The Prostate 66:536 ^545 (2006)

MurineAndrogen-IndependentNeuroendocrineCarcinomaPromotesMetastasis ofHuman

ProstateCancerCell Line LNCaP

Kohsuke Uchida,1 Naoya Masumori,1* Atsushi Takahashi,1 Naoki Itoh,1

Kazunori Kato,2 Robert J. Matusik,3 and Taiji Tsukamoto1

1DepartmentofUrologic SurgeryandAndrology, SapporoMedicalUniversity SchoolofMedicine, Sapporo, Japan2DepartmentofMolecularMedicine, SapporoMedicalUniversity SchoolofMedicine, Sapporo, Japan

3DepartmentofUrologic Surgery,Vanderbilt Prostate Cancer Center,Vanderbilt UniversityMedical Center,Nashville,Tennessee

BACKGROUND. Although neuroendocrine (NE) cells in prostate cancer have beenspeculated to accelerate the growth and progression of surrounding cancer cells, the evidenceis as yet inconclusive.We investigated the effect of anNE allograft (NE-10) and its cell line, NE-CS, which were established from the prostate of the LPB-Tag 12T-10 transgenic mouse, onhuman prostate cancer cell line LNCaP.METHODS. The proliferation and pulmonary metastasis of LNCaP xenografts in athymicmicewith andwithoutNE-10 allograftswere evaluated. Boyden chamber assay andmicroarrayanalysis were performed to investigate changes in invasion/migration and mRNA of LNCaPcells under the influence of the NE cells, respectively.RESULTS. NE-10 did not influence the proliferation of LNCaP. The pulmonary metastasis ofLNCaP with NE-10 significantly increased compared to mice without it. The NE-CS cellsaccelerated the in vitro invasion/migration of adenocarcinoma cells. Increased expression ofmRNA of gelsolin was observed in LNCaP cells incubatedwith the supernatant of NE-CS cells.CONCLUSIONS. The NE-10 allograft promotes pulmonary metastasis of subcutaneouslyinoculated LNCaP cells by facilitating cell invasion. Secretions from NE cells upregulate theexpression of gelsolin, which is an actin-binding protein, resulting in acceleration of themigration of LNCaP cells. Prostate 66: 536–545, 2006. # 2005 Wiley-Liss, Inc.

KEY WORDS: prostate cancer; neuroendocrine; gelsolin

INTRODUCTION

To elucidate the molecular factors that lead to lossof androgen-dependence as well as the progression ofprostate cancer has become one of the major tasks ofcurrent research on prostate cancer. Recently, it hasfocused on epigenetic events and cellular interactionsbetween cancer cells and the surrounding cells. Therole of neuroendocrine (NE) differentiated cells inprostate cancer inparticular has attracted agreatdeal ofattention. NE cells are identified as a component ofconventional prostatic adenocarcinoma, occurring in30–100%of tumors [1,2].Androgenablation induces anincreased number of NE cells in prostate cancer [3] andthe frequency and density of NE cells are more pro-nounced in hormone-refractory prostate cancer [4].

SinceNE cells frequently lack androgen receptors (AR),they do not respond to androgen ablation therapy.However, NE cells do secrete growth-modulating

Grant sponsor: Stiftelsen Japanese-Swedish Cooperative Founda-tion; Grant sponsor: Japan Society for the Promotion of Science;Grant numbers: 13671660, 16591606; Grant sponsor: NIH; Grantnumber: CA-76142; Grant sponsor: Frances Williams PrestonLaboratory of the T.J. Martell Foundation.

*Correspondence to: NaoyaMasumori, MD, Department of Urology,Sapporo Medical University School of Medicine, S-1, W-16, Chuo-Ku, Sapporo 060-8543, Japan. E-mail: masumori@sapmed.ac.jpReceived 13 May 2005; Accepted 24 August 2005DOI 10.1002/pros.20369Published online 21 December 2005 in Wiley InterScience(www.interscience.wiley.com).

� 2005 Wiley-Liss, Inc.

neuropeptides. These observations have led to thehypothesis that NE cells are providers of paracrinefactors that accelerate the growth and progression ofsurrounding prostate cancer cells toward an androgen-independent state. However, the relationship betweenNE cells and prostatic adenocarcinoma cells remainsobscure. It is controversial whether the appearance ofNE cells indicates a poor prognosis due to their abilityto promote the progression of adenocarcinoma toandrogen independence and metastasis [4,5].

We recently developed a transgenicmousemodel ofprostate cancer composed of NE carcinoma, using arecombinant gene expressing an SV40 large T-antigen(Tag) transforming sequence under the regulatorycontrol of the rat large probasin promoter (LPB) [6].An NE-10 allograft that is transplantable into athymicmice was established from a ventral lobe of the murineprostate [7]. This allograft exhibits NE features histo-logically and immunohistologically. Cells in the NE-10allograft areweakly positive or negative for AR and thetumor shows androgen-independent growth in vivo. Inaddition, the in vitro cell line NE-CS was establishedfrom the NE-10 allograft [8]. The cells have character-istics ofNEdifferentiation similar to theNE-10 allograftexcept they no longer express Tag. Transplantation ofNE-CS cells into athymic mice indicates their tumori-genic ability.

We have reported that the NE-10 allograft promotesLNCaP (human prostate adenocarcinoma cell line)tumor progression to androgen independence [9].LNCaP tumor growth decreases in mice bearingLNCaP alone after castration. In contrast, LNCaPtumors continue to grow in those bearing both LNCaPand NE-10 tumors. On the other hand, it remainsunknownwhetherNE-10 is involved in theprogressionof prostatic adenocarcinoma in normal androgenstatus. In this study, we investigated whether the NE-10 allograft and NE-CS cell line affected the prolifera-tion, adhesion, migration, invasion, and metastaticpotential of LNCaP.

