant functions) obliterates metastaticbehavior in both nude mice and isogeneic animals, suggesting that thelevels expressed in CREF cells were close to the lower limit required.

Splice control in vivo is extremely stringent. Specific variantsof CD44 are expressed only under circumscribed restrictive conditions (24). Low-level synthesis of CD44 variants (detected by

MAbl.lASML or by hybridization) may result from leakiness of thesplice control, which in turn may vary from cell to cell, ras-trans-

formed rat cells (not shown) produce no detectable variant whileras-transformed CREF cells synthesize low but detectable levels. A

possible direct influence of ras on splice control does need consideration. Upon transient overexpression of ras, the ratio CD44v/totalCD44 increases. The effects are modest, however, and in stable rastransfectants we have detected no shift in CD44v/total CD44 ratio ascompared to the recipient cells. Further, the synthesis of many different CD44v RNAs speaks for inhibition or relaxation of splice controlrather than specific ras-dependent splice regulation.

The turn-on of CD44v in human colon carcinogenesis is rather early(24) and may be raÃrelated since c-Ha-rcw is mutated early in col-

orectal carcinogenesis (4). The tremendous CD44v expression in thepancreatic carcinoma cell line BSp73ASML must, however, be causedby mutation, presumably in cw-acting elements of the splice mecha

nism, since CD44s expression in these cells is particularly low.The low-level CD44v in CREF ras transfectants do suffice to

mediate metastatic behavior. The low level may be close to a threshold

1520



TOÃ•INDUCTION OF MKTASTASIS-SPEriFIC VARIANTS OF CD44

required for lymphogenic spread thus reducing the efficiency as compared to highly metastatic cell lines, e.g., BSp73ASML. In all previous transfcctions using CD44v cDNA (19) CD44v expression hasalways been above the threshold. The fact that ras influences CD44expression to different degrees in different cell lines could explain thatactivation of nix does not automatically lead to full metastatic properties in most tumors.

ACKNOWLEDGMENTS

We thank LI. Rahmsdorf. C. Zahn. L. Klein, and H. Zwecker for technicalassistance and reprographic services and 1. Kammerer and C. Heinold forarrangement of text and tables.

REFERENCES

1. Greig. R G.. Koesiler. T. P.. Trainer. D. L.. Conviti. S. P.. Miles. L.. Kline. T.. Sweet.R.. Yokoyuma. S.. and Poste. G. Tumorigenic and metastatic properties of "normal"

and ra.Ã¯-transfected NIH/3T3 cells. Prix:. Nati. Acad. Sci. USA. Â«2:3698-3701. 1985.2. Marshall. C. J. How does p21""- transform cells? Trends Genet.. 7: 91-95, 1991.

3. Brown. K., Quintanilla. M.. Ramsden. M., Kcrr. I. B., Young. S., and Balmain. A.v-ra.v genes from Harvey and BALB murine sarcoma viruses can net as initiators oftwo stage mouse skin earcinogenesis. Cell, 46: 447^156, 1986.

4. Bos. J. I... Fearon. E. R.. Hamilton. S. R.. Verlaan-de Vries. M.. van Boom. J. H., vander Eh, A., and Vogelstein. B. Prevalence of ra.v gene mutations in human colorectalcancers. Nature (Lond.l. )27: 293-297. 1987.

5. Harper. J. R.. Roop. D. R.. and Yuspa. S. H. Transfection of the EJ rn\ Hu gene intokeratinocvtcs derived from carcinogen-induced mouse papillomas causes malignantprogression. Mol. Cell. Biol.. 6: 3144-3149. 1986

6. Boylan. J. F., Jackson. J.. Steiner. M. R., Shin. T. Y.. Duigou. G. J., Ros/man. T.,Fisher. P. B., and /immer, S. G. Role of the Ha-ra.v (Ru*fl) oncogene in mediatingprogression of the tumor cell phenotype. Anticancer Res.. 10: 717-724, 1990.

