Abnormal Expression of Perlecan Proteoglycan in Metastatic Melanomas1

(CAN@RRESEARCH54,5771-5774,November15,19941 Advances in Brief

Isabelle R. Cohen, Alan D. MUI'dOCh, Michael K Naso, Dario Marchetti, David Berd, and Renato V. Iozzo2 Department ofPathology and Cell Biology, Thomas Jefferson University, Philadelphia@Pennsylvania19107 (1. R. C., A. D. M., M. F. N., R. V. 1.]; Department ofTumor Biology, University 4f Texas. M. D. Anderson Cancer Center. Houston@Texas 77030 (D. MI; and Department of Medicine (D. B.J, Jefferson Cancer Institute (D. B., R. V. I.], Thomas Jefferson University, Philadelphia, Pennsylvania 19107

Abstract (7). In addition, tumor cytokines that influence the turnover and biosynthesis of proteoglycans in host cells are known to bind perlecan Abnonnal expression of proteoglycans has been implicated in cancer and can act as a reservoir of growth factors to stimulate tumor growth and metastasis primarily because these macromolecules are Involved in and angiogenesis (3). the control of cell growth and matrix assembly. In this report, we have Investigated the expression and Immunolocalizationof perlecan, a major In this report, we have investigated perlecan expression in invasive hepsran sulfate proteoglycan of basement membranes and pericellular melanomas and found that the perlecan mRNA steady-state levels mitric@, In human metastatic melanomas. Twenty-six of the 27 tumor were elevated up to 15-fold. This increase correlated with abundant ssm@ showed a significant Increase (up to 15-fold) In the perlecan deposition of perlecan proteoglycan in the pericellular matrix of the niRNA levels when compared with normal tissue. This change correlated tumor cells. Using the human melanoma cell line MeWo and its more whit a vast deposition of perlecan protein core In the pericellular matrix actively metastatic variant 70W, we were able to establish a correla ofmetastatic melanomas. Furthermore, we have established a relationship tion between perlecan up-regulation and invasiveness of the tumor between perlecan expression in clonal melanoma cells (70W) StimUlated cells. The 70W subline is able to form brain colonies in nude mice (8, with neurotrophlns and their increased invasiveness. Interestingly, perle 9) and exhibits an increased ability to invade the reconstituted base can mRNA levels were up-regulated within 10 mix of neurotrophin slim ment membrane Matrigel following NT treatment (10). Here we ulaftois, Indicating that perlecan is an early response gene This up report that the expression of perlecan protein core is up-regulated in regulation also occurred prior to heparanase production, suggesting that the 70W cells on stimulation with two different NTs, NGF and NT-3. perlecan expression and Its regulation might play a pivotal role in the Initial onset of invaslon The transcript levels were up-regulated within 10 min of NT treat ment, indicating that perlecan is an early response gene in this system. Introduction Secretion of the perlecan protein core appeared within 30 mm, there fore preceding the secretion of heparanase. These results indicate that The vascularbasementmembraneisthe firstbarriertumorcells perlecan may play a major role in the initial stages of invasion and need to cross to colonize distant organs. The major components of this during the progressive pathways of human melanoma into the inva extracellular matrix include type IV collagen, laminin, entactin, and sive phenotype. perlecan (1). During tumor invasion, the basement membrane be comes disorganized through three major steps: changes in the expres Materials and Methods sion of the constituents, alterations in their assembly, and enzymatic degradation of the matrix (2). In this cascade of events, the role of Northern