MATERIALSANDMETHODS

Cell Cultures

The human prostate adenocarcinoma cell lineLNCaP was obtained from the American Type CultureCollection, and usedwith passage numbers between 44and 49. The murine prostate neuroendocrine cancerallograft (NE-10) and its cell line (NE-CS) were estab-lished from the LPB-Tag transgenic mouse line 12T-10[7,8]. The LNCaP and NE-CS cells were maintained inculture medium [RPMI-1640 (Gibco BRL, Breda, TheNetherlands) supplemented with MEM non-essentialamino acid (10 ml/L, Gibco BRL), MEM sodiumpyruvate, penicillin-streptomycin (10 ml/L, Gibco

BRL), and containing 10% fetal bovine serum (FBS,ICNBiomedicals, CostaMesa, CA), and 7.5%NaHCO3]in 5% CO2 in a humidified incubator.

InVivo Study

LNCaP cells (5� 106 suspended in 75 ml of serum-free RPMI-1640 medium) were mixed with 75 ml ofMatrigel (Becton Dickinson, Sunnyvale, CA) in asyringe. These cells were subcutaneously injected intothe backs of 6-week-old athymic male mice (Balb/c,nu/nu, Sankyo Labo, Tokyo, Japan). At 7 weeks afterinjection, the mice with LNCaP xenografts wererandomly divided into two groups (LNCaP-only andLNCaPþNE-10). In the LNCaPþNE-10 group, a 50 mgtissue block of the NE-10 allograft was inoculated intothe contralateral back of each mouse with LNCaP. Thetumor volume of the LNCaP xenograft was measuredevery other week. The tumor volume (mm3) of LNCaPwas calculated by the formula 0.523� long diameter(mm)2� short diameter (mm). At 17 weeks afterinjection of LNCaP cells, the mice were killed. Bloodsamples were taken from the ventricle, and the lungswere removed. Bromodeoxyuridine (BrdU, RocheMolecular Biochemicals, Mannheim, Germany), 1 ml/100 g (body weight), was administered into theabdominal cavity of eachmouse at 1 hr before sacrifice.

The lung tissues were fixed in 10% formalin andembedded in paraffin to assess incorporation ofBrdU and immunohistochemistry for prostate-specific antigen (PSA) and Tag. Microscopic images ofthe PSA immunostained sections were obtained usinga digital camera equipped with a microscope. Theimages of each specimen with the largest area wereprocessed as the whole area of the lung for analysisusing NIH Image (National Institutes of Health,Bethesda, Maryland).

Serum levels of PSA were measured in blood fromthe mice with the NE-10 allograft and control mice byusing the radioimmunoassay Tandem-R PSA (SRL,Tokyo, Japan).

Immunohistochemistry for PSA,T-Antigen,BrdU, andCD31.

Tissue sections were stained with an antibodyagainst Tag using a technique reported previously [6].PSAandCD31 immunostainingwasperformedusing arabbit polyclonal anti-human antibody against PSA(Code No. A0562, 1:2000, DakoCytomation, Glostrup,Denmark) and monoclonal anti-human antibodyagainst CD31 (Code No. M0823, 1:40, DakoCytoma-tion), respectively. BrdU immunostaining was per-formed using a 5-Bromo-20-deoxy-uridine Labelingand Detection Kit II (Roche, Mannheim, Germany)according to the manufacturer’s instructions.

Neuroendocrine Cells PromoteMetastasis of AdenocarcinomaCells 537

Preparation of theMediumfor InVitro Studies

The supernatant of NE-CS cells that were incubatedin culturemedium in 5%CO2 in a humidified incubatorfor 36 hr was used as NE medium. Culture mediumwithout NE-CS cells that was left in culture dish underthe same conditions for 36 hr was prepared as controlmedium. NE medium and control medium wereused for cell proliferation assay, cell migration/invasion assay, cell adhesion assay, microarray analy-sis, Northern blot analysis, and florescence staining foractin filaments.

Cell ProliferationAssay

LNCaP cells (2� 104 cells/well) were suspended inwells of a 96-well plate. At 3 days after suspension inculture medium, it was changed to NEmedium (n¼ 9)or control medium (n¼ 9). After culture for anadditional 3 days, MTT assay was performed forestimating cellular viability using a commerciallyavailable kit (Cell Counting Kit-8, Dojindo Labora-tories, Kumamoto, Japan).

CellMigration and InvasionAssay

In vitro invasiveness of LNCaP was determinedaccording to the method described by Albini et al. [10].Briefly, in a Transwell culture chamber (Coster Science,Cambridge, MA), a polyvinylpyrrolidone-free poly-carbonate filter with an 8.0 mmpore size was precoatedwith 5 mg of fibronectin (Biomedical Technologies,Stoughton, MA) on the lower surface and 10 mg of thereconstituted basement membrane material Matrigel(Becton Dickinson, Sunnyvale, CA) on the upper sur-face. The cells that invaded across the pores werecounted under a microscope after hematoxylin andeosin staining. The experiments were carried out intriplicate. Cell migration assay was performed usingthe filter without Matrigel coating.