7. Gunthert. I'.. Hofmann. M.. Rudy. W.. Rcher. S.. ZÃ¶ller.M.. Haussmann, I., Matzku.

S.. Wenzel. A.. Ponta. H.. and Herrlich. P. A new variant of glycoprotein CD44 confersmetastatic potential to rat carcinoma cells. Cell. 65: 13-24. 1991.

8. Hofmann. M.. Rudy. W.. ZÃ¶ller,M.. TÃ¶lg.C.. Ponta. H.. Herrlich. R. and GÃ¼nther!,U.CD44 splice variants confer metastatic behavior in rats: homologous sequences areexpressed in human tumor cell lines. Cancer Res.. 51: 5292-5297. 1991.

9. Mat/.ku. S.. Komitowski. D.. Mildenherger. M., and ZÃ¶ller,M. Characterisation ofBsp 73. a spontaneous rat tumor and its in w'w; selected variants showing different

metastasi/ing capacities. Invasion Metastasis, ,f: 109-123, 1983.10. Neri. A.. Welch. D.. Kawaguchi. T.. and Nicolson. G. L. Development and biologic-

properties of malignant cell sublines and clones of spontaneously metastasizing ratmammary adenocarcinoma. J. Nati. Cancer Inst.. 6K: 507-517, 1982.

11. Fisher. P. B.. Babiss, L. E., Weinstein. I. B., and Ginsberg. H. S. Analysis of type 5adenovirus transformation with a cloned rat embryo, cell line (CREF). Proc. Nati.Acad. Sci. USA. 79: 3527-3531. 1982.

12. Gritz. L.. and Davies. J. Plasmid-encoded hygromycin B resistance: the sequence ofhygromycin B phosphotransferase gene and its expression in /-.V/iÃT/r/iÃÂ«colt and.VrtÃr/jÃ/mwvrrvcern'i.'iiae. Gene. 25: 179-188. 1983.

13. Jaggi. R.. Salmons. B.. Muellener. D., and Groner. B. The v-mra and H-ra.v oncogeneexpression represses glucocorticoid hormone-dependent transcription from the mouse

mammary tumor virus LTR. EMBO J.. 5: 2609-2616. 1986.14. Meyviseh. C.. and Mareel. M. Influence of implantation site of MC)4cell aggregates

in the formation of metastasis. Invasion Metastasis, 2: 51-60. 1982.

15. Reddy. E. P., Reynolds. R. K., Santos, E., and Barbacid. M. A point mutation isresponsible for the acquisition of transforming properties by the T24 human bladdercarcinoma oncogene. Nature (Lond.l, 300: 149-152, 1982.

16. Capon, D. J.. Chen. E. Y. Levinson. A. D., Seeburg. P. H., and Goeddel. D. V.Complete nucleotide sequences of the T24 human bladder carcinoma oncogene and itsnormal homologue. Nature (Lond.), 302: 33-37. 1983.

17. Shin, T. Y., Weeks, M. O.. Young, H. A., and Scolnick. E. M. Identification of asarcoma virus-coded phosphoprotein in nonproducer cells transformed by Kirsten orHarvey murine sarcoma virus. Virology, 96: 64-79. 1979.

18. Sy. M. S., Guo, Y-J., and Stamenkovic. I. Distinct effects of two CD44 isoforms ontumor growth in vim. J. Exp. Med.. 174: 859-866. 1991.

19. Herrlich. P.. Rudy. P.. Hofmann. M.. ZÃ¶ller.M.. Zawadzki. V. Tolg. C., Hekele. A.,Pals. S. T. Heider. K-H.. Slceman. J.. and Ponta. H. CD44 and splice variants ofCD44 in normal differentiation and tumor progression. In: D. M. Livingston. M.Hemler. and E. Mihich (eds.). Cell Adhesion Molecules. Rome: Edigraf, in press.1993.