Blotting. RNA from invasive melanomas and metastatic cell lines was preparedusingthe Tri-Reagent-LSprocedure(MolecularResearch perlecan, the major heparan sulfate proteoglycan of basement mem Center,inc., Cincinnati,OH) accordingto the manufacturer'srecommenda brane, has remained undefined. Perlecan is known to be involved in tions. We have analyzed27 metastases:15 to the brain,10 to lymphnodes, 1 cell migration, adhesion, and proliferation (3). Such diversity of to theaxillarylymphnodes,and1 to thestomach.As controltissues,we tested functions is probably a consequence of its chimeric structure of five 8 skin and 2 brainsamples, as well as one primarymelanomaof the foot. distinct domains, with homologies to the low-density lipoprotein Metastatic melanoma ceHs were treated with either NTs NGF or NT-3 (2 aM receptor, laminin, epidermal growth factor, and the neural cell adhe for 10 mmin serum-freeconditionsat 37Â°C)andcells werecollectedfor RNA sion molecule N-CAM3 (4). Perlecan is a proteoglycanwidely cx preparation. A 20-g.@galiquot of total RNA was separated on a 0.75% dena pressed in all basement membranes, in particular in the turing agarose gel containing formaldehyde and transferred to a nylon mem brane.The blots were hybridizedat 42Â°Cin50% formamidewith an a-32P- subendotheium of blood vessels, where it would hold a key position labeled polymerasechain reaction-generatedprobespanningperlecancDNA to influence extravasation of malignant cells (5). Several reports have between 11,225 and 12,725 base pairs (4). The membranes were washed under pointed to qualitative and quantitative changes in perlecan expression stringent conditions in 0.5X standard saline-citrate-0.1% sodium dodecyl linked to the neoplastic phenotype (3, 6). Degradation of the heparan sulfate at 65Â°Cfor1 h and exposed as indicated.For quantitation,RNAslot sulfate chains following the release of degradative enzymes such as blots were performed.A 15-@tgaliquotof totalRNA was appliedto nitrocel heparanase has been well described in the progression to malignancy lulose usingtheminifoldslot-blotsystem(SchleicherandSchuell,Inc.,Keene, NH) and air dried. Hybridization and washing were performed as described Received7t27i94;accepted9t26/94. above. Quantitationwasdeterminedby measuringlight absorbancein a den Thecostsof publicationofthisarticleweredefrayedinpartby thepaymentofpage sitometer,and normalizationwasmadeon the amountof total RNA. charges.Thisarticlemustthereforebeherebymarkedadvertisementinaccordancewith Immunohistochemistry. Frozen sections of surgically obtained tissues 18U.SC.Section1734solelyto indicatethisfact were reactedwith monoclonalantibody7B5 in a 1:10,000 dilutionof ascites 1 ThM work was supported in part by NIH Grants CA-39481 and CA-47282 (R. V. I.) and CA-39248(D. B.). R.V.I.is a recipientof a FacultyResearchAwardfrom the mediumin PBS. The boundantibodieswere reactedwith biotinylatedhorse American Cancer Society. anti-mouse secondary antibodies (Vectastain ABC kit; Vector Laboratories, 2Towhomrequestsforreprintsshouldbeaddressed,atDepartmentofPathologyand Burlingame, CA). All reagents were diluted in PBS and incubated with the Cell Biology, Thomas Jefferson University, 1020 Locust Street, JAB 249, Philadelphia, sections for 60 mm, followed by two washes in PBS. The sections were PA 19107. subsequently incubated with avidin and biotinylated horseradish peroxidase for 3 The abbreviations used are: N-CAM, neural cell adhesion molecule; NTs, neurotro _; NOF,nervegrowthfactor,NT-3,neurotrophin.3;PBS,phosphate-bufferedsaline; 10 mm, followed by an 8-mm incubationwith the peroxidase substrate. PBST, 0.05% phosphate-buffered saline-Tween 20; cDNA, complementary DNA. Sections were counterstained with hematoxylin (5). 5771