Two different experiments were performed. InExperiment 1, LNCaP cells (1� 105) were harvestedand placed in the upper chamber of a Boyden chamberwith 100 ml of NE medium or control medium. In thelower chamber, 600 ml of culturemediumwas added. InExperiment 2, 600 ml of culturemediumwith orwithoutNE-CS cells was incubated in the lower chamber for36hr and thenLNCaPcells (1� 105) suspended in100mlof culture medium were placed in the upper chamber.The numbers of migrating and invading cells werecounted at 2, 4, 6, and 8 hr and 4, 8, 12, and 20 hr,respectively.

Cell AdhesionAssay

Harvested LNCaP cells (1� 103) were resuspendedinNEmediumor controlmediumandplated in 24-well

plates precoated with collagen type IV (BectonDickinson, Franklin Lakes, NJ) and incubated for 1 or2 hr. Non-adherent cells were removed by aspiration,and washed three times with PBS. The total proteincontent of adherent cells was determined by the BCAmethod (Pierce, Rockford, IL). The absorbance wasdetermined at a wavelength of 562 nm in a modelmicroplate reader in triplicate experiments.

Isolation of RNAandMicroarray

LNCaP cells cultured in culture medium were har-vested when they reached subconfluence. The LNCaPcells suspended in the NE medium or control mediumwere incubated. Before (0 hr) and after treatment withNE medium or control medium for 4 hr and 8 hr, totalRNA was extracted using an RNeasy kit (Qiagen,Valencia, CA) according to the manufacturer’s instruc-tions. DNA microarray analysis was conducted byKurabo Corp. Target preparation, CodeLinkTM DNAmicroarray hybridization (Amersham Bioscience,Piscataway, NJ), and processing were performed asdescribed previously [11]. The fragmented targetcRNA was used for hybridization of each UniSetHuman 1 Expression Bioarray chip. In this analysis,1.5-fold up or downregulation was defined as a signi-ficant change.

Northern Blot Analysis

To confirm the reliability of results obtained from themicroarray, Northern blot analysis was performedusing total RNA derived from LNCaP cells at 4 hr and8 hr after incubation with NE medium or controlmedium. Electrophoresis, blotting, and hybridizationwere done according to a standard protocol. Comple-mentary DNA probes were generated by RT-PCR fromtotal RNA of LNCaP cells. After amplification bythe conventional PCR technique (Gelsolin: forward50-ACTGGTCTACTGTGTCTCTA-30 and reverse 50-TCTTCAGCCCACACTTTCTG-30), the PCR productswere purified using a Gel Extraction Kit (Qiagen) and arandom hexamer labeled with alkaline phosphatase(AmershamAlkphosDirect). The hybridization signalswere detected with CDP-Star Chemiluminescent De-tectionReagent and exposed tohyperfilm (Amersham).Ethidium bromide staining of 28 S RNA was used tocompare the amounts of loading.

Fluorescence Staining forActin Filaments

Harvested LNCaP cells were resuspended in NEmedium or control medium and incubated in a 2-wellculture slide precoatedwith fibronectin for 5 hr (BectonDickinson, Franklin Lakes, NJ). After discarding themedium, adherent cells were fixed with 4% parafor-maldehyde, permeabilized with 0.5% Triton X-100 for

538 Uchida et al.

5 min, and stained with 0.2 mg/ml FITC-labeledphalloidin (Sigma, St. Louis, MO), which binds actinfilaments for 20 min. After washing, the slides werecoversliped and observed under a Zeiss LSM510confocalmicroscope (Carl Zeiss, Inc., Thornwood,NY).

Statistical Analysis

Statistical analysis was performed with StatView4.5 (Abacus Concepts, Inc., Berkeley, CA). The Mann–Whitney U-test was applied to compare resultsbetween two different groups. Repeated-measuresANOVA was used when comparing the in vitro cellmigration and invasiveness in an individual group.Statistical significance was assigned at P< 0.05.

RESULTS

NECellsHaveno Effecton the Proliferation ofAdenocarcinomaCells

To investigate whether the NE-10 allograft alteredthe growthof theLNCaPxenograft in vivo, the allograftwas co-transplanted into the contralateral back inathymic mice with the LNCaP xenograft. There wasno significant difference in the tumor size of LNCaPxenografts between the mice with NE-10 and thosewithout NE-10 for 10 weeks (Fig. 1A). The supernatantof NE-CS cells (NE medium) did not affect the growthof LNCaP cells in vitro on MTT assay (Fig. 1B), whichwas consistent with the result of the in vivo study. Thelevels of serum PSA were 189� 86 ng/ml in the micewith NE-10 and 260� 287 ng/ml in the mice withoutNE-10 (mean� standarddeviation),with no significantdifference between the two groups. These resultsindicated that NE cells had no effect on the growth ofLNCaP cells in vitro or in vivo.

NECells PromoteMetastasis ofAdenocarcinomaCells

We examined whether the NE-10 allograft influ-enced the metastatic ability of LNCaP cells. PSA-positive cells, which indicated LNCaP cells in thelungs, were detected in six of nine mice in the LNCaP-only group, and infiveofninemice in theLNCaPþNE-10 group (Fig. 2A). Although, it has been reported thatmetastasis from the subcutaneously inoculated LNCaPxenograft is extremely rare if examined by routinehistological examination with hematoxylin and eosinstaining [12], careful examination by PSA immunos-taining detected PSA-positive cells in lungs in theLNCaP-only group.Another set of experiments using adifferent lot number of LNCaP cells showed similarresults (data not shown). The number of the clusters ofthe PSA-positive cells in the LNCaPþNE-10 group

was significantly greater than that in the LNCaP-onlygroup (P¼ 0.012, Fig. 2B-1). Likewise, the total area ofPSA-positive cells per whole lung was significantlylarger than that in the LNCaP-only group (P¼ 0.012,Fig. 2B-2). The PSA-positive cells in the LNCaPþNE-10 group were not located adjacent to the Tag-positivecells representing metastases from the NE-10 allograft(Fig. 3).