20. SchÃ¶nthal.A.. Herrlich. P.. Rahmsdorf. H. J.. and Ponta. H. Requirement for ftts geneexpression in the transcriptional activation of collagenase by other oncogenes andphorhol esters. Cell. 54: 325-334. 1988.

21. Garbisa. S., Pozzatti, R., Muschel. R. J.. Saffiotti, U.. Ballin. M.. Goldfarb, R. H.,Khoury, G.. and Liotta. L. A. Secretion of type IV collagenolytic protease andmetastatic phenotype: Induction by transfection with c-Ha-ri/.v hut not c-Ha-rm plusAd2-Ela. Cancer Res., 47: 1523-1528. 1987.

22. Kerr. L. D., Holt. J. T.. and Matrisian. L. M. Growth factors regulate transin geneexpression by c-/m--dependent and c-/ii.v-independent pathways. Science (Washington

DC). 242: 1424-1427. 1988.23. Denhardt. D. T.. Greenberg. A. H.. Egan. S. E.. Hamilton. R. T.. and Wright. J. A.

Cysteine proteinase cathepsin L expression correlates closely with the metastaticpotential of H-rÂ«.v-transformed murine fibroblasts. Oncogene. 2: 55-59. 1987.

24. Heider. K-H.. Hofmann. M.. Horst. E., van den Berg. F., Ponta. H., Herrlich, P.. andPals. S. T. A human homologue of the rat metastasis-associated variant of CD44 is

expressed in colorectal carcinomas and adenomatous polyps. J. Cell Biol.. 120:227-233. 1993.

25. Frisch. S. M. Antioncogenic effect of adenovirus EIA in human tumor cells. Proc.Nati. Acad. Sci. USA, 8ft: 9077-9081, 1991.

26. Offringa. R., Gebel, S.. van Dam. H.. Timmers. M.. Smits. A.. Zwart. R.. Stein. B..Bos. J. L.. van der Eb, A., and Herrlich. P. A novel function of the transformingdomain of Eia: repression of AP-I activity. Cell, 62: 527-538, 1990.

27. Fort. P.. Marty. L.. Piechaczyk. M.. El Sabrouty. S.. Dani. C.. Jeanteur. P.. andBlanchard, J. M. Various rat adult tissues express only one major mRNA species fromthe glyceraldehyde-3-phosphate dehydrogenase multigenic family. Nucleic AcidsRes., /.i: 143I-I442. 1985.

28. Boghaert. E. R.. Austin. V.. and Zimmer. S. G. The influence of the presence ofadenovirus 5 Eia and Elb sequences on the pathology of rat embryonic fibroblaststransfected with activated c-Ha-rÂ«.\and \-rtts. Clin. Exp. Metastasis, 9: 231-243,

1991.29. Finney. J. The Fisher-Yates test of significance in 2 X 2 contingency table.

Biometrika. .15: 145. 1948.30. Jonat. C., Rahmsdorf. H. J., Park. K-K., Cato, A. C. B., Gebel. S.. Ponta, H., and

Herrlich. P. Anti-tumor promotion and antiinflammation: down-modulation of AP-I(FoslJuni activity by glucocorticoid hormone. Cell. 62: 1189-1204. 1990.

31. Shtivelman, E., and Bishop. M. Expression of CD44 is repressed in neuroblastomacells. Mol. Cell. Biol.. //: 5446-5453. 1991.

1521



1993;53:1516-1521. Cancer Res Martin Hofmann, Wolfgang Rudy, Ursula Günthert, et al. CellsExpression of Metastasis-specific Variants of CD44 in CREFInduces CD44 Promoter Activity and Leads to Low-Level

ras and Metastatic Behavior of Tumor Cells: rasA Link between

Updated version

http://cancerres.aacrjournals.org/content/53/7/1516

Access the most recent version of this article at:

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

Subscriptions

Reprints and

[email protected] at

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

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/53/7/1516To request permission to re-use all or part of this article, use this link



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

mailto:[email protected]



a link between ras and metastatic behavior of tumor cells ... · a link between ras and metastatic...

Documents