Western Blotting. Cultures of metastatic melanoma cells were treated with melanomasAllTable1 Perlecanexpressioninnormaltissuesandmetastatic eitherNTs NOF or NT-3 andthe mediumwas collectedat varioustimes after of the samples were obtained from surgical procedures and were frozen immedi stimulation. Serial dilutions of media were applied to nitrocellulose using the atelyextraction.Fifteen upon removal. Approximately 100 mg of wet tissue were used for RNA apparatus.Hybridizationg@goftotal RNA were loaded on a nylon membrane using a slot-blot minifold slot-blot system (Schleicher and Schuell, Inc.) and air dried. The at42Â°C, was performed ovemight under stringent conditions in 50% formamide membranewasblockedwith4%Carnationnonfatmilkin PBSTandincubated bydensitometrywitha probespanningtheN-CAMregionofperlecan.Quantitationwasdone with a monoclonal antibody directed against domain III of perlecan (5). After and computerRNA.Tissue integration, and the data were normalized on total three washes in PBST-0.1% milk, the membrane was incubated with an RNASkin(n=7) origin Absorbance units/hg antimouse lgG labeled with horseradish peroxidase and then was washed again 3.8Â±1.laBrain(n=2) in PBST-0.1% milk. The chemiluminescent reagent was prepared according to 4.6Â±3.7Primary the manufacturer'sinstructions(Dupont-NEN,Wilmington, DE) and inca tumor (n = 1) 5.6 batedwith the membranefor1 mm at roomtemperature.Exposureof the film Metastatic11.2Metastatic melanomas, paired (n = 10)b 19.0 Â± variedfrom 10 to 60 s. melanomas, unpaired (n = 17)c 20.6 Â±10.6 a Mean Â± SD. ResUlts and Discussion b Metastatic melanomas were paired with either normal tissue (skin, brain) or primary tumor. CMetastaticmelanomacasesforwhichcontrolsampleswerenotavailable. Elevated Expression of Perlecan in Invasive Melanomas. We have investigated tumor samples from patients with metastatic mela nomas to determine the levels of perlecan message compared with those of normal tissue or primary tumors from the same patient. Fig. perlecan (5), we localized perlecan in frozen sections of tumor biop 1 shows a typical Northern blotting analysis of a primary melanoma sies. Perlecan was widely distributed in the pericellular matrix of the (Lane 1) and the metastasis of the same patient (Lane 2). Perlecan tumor from two separate cases: one amelanotic (Fig. 2C) and one was increased about 5-fold in the metastatic sample compared to the melanotic (Fig. 2D). Strong immunoreactivity was detected particu primary tumor, when normalized on (3-actin levels. A single transcript larly aroundthe melanoma cells (arrowheads) and the sites of vas of â€”14.5kilobase was detected in all samples tested. In addition, cular proliferation (arrows). Perlecan was also present in the dermal/ reverse transcriptase-polymerase chain reaction analysis of the perle epidermal junction of the skin (Fig. 14, arrowheads) and in the can transcript detected the expected fragment (not shown), indicating subendothelialbasement membraneof the microvessels in the brain no significant alternatively spliced variant of perlecan in these tumors. cortex and the dermis (Fig. 2, A and B, arrows). Control sections Most of the metastatic tumors we examined (26/27) were found to showed no reactivitywhen the primaryperlecanantibodywas omitted express perlecan at higher levels than the normal or primary samples. (Fig. 2, E and F). There are numerous cell-matrix interactions which Quantitation of RNA slot blots by densitometric analysis indicated up occur in the multistep process leading to metastasis (2). These include to a 15-fold increase in perlecan expression in the invasive tumors the ability of the tumorcells to attachto the subendothelialbasement when compared with the normal tissue (skin, brain) or primary coun membraneof microvessels and migrate throughthe matrix into the terparts. Only one tumor tissue expressed perlecan at a level lower bloodstream. Several components of the matrix are affected during than that of the control. Table 1 indicates the relative values obtained this process, and this is the first report of an increase in the protein after quantitation for the population of tumors we analyzed. core of the proteoglycan. Disruption of the balance between the Immunolocalizatlon of Perlecan In Invasive Melanomas. Using differentcomponentsof the basementmembranecould participatein a monoclonal antibody directed against the laminin-like domain Ill of its destabilization and facilitate migration of tumor cells through the vascular wall and interstitial stroma. This elevated expression of perlecan in invasive tumors may also represent a marker for a more aggressive phenotype. Perlecan Expression Is Elevated in Melanoma Cell Lines E Treated with NTs. We then attempted to reproduce in vitro the results obtained in the tumor samples by investigating two melanoma U cell lines of high-metastatic and low metastatic potential. The 70W 0 â€˜- melanoma cells of the MeWo cell line have been characterizedfor @m i@;4 their ability to develop cerebral metastases (8, 9). When treated with A A -@ NTs, such as NGF, these cells invaded a Transwell filter coated with @ 0 a reconstituted basement membrane, thereby mimicking advanced stages of melanoma metastasis (10, 12). Furthermore, the 70W cells invade a Transwell filter as efficiently, whether it is coated with purified mouse perlecan or Matrigel, an extract of the EHS tumor @@@ 14 5 kb -+ 1 Perlecan (10). Here we examinedmelanomacell lines of the MeWo series for perlecan expression to determine any association with invasion fol lowing treatmentwith two differentNTs, NGF and NT-3. The inca bationof 70W cells with eitherNT resultedin a dramaticincreasein perlecan mRNA (Fig. 3A, Lanes 5 and 6), whereas the level of @@ â€”1@ @-actin perlecanmessage in unstimulated70W was barely detectable(Lane 4). Thesedatacorrelatedwell with thedegreeof invasionof the70W cells on stimulation with the cytokines (10). The increase in perlecan I 2 3 message occurred within a 10-mm induction, suggesting that perl Fig. 1. Northernblotof a malignantmelanomaversusthe primarytumorof the same ecan may be an early player in the invasion process. In contrast, patient. Twenty @xgoftotal RNA were separated on denaturing agarose gel and transferred to a nylon membrane.Hybridizationwasperformedusinga probespanningthe N-CAM perlecan expression was not modulated in the parental MeWo cells region of perlecan (4). Arrow, position of the 14.5-kilobase message. The same blot was (Lanes 1-3). rehybridized to a @-actinprobefor normalization. Lane 1, primary melanoma of the foot; The NT-induced up-regulation of perlecan was also detected at the Lane 2, metastatic melanoma to the cerebellum from the same patient; Lane 3, CRL1262 fibroblast cell line. proteinlevel. A serialdilutionof the mediumfromunstimulated70W 5772