Fig. 1. A:GrowthrateofLNCaPxenograftinathymicmalemice.LNCaP cells (5�106) with Matrigel were subcutaneously injectedinto the backs of 6-week-old athymicmice. At 7weeks after injec-tion, the mice were randomly divided into two groups (open box,LNCaP-only, n¼ 9; open circle, LNCaPþNE-10, n¼ 9). For theLNCaPþNE-10 group, a 50 mg tissue block of the NE-10 allograftwas inoculated into the contralateral back inmicewith the LNCaPxenograft.Tumor volume (mm3) is represented as themean of ninemiceineachgroup.Verticalbars indicate standarddeviations.Therewas no significant difference in the size of the LNCaP xenograftbetween the two groups. B: Proliferation of LNCaP cells in aninvitro study.Cellproliferationwas evaluatedusingMTTassay.TherewasnosignificantdifferenceinopticaldensitybetweenLNCaPcellsinNEmedium and those in controlmedium (P¼ 0.531,Mann^WhitneyU-test).Boxpresents25th^75thpercentiles,aswellasmedian(centerline).Barsindicate5thand95thpercentiles.O.D.,opticaldensity.

NeuroendocrineCells PromoteMetastasis of AdenocarcinomaCells 539

We examined whether the PSA-positive cells in thelungs were metastases or just tumor emboli of the cellsfrom the LNCaP xenograft. The nuclei of the PSA-positive cells in serial sections were stained a uniform,dense blue color on BrdU immunohistochemistrywithout Tag expression (Fig. 3A–C). Immunostainingfor CD31 showed that some clusters of PSA-positive

cells lacked the surroundingCD31-positive endothelialcells (Fig. 3D). On the other hand, obvious invasioninto the alveoli was not detected when the clusters ofPSA-positive cells were relatively small. Thus, not allclusters of PSA-positive cells in the lungs weremetastases defined as lesions accompanied by escapefrom the vascular lumina into the alveoli in the presentstudy. However, if we maintained the mice for morethan 17 weeks after the inoculation of LNCaP cells, thesmall tumor emboli that are prerequisite for metastaticdevelopment might have become apparent metastasesas their tumor size increased since the PSA-positivecells in the lungs had proliferative ability and LNCaPcells possessed invasive potential and could enter intothe vascular circulation at the subcutaneously inocu-lated primary site. Furthermore, it was unlikely thatmechanical destruction of the pulmonary tissues byNE-10 metastasis provided a scaffold for LNCaPmetastasis because no Tag-positive cells (NE-10 cells)were observed around the metastatic nodules ofLNCaP. Thus, the NE-10 allograft promoted pulmon-ary metastasis of the subcutaneously inoculatedLNCaP xenograft.

Fig. 2. A: Representative NIH images and immunohistochemis-try for PSA.Themetastatic abilityof LNCaP affectedby theNE-10allograft was evaluated in an in vivo study.The LNCaP cells in thelungsof theathymicmiceweredetectedusingimmunohistochemis-try forPSA(leftpanels).ImagesofPSA-immunostainedslideswereobtained by digital camera and converted to NIH images (rightpanels). Black areas in NIH images show clusters of PSA-positivecells per unit lung area. Many clusters of PSA-positive cells wereobserved in the LNCaPþNE-10 group compared to the LNCaP-only group.The clusters of PSA-positive cells in the LNCaPþNEgroup were generally larger than those of the LNCaP-only group.Reducedfrom�200.B:Quantitative analysis of thenumberofclus-ters of PSA-positive cells (B-1) and the percentage of PSA-positivecells per whole lung section (B-2).The number and percentage ofclusters of PSA-positive cells per whole lung section were quanti-fied.The PSA-positive clusters in the LNCaPþNE-10 group weresignificantlygreater in number (B-1) and extent (B-2) than those oftheLNCaP-onlygroup(a,P¼ 0.012;b,P¼ 0.012;Mann^WhitneyU-test).Boxpresents25th^75thpercentiles, aswellasmedian(centerline).Barsindicate5thand95thpercentiles.

Fig. 3. A: Immunohistochemistry for PSA; (B) Immunohisto-chemistry forT-antigen; (C) Immunohistochemistry for BrdU; (D)Immunohistochemistry for CD31. (A, B, and Cwere in serial sec-tions).The clusters of PSA-positive cells show positive staining forBrdU.The PSA-positive cells (LNCaP) do notneighborTag-positivecells (NE cells). Some clusters of PSA-positive cells lacked CD31-positive endothelial cells (arrow). The results indicated that thePSA-positive cells, which hadproliferative activity accompaniedbyescape from the vascular lumina into the alveoli, weremetastasesfrom the LNCaP xenografts thatwere inoculated into the backs ofathymicmice.Reducedfrom�400.

540 Uchida et al.

NECells Accelerate the InVitro Invasion andMigration of AdenocarcinomaCells, but

NotCell Adhesion

To identify the mechanism by which NE-10 pro-moted metastasis of LNCaP, we assessed the effect ofNE cells on the migration and invasion of LNCaP cellsusing a Boyden chamber assay.WhenNEmediumwasadded to the upper chamber with LNCaP cells(Experiment 1), the number of LNCaP cells penetratingthrough the porous filter without Matrigel coating wassignificantly greater than that of LNCaP cells in controlmedium at each given point (Fig. 4A-1). Likewise,when the LNCaP cells were co-cultured with NE-CScells in the lower chamber (Experiment 2), the numberof penetrating LNCaP cells significantly increased(Fig. 4A-2). These results indicated that NE cells

enhanced the migration of LNCaP cells. Similarly, thenumber of LNCaP cells that invaded through the filterwith Matrigel significantly increased in the NE med-ium (Fig. 4B-1, Experiment 1) and in cocultured withNE-CS cells (Fig. 4B-2, Experiment 2), suggesting thatNE cells enhanced the invasion of LNCaP cells. Thus, itwas speculated that secretions from the NE-CS cellsmight enhance the migration and in vitro invasion ofLNCaP cells. In contrast, the cell adhesion on collagentype IV did not significantly differ between the LNCaPcellswith andwithoutNEmedium (Fig. 4C). Therefore,the NE-10 allograft promoted pulmonary metastasis ofLNCaP cells through enhancement of invasion byfacilitating their migration. On the other hand, NE cellshad no effect on the proliferation and adhesion to theextracellular matrix of LNCaP cells.

NECells Increase theRNAExpressionof EightGenesof AdenocarcinomaCellsThat Include the

Actin-Regulating ProteinGelsolin

We examined which genes in LNCaP cells wereinfluenced by treatment with NE medium, usingCodeLinkTM DNA microarray analysis. The micro-array for 10,458 gene profiles detected eight genes withupregulation by NE medium (Table I). Of the eight,we focusedon gelsolin since it is an important regulatorof actin cytoskeleton dynamics required for cellmigration. The expression of mRNA of gelsolin wasincreased by NE medium 1.9- and 1.5-fold at 4 hr and8 hr, respectively (Fig. 5A). The increased expression ofgelsolin induced by NE medium was confirmed using

Fig. 4. A:MigrationofLNCaPcells.Cellmigrationassaywasper-formedusingaporousfilterwithoutMatrigelcoating.InExperiment1(A-1),1�105ofLNCaPcellswereplacedintotheupperchamberofaBoyden chamberwithNEmedium (open circle) or controlmedium(opensquare).InExperiment2(A-2),culturemediumwith(opencir-cle)orwithoutNE-CScells(opensquare)wasincubatedinthelowerchamber for 36 hr and then LNCaPcells suspended in culturemed-ium were placed into the upper chamber.The NE-CS cells signifi-cantly facilitated the migration of LNCaP cells (a, P¼ 0.005; b,P< 0.001; repeated-measures ANOVA).Vertical bars indicate stan-dard deviations of triplicate experiments. B: In vitro invasion ofLNCaPcells.Invasionassaywasperformedusingaporous filterwithMatrigel coating.Protocols of experiments1 (B-1) and 2 (B-2)werethesameas thosefor thecellmigrationassay.TheNE-CScells signif-icantly enhanced the invasion of LNCaP cells (c, P< 0.001; d,P< 0.001; repeated-measures ANOVA).Vertical bars indicate stan-dard deviations of triplicate experiments.C: Adhesion of LNCaPcells.LNCaP cells (1�103) were seeded ontowells coatedwith col-lagentypeIVandincubatedat378Cfor1hrandtwohr.NEmediumdidnot significantly stimulate theadhesionofLNCaPcells (e,P¼ 0.191; f,P¼ 0.239;Mann^WhitneyU-test).Boxpresents25th^75thpercen-tiles, aswell asmedian (center line) of triplicate experiments.O.D.,opticaldensity;C-M,controlmedium;NE-M,NEmedium.

Neuroendocrine Cells PromoteMetastasis of AdenocarcinomaCells 541

Northern blot analysis (Fig. 5B). Thus, itwas speculatedthat the upregulation of gelsolin in LNCaP cells bysecretions from NE cells was involved in facilitation ofmigration of LNCaP cells.

NECells Change theMorphologyofAdenocarcinomaCells

We analyzed how NE medium influenced the actinfilament dynamics in LNCaP cells. It increased theproportion of LNCaP cells with obvious protrusions(Fig. 6). The conversion of the actin filaments occurredat the bases of protrusions of LNCaP cells. These resultssuggested that the morphology of LNCaP cells waschanged by treatment with NE medium.

DISCUSSION

Although much progress has recently been made inidentifying molecular events leading to the develop-ment of prostate cancer, the exact mechanisms under-lying the acquisition of androgen independence remainpoorly understood [13,14]. The possible mechanismshave been reported to be alternations of the activity,function and specificity of AR by mutation or ampli-fication of the AR gene [15,16], and activation ofintracellular signal transduction pathways that stimu-late AR [17,18]. Thus, the mechanism of androgenindependence is multifactorial. In addition, recentreports have posited a role of NE differentiated cellsin the progression and androgen independence ofprostate cancer [9].

NE cells produce growth factors, including vascularendothelial growth factor and transforming growthfactor a, which may stimulate growth and accelerateprogression of the surrounding adenocarcinoma cellsin a paracrine fashion [19]. In addition, NE cells secrete

TABLE I. Genes Signif|cantlyOverexpressedin LNCaPCells

ACCa Gene name

Expression ratio

4 hr 8 hr

AB020716 KIAA0909 Protein 1.6 1.6NM_000177 Gelsolin 1.9 1.5NM_003004 Secreted and transmembrane 1 (SECTM1) 2.1 1.6NM_005224 Dead ringer-like 1 (Drosophila) (DRIL1) 2.0 1.9NM_005889 Apolipoprotein B mRNA editing enzyme, catalytic polypeptide 1 (APOBEC1),

transcript variant 21.5 1.8

NM_015196 KIAA0922 protein 1.6 1.5NM_018321 Hypothetical protein FLJ11100 (FLJ11100) 1.5 1.7NM_021270 LE hypothetical protein FLJ11100 (FLJ11100) leukocyte-associated IG-like receptor 2

(LAIR2), transcript variant 21.9 1.5

Genes with an expression ratio >1.5 were considered to be significantly overexpressed.aGeneBank accession number.