PERLECAN EXPRESSION IN MELANOMAS

@ â€˜C' â€˜@: . @_ :â€˜ , .t'.:

@, 4;@@e@â€¢ w' @. @@@ B. a @. yâ€˜@J@L: . â€¢@@â€˜Y

C @,4@, â€˜be' S â€¢@ â€¢@ @â€¢ â€˜â€¢e%@'b 4 @ .â€˜ @.,-.â€˜b ..)@.%â€¢ fI@. @ . # . @s,' S â€˜, e,@. @ :Ã§' . e@':@'. â€˜

0, Â£â€˜ â€¢ . .

S. .. I. @ F

@@@ Ftt@@2. I!11r@1tn I@ @li/tHn t i n in 11i@@@i@ ui in rn@ iirI@ un@ Iii ii ntibody directed againstdomain lilA, skin showing linear immunostaining along the basement @@@@@@@@@@ iii@rnH iri@ i@mc â€¢t!1i@J@i rii,i I idci ii@l un@ti n : N i@ @it\ @tiining in the de@al imcmvessels (arrow). B, brain cotiex showing positivity oaly in the

@@@@ 11tCT@)\ t@ULttUR ( md 1) 1 in@ nkn@ t nt@i@ i@ tn mt@ L iii@ are shown: one ainelanotic (C) and one melanotic (D, asterisks). Notice the large deposits @@@@@@@@@@@@@@@@@@@@@@@@@ @tf)C1ICLi!1 .1r @tiiiJ HJI\ IJuiI\ IH @t 1k ,@nJ @i h k . Control section of melanoma (E) and brain cortex (F) lacking the primary antibody show @@@@@@ @( I 111 fl 1 I@ 1 \I \ â€¢

cells or NT-treated cells was immunoreacted with an anti-perlecan within the basement membrane. Local proteolytic activity by tumor monoclonal antibody directed against domain III of the perlecan cells leading to the release of glycosaminoglycan chains/growth factor protein core (Fig. 3B). Perlecan protein increased as early as 30 min complexes, in conjunction with changes in the sulfation pattern of the (not shown) and enhanced secretion persisted for at least 24 h (Fig. chains, is bound to influence the autocrine growth of the tumor and 3B, Lanes 3 and 5). These data also indicate that perlecan is up neovascularization (3, 6, 14). regUlated prior to the induction of heparanase activity, which appears Perlecan also has the ability to self-associate (15) and interact with only after 24 h of NT stimulation (10). type N collagen (16), laminin (17), and entactin (18, 19). With the To produceatumorata secondarysite,interactionsofthetumor homologies to the low-density lipoprotein receptor, laminin, and cells with their environment is crucial and the destabilization of the N-CAM in its cDNA (4) and gene organization (20), the release of basement membrane is one of the major events in this process. The high amounts of perlecan in the basement membrane could lead to increase in perlecan expression that we detect in the invasive tumors dramatic changes in the molecular architecture of this matrix. Gain of is very significant, but why should a gene-encoding basement mem additional adhesion and matrix binding sites might be decisive in the brane perlecan be induced as an early response gene? Signal trans invasion process. Recently, the N-CAM and laminin-globular subdo duction in response to NTs indicates association of the p'75-NT mains of perlecan have been implicated in cell adhesion and spreading receptor with a NGF-stimulated purine analogue-sensitive protein via a f31 integrin receptor interaction (21), suggesting a putative role kinase, which could modulate expression of genes involved in inva of this region of perlecan in adhesion. In addition, alternatively sion (12). Similarly, perlecan is known to bind growth and angiogenic spliced variants of perlecan, lacking domain IV and V of the mole factors through its heparan sulfate chains (13) and allows their storage cule, have been identified in Caenorhabditis elegans development, 5773