Fig. 5. A: Images ofDNAmicroarrays including gelsolin spots inLNCaPcells.The slides show several spots stainedwithCy5-strep-tavidin conjugate (red circles). LNCaP cells with NEmedium showprominent signal intensity for gelsolin (arrow) compared to thosewithcontrolmediumat4and8hr incubation.B:Northernblotana-lysis of mRNA of gelsolin in LNCaP cells. Northern blot analysisdemonstratesmarkedly increased expression of gelsolin in LNCaPcellswithNEmediumcomparedto controlmedium.

542 Uchida et al.

a variety of neuropeptides and biogenic amines such asserotonin [1]. It has been clinically reported that thenumber of NE cells that lack AR increases duringandrogen ablation therapy [3]. These results indicatethat NE cells might play an important role in the devel-opment and progression of prostate cancer. However,the prognostic value of NE differentiation is clinicallycontroversial [2,20]. Thus, to investigate the relation-ship between NE cells and adenocarcinoma cells aremandatory in basic research.

The most critical problem in studying the progres-sion of prostate cancer is the lack of adequate modelsystems. Although new cell lines and xenografts forhuman models for NE prostatic carcinoma have beendescribed [21,22], there has been no direct in vivoevidence on whether NE cells influence the develop-ment and progression of prostate cancer and androgenindependence. Recently, we established the NE-10allograft and the NE-CS cell line from the ventralprostate of LPB-Tag transgenic mouse line 10 (12T-10)[7,8]. NE-10/NE-CS has the NE property representedby dense core granules in the cytoplasm and androgen-independent growth due to being negative or weaklypositive for AR. The development of this NE allograftand NE-CS cell line has provided us with an opportu-nity to investigate the role ofNE cells in prostate cancer.

Previously, we demonstrated that the NE-10 allo-graft promoted LNCaP tumor growth in castratedmice[9]. In vitro and in vivo studies demonstrated that thiseffect was mediated by increased AR level/activity inLNCaP cells by NE secretions combined with a lowandrogen concentration. In the present study, wedemonstrated that the NE-10 allograft promoted themetastasis of the LNCaP xenograft in mice having anormal testicular androgen level. The metastaticprocess consists of multiple steps, including growthat the primary site, invasion into vessels, circulation tothe metastatic site, extravasation and growth at distant

organs [23]. Cancer cell invasion is a crucial phenom-enon for metastasis. It is composed of three steps:cancer cell attachment to the basement membrane,degradation of the extracellular matrix by proteolyticenzymes, and cell migration [24]. We found that theNE-CE cells promoted migration and invasion ofLNCaP cells but not cell proliferation and attachment.Thus, it was speculated that the NE cells promotedpulmonary metastasis of LNCaP cells throughenhancement of cell invasion by facilitating theirmigration.

The DNA microarray and Northern blot analysisshowed that expression ofmRNAof gelsolin in LNCaPcells was increased by the supernatant of NE-CS cells.Gelsolin is an actin-binding protein with well-char-acterized functions for cytoskeltal reorganization, cellmorphology, and motility. Activated gelsolin binds toassembled actin filaments and severs the actin intosmall fragments (severing). After severing, gelsolinremains attached to the barbed ends of the fragments asa cap to prevent reannealing with each other orelongation at the barbed end (capping). Gelsolin alsopromotes polymerization of actin monomers to elon-gate actin filaments (nucleating). Thus, gelsolin med-iates thedynamic changes in the actin cytoskeleton for avariety of forms of cell motility [25,26]. In addition, it isknown that gelsolin is involved in cellular apoptosis,since it is identified as a substrate of caspase-3 [27].Gelsolin expression is frequently downregulated inseveral types of human cancer [28,29], includingprostate cancer [30]. Although gelsolin is consideredto be a candidate tumor suppressor gene, its role incarcinogenesis remains unclear. Recently, it has beenreported that higher expression of gelsolin is associatedwith a higher risk of recurrence in early stage non-smallcell lung cancer and breast cancer [31,32]. In a study onurothelial carcinoma, Rao et al. [33] reported thatgelsolin expression decreased in carcinoma in situ anddysplastic lesions compared to benign areas, whereas itgradually increased as the grade and stage becamehigher. They also demonstrated that higher expressionof gelsolinwas an independent predictor for recurrenceand progression.

In vitro studies have shown that increased expres-sion of gelsolin in cultured fibroblasts results in anincrease of cell migration [34,35]. Similarly, it has beendemonstrated that overexpression of gelsolin promotesinvasion of epithelial cells (MDCK andHEK293T cells)[36]. In the present study, NE cells facilitated themigration of LNCaP cells and increased expression ofgelsolin mRNA in LNCaP cells. We also observed thatthe NE cells enhanced remodeling of the cytoskeletalorganization of LNCaP cells. Based upon our evidenceand previous studies, it is likely that gelsolin plays animportant role in LNCaP tumor cell motility and

Fig. 6. The LNCaP cells were cultured on fibronectin-coatedchamberslidesfor6hrincontrolmedium(A)orNEmedium(B).Fila-mentous actinwasvisualizedusingTRITC-labeledphalloidinunder aconfocal lasermicroscope.The LNCaP cells with NEmedium pos-sessmoreprominentprotrusionswithconvertingactin filaments atthebase thanthosewithcontrol-medium.

Neuroendocrine Cells PromoteMetastasis of AdenocarcinomaCells 543

invasion, although further study is mandatory toconfirm the relationship between upregulation ofgelsolin and cell motility in our model.

Nishimura et al. [37] reported that androgenablationincreased the expression of gelsolin inLNCaP cells. Theincreasing expression of gelsolin enhanced the ARactivity under low androgen level, which suggestedthat gelsolin contributed in maintaining a functionalAR signaling pathway even in a low androgenenvironment. Thus, gelsolin may have a biphasicfunction in the carcinogenesis and progression ofprostate cancer. Expression of gelsolin may be down-regulated in the early stage of carcinogenesis, but maylater switch to upregulation and promotemetastasis byfacilitating tumor cell motility. The surrounding NEcells may contribute to upregulation of mRNA ofgelsolin in adenocarcinoma cells as well as a change inthe characteristics of adenocarcinoma cells per se byandrogen ablation since androgen ablation inducesLNCaP cells to transdifferentiate into an NE-likephenotype [38]. Our previous study also demonstratedthat NE secretions activated AR of LNCaP cells underlow androgen levels [9]. Gelsolin, which is upregulatedby ablation of testicular androgens may be involved inthe development of hormone-refractory prostate can-cer where the cancer can respond to low levels ofadrenal androgens.

There are several limitations to this study. TheNE-10allograft and the NE-CS cell line, which were derivedfrom the mouse prostate, are of different origin fromhuman adenocarcinoma cell line LNCaP. The role ofhuman NE cells in human prostate cancer may not bethe same as mouse NE cells. In addition, the character-istics of the established cell line, NE-CS could bedifferent from those of the original NE-10 allograftbecause cells suitable for survival in vitrowere selectedduring establishment of the cell line. Thus, peptides oramines secreted from NE cells may be different in NE-10 allograft and NE-CS cells. There are no ideal humanlines for which both in vitro and in vivo NEmodels areavailable. In addition, it is unknown which factorssecreted from NE cells are involved in increasedexpression of gelsolin. Because NE-10/NE-CS pro-duces several neuropeptides such as chromogranin A,serotonin, and somatostatin, it will be necessary toinvestigate if these or other factors are crucial forupregulation of gelsolin in LNCaP cells.

Although we need to generate further evidenceabout the importance of the interaction betweenadenocarcinoma cells and NE cells in the field ofprostate cancer, our results establish a significant rolefor NE cells in vivo. Learning how to target NE cellsmay lead to a breakthrough in controlling progressionof prostate cancer tometastasis and becominghormonerefractory to androgen ablation treatment.

REFERENCES

1. di Sant’Agnese PA. Neuroendocrine differentiation in carci-noma of the prostate. Diagnostic, prognostic, and therapeuticimplications. Cancer 1992;70:254–268.

2. Abrahamsson PA. Neuroendocrine differentiation in prostaticcarcinoma. Prostate 1999;39:135–148.

3. Jiborn T, Bjartell A, Abrahamsson PA. Neuroendocrine differ-entiation in prostatic carcinoma during hormonal treatment.Urology 1998;51:585–589.

4. CasellaR,BubendorfL,SauterG,MochH,MihatschMJ,GasserTC.Focalneuroendocrinedifferentiation lacksprognostic significancein prostate core needle biopsies. J Urol 1998;160:406–410.

5. Abrahamsson PA, Cockett AT, di Sant’Agnese PA. Prognosticsignificance of neuroendocrine differentiation in clinicallylocalized prostatic carcinoma. Prostate 1998;8:37–42.

6. MasumoriN,ThomasTZ,ChaurandP,CaseT, PaulM,Kasper S,Caprioli RM,TsukamotoT, Shappell SB,MatusikRJ.Aprobasin-large T antigen transgenic mouse line develops prostateadenocarcinomaandneuroendocrine carcinomawithmetastaticpotential. Cancer Res 2001;61:2239–2249.

7. MasumoriN,TsuchiyaK,TuWH,LeeC,KasperS, TsukamotoT,Shappell SB, Matusik RJ. An allograft model of androgenindependent prostatic neuroendocrine carcinoma derived froma large probasin promoter-T antigen transgenic mouse line. JUrol 2004;171:439–442.

8. Uchida K, Masumori N, Takahashi A, Itoh N, Tsukamoto T.Characterization of prostatic neuroendocrine cell line estab-lished from neuroendocrine carcinoma of transgenic mouseallograft model. Prostate 2005;62:40–48.

9. Jin RJ,Wang Y,Masumori N, Ishii K, Tsukamoto T, Shappell SB,Hayward SW, Kasper S, Matusik RJ. NE-10 neuroendocrinecancer promotes the LNCaP xenograft growth in castratedmice.Cancer Res 2004;64:5489–5495.

10. Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson SA,Kozlowski JM, McEwan RN. A rapid in vitro assay forquantitating the invasive potential of tumor cells. Cancer Res1987;47:3239–3245.

11. Dorris DR, Ramakrishnan R, Trakas D, Dudzik F, Belval R, ZhaoC, Nguyen A, Domanus M, Mazumder A. A highly reproduci-ble, linear, and automated sample preparationmethod for DNAmicroarrays. Genome Res 2002;12:976–984.

12. Lim DJ, Liu XL, Sutkowski DM, Braun EJ, Lee C, Kozlowski JM.Growth of an androgen-sensitive human prostate cancer cellline, LNCaP, in nude mice. Prostate 1993;22:109–118.