Acknowledgments A MeWo 70W We thankInge Eichstetterforexcellent technicalassistance. References

1. Yurchenco, P. D., and Schnitty, J. C. Molecular architecture ofbasement membranes. FASEB J., 4: 1577-1590, 1990. 2. Liofta, L A., Steeg, P. 5., and Stetler-Stevenson W. G. Cancer metastasis and angiogenesis: an imbalance of positive and negative regulation. Cell, 64: 327â€”336, @@ Perlecan â€” . I @*â€”14.5 kb 1991. @ mRNA i'@ 3. lozzo, R. V., Cohen, I. R., Grissel, S., and Murdoch, A. D. The biology of perlecan: @ ,@. themultifacetedheparansulfateproteoglycanofbasementmembranesandpericel @ - t& lular matrices. Biochem. J., 302: 625-639, 1994. 4. Murdoch, A. D., Dodge, 0. R., Cohen, I., Than, R. S., and lozzo R. V. Primary structure of the heparan sulfate proteoglycan from basement membrane (HSPG2/ rRNA perlecan): a chimeric molecule with multiple domains homologous to the low density lipoprotein receptor, laminin, neural cell adhesion molecules and epidermal growth factor. J. Biol. Chem., 267: 8544â€”8557, 1992. I 23456 7 5. Murdoch, A. D., Liu, B., Schwarting, R., Tuan, R. S., and loam, R. V. Widespread expression of perlecan proteoglycan in basement membranes and extracellular mat rices of human tissues as detected by a novel monoclonal antibody against domain 111 and by in situ hybridization. J. Histochem. Cytochem., 42: 239â€”249,1994. 6. lozzo,R.V.Proteoglycans:structure,functionandroleinneoplasia.Lab.Invest.,53: 373â€”396,1985. 7. Nakajima, M., Irimura, T., and Nicolson, G. L Heparanase and tumor metastasis B Hoursoftreatment J.Cell.Biochem.,36:157â€”167,1986. 8. Ishikawa, M., Dennis, J. W., Man, M. S., and Kerbel, R. S. Isolation and character 0 5 24 5 24 ization of spontaneous wheat germ agglutinin-resistant human melanoma mutants displaying remarkably different metastatic profiles in nude mice. Cancer Rca., 48: 665â€”670,1988. + + - - NGF 9. Ishikawa,M.,Fernandez,B.,andKerbel,R.S. Highlypigmentedhumanmelanoma + +NT-3 variantwhich metastasizeswidely in nudemice, includingto skin andbrain.Cancer none Res., 48: 4897â€”4903,1988. @ â€” 10. Marchetti,D.,Menter,D.,Jin,L, Nakajima,M.,andNicolson,0.L Nervegrowth C factor effects on human and mouse melanoma cell invasion and heparanase produc @ 0 â€” .@ tion.Int.J.Cancer.,55:692â€”699,1993. -@ _ _ 11. Herlyn, M. Human melanoma: development and progression. Cancer Metastasis Rev., 9: 101â€”112,1990. 1:8 12. Hermann, J. H., Menter, D. G., Marchetti, D., Hamada, J., and Nicolson, G. L Mediation of NGF-stimulated extracellular matrix invasion by the human melanoma 1 2 3 4 5 low-affinity p75 neurotrophin receptor: melanoma p75 functions independently of trkA. Mol. Biol. Cell, 4: 1205â€”1216,1993. Fig. 3. A, Northern blot of the human melanoma cell lines MeWo and 70W following 13. Vigny, M., Ollier-Hartmann, M. P., Lavigne, M., Fayein, N., Jeanny, I. C., Laurent, stimulationwith eitherNOF or NT-3. Confluentcultureswere treatedwith 50 ng/ml (2 M., and Courtois, Y. Specific binding of basic fibroblast growth factor to basement mA)of NTs for 10 mm at 37Â°Cand total RNA was isolated as indicated in â€œMaterialsand membrane-like structures and to purified heparan sulfate proteoglycan of the EHS Methods.â€•Fifteen g@gof total RNA were loaded in each lane and hybridized to a probe tumor. J. Cell Physiol., 137: 321â€”328, 1988. spanningthe N-CAM region of perlecan.Following stringentwashing conditions, the 14. Flaumenhaft, R., and Rifkin, D. Extracellular matrix regulation of growth factor and autoradiograph was exposed overnight. Lanes 1 and 4, unstimulated cells; Lanes 2 and 5, protease activity. Curr. Opin. Cell Biol., 3: 817â€”823,1991. cells stimulated with 2 [email protected];Lanes 3 and 6, cells stimulated with 2 ni@iNT-3;Lane 7, 15. Yurchenco,P. D., Cheng,Y. S., andRuben,G. C. Self-assemblyof a high molecular CRL-1262 fibroblast cell line (positive control). rRNA is shown in the lower panel. B, weight basement membrane heparan sulfate proteoglycan into dimers and oligomers. Western slot-blot analysis of the media from 70W cells following stimulation with either J. Biol. Chem., 262: 17668â€”17676, 1987. NOF or NT-3. Mediumfrom unstimulatedor stimulatedcells was collected at various 16. Fujiwara, S., Wiedemann, H., TimpI, R., Lustig, A., and Engel, J. Structure and intervals and a serial dilution of the medium was applied onto nitrocellulose membrane. interactions of heparan sulfate proteoglycans from a mouse tumor membrane. Eur. J. The blot was reacted with a monoclonal antibody directed against a laminin-like region of Biochem., 143: 145â€”157,1984. perlecan. Following chemiluminescence development, exposure time was 10 s. Lane 1, unstimulated cells; Lanes 2 and 3, cells stimulated with NGF for 5 and 24 h, respectively; 17. Timpl, R. Structure and biological activity of basement membrane proteins. Eur. J. Lanes 4 and 5, cells stimulated with NT-3 for 5 and 24 h, respectively. Biochem., 180: 487â€”502,1989. 18. Battaglia,C., Mayer,U., Aumailey, M., andTimpl,R. Basement-membraneheparan sulfateproteoglycanbindsto lamininby its heparansulfatechainsandto nidogenby sites in the protein core. Eur. J. Biochem., 208: 359â€”366,1992. causing disruption in muscle cell-body wall interactions (22). Al 19. Reinhardt, D., Mann, K., Nischt, R., Fox, J. W., Chu, M-L, Krieg, T., and Timpl, R. though it not clear which function the different domains of perlecan Mapping ofnidogen binding sites for collagen type IV, heparan sulfate proteoglycan, and zinc. J. Biol. Chem., 268: 10881â€”10887,1993. play in the invasive process, the results presented here indicate that 20. Cohen,I. R., Grassel,S., Murdoch,A. D., andlozzo, R. V. Structuralcharacterization enrichment in basement membrane perlecan is associated with the of the complete human perlecan gene and its promoter. Proc. NatL Aced. Sd. USA, metastatic potential of human melanoma cells. Elevated expression of 90: 10404â€”10408,1993. 21. Battaglia, C., Aumailley, M., Mann, K, Mayer, U., and Timpl, R. Structural basis of perlecan in invasive melanomas may represent a marker for a more @-1integrin-mediated cell adhesion to a large heparan sulfate proteoglycan from aggressive phenotype and correlate with our in vitro data. Further basement membranes. Eur. J. Cell Biol., 61: 92â€”99,1993. investigations are necessary to determine how perlecan affects inva 22. Rogalski, T. M., Williams, B. D., Mullen, U. P., and Moerman, D. G. The products of the unc-52 gene in Caenorhabditis elegans are homologous to the core protein of sion of tumor cells and which domains of the molecule may be the mammalian basement membrane heparan sulfate proteoglycan. Genes Dcv., 7: implicated in this process. 1471â€”1484,1993.

5774

Isabelle R. Cohen, Alan D. Murdoch, Michael F. Naso, et al.

Cancer Res 1994;54:5771-5774.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/54/22/5771

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/54/22/5771. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.