13. NelsonWG, DeMarzo AM, IsaacsWB. Prostate cancer. N Engl JMed 2003;349:366–381.

14. Grossmann ME, Huang H, Tindall DJ. Androgen receptorsignaling in androgen-refractory prostate cancer. J Natl CancerInst 2001;93:1687–1697.

15. Taplin ME, Bubley GJ, Shuster TD, Frantz ME, Spooner AE,OgataGK,KeerHN, Balk SP.Mutation of the androgen-receptorgene in metastatic androgen-independent prostate cancer. NEngl J Med 1995;332:1393–1398.

16. Thompson J, Hyytinen ER,HaapalaK, Rantala I, HelinHJ, JanneOA, Palvimo JJ, Koivisto PA. Androgen receptor mutations inhigh-grade prostate cancer before hormonal therapy. Lab Invest2003;83:1709–1713.

17. Craft N, Shostak Y, Carey M, Sawyers CL. A mechanism forhormone-independent prostate cancer through modulation ofandrogen receptor signaling by the HER-2/neu tyrosine kinase.Nat Med 1999;5:280–285.

544 Uchida et al.

18. Gioeli D, Mandell JW, Petroni GR, Frierson HF Jr, Weber MJ.Activation of mitogen-activated protein kinase associated withprostate cancer progression. Cancer Res 1999;59:279–284.

19. Harper ME, Glynne-Jones E, Goddard L, Thurston VJ, GriffithsK. Vascular endothelial growth factor (VEGF) expression inprostatic tumours and its relationship to neuroendocrine cells.Br J Cancer 1996;74:910–916.

20. di Sant’Agnese PA. Neuroendocrine differentiation in prostaticcarcinoma: An update. Prostate 1998;8:74–79.

21. Pinthus JH, Waks T, Schindler DG, Harmelin A, Said JW,Belldegrun A, Ramon J, Eshhar Z. WISH-PC2: A uniquexenograft model of human prostatic small cell carcinoma.Cancer Res 2000;60:6563–6567.

22. Okada H, Shirakawa T, Miyake H, Gotoh A, Fujisawa M,Arakawa S, Kamidono S. Establishment of a prostatic small-cellcarcinoma cell line (SO-MI). Prostate 2003;56:231–238.

23. Fidler IJ. Critical factors in the biology of human cancermetastasis. Cancer Res 1990;50:6130–6138.

24. Liotta LA. Tumor invasion and metastases- role of theextracellular matrix. Cancer Res 1986;46:1–7.

25. Kwiatkowski DJ. Functions of gelsolin: Motility, signaling,apoptosis, cancer. Curr Opin Cell Biol 1999;11:103–108.

26. Sun HQ, Yamamoto M, Mejillano M, Yin HL. Gelsolin, amultifunctional actin regulatory protein. J Biol Chem 1999;274:33179–33182.

27. Kothakota S, Azuma T, Reinhard C, Klippel A, Tang J, Chu K,McGarry TJ, KirschnerMW,KothsK, Kwiatkowski DJ,WilliamsLT. Caspase-3-generated fragment of gelsolin: Effector ofmorphological change in apoptosis. Science 1997;278:294–298.

28. Asch HL, Head K, Dong Y, Natoli F, Winston JS, Connolly JL,Asch BB. Widespread loss of gelsolin in breast cancers ofhumans, mice, and rats. Cancer Res 1996;56:4841–4845.

29. TanakaM, Mullauer L, Ogiso Y, Fujita H, Moriya S, Furuuchi K,Harabayashi T, Shinohara N, Koyanagi T, Kuzumaki N.

Gelsolin: A candidate for suppressor of human bladder cancer.Cancer Res 1995;55:3228–3232.

30. Lee HK, Driscoll D, Asch H, Asch B, Zhang PJ. Downregulatedgelsolin expression in hyperplastic and neoplastic lesions of theprostate. Prostate 1999;40:14–19.

31. Shieh DB, Godleski J, Herndon JE 2nd, Azuma T, Mercer H,Sugarbaker DJ, Kwiatkowski DJ. Cell motility as a prognosticfactor in Stage I nonsmall cell lung carcinoma: The role ofgelsolin expression. Cancer 1999;85:47–57.

32. Thor AD, Edgerton SM, Liu S, Moore DH 2nd, Kwiatkowski DJ.Gelsolin as anegativeprognostic factor andeffector ofmotility inerbB-2-positive epidermal growth factor receptor-positivebreast cancers. Clin Cancer Res 2001;7:2415–2424.

33. Rao J, Seligson D, Visapaa H, Horvath S, Eeva M, Michel K,Pantuck A, Belldegrun A, Palotie A. Tissue microarray analysisof cytoskeletal actin-associated biomarkers gelsolin and E-cadherin in urothelial carcinoma. Cancer 2002;95:1247–1257.

34. Cunningham CC, Stossel TP, Kwiatkowski DJ. Enhancedmotility inNIH3T3fibroblasts that overexpress gelsolin. Science1991;251:1233–1236.

35. Azuma T, Witke W, Stossel TP, Hartwig JH, Kwiatkowski DJ.Gelsolin is a downstream effector of rac for fibroblast motility.EMBO J 1998;17:1362–1370.

36. De Corte V, Bruyneel E, Boucherie C, Mareel M, Vandekerc-khove J, Gettemans J. Gelsolin-induced epithelial cell invasion isdependent on Ras-Rac signaling. EMBO J 2002;21:6781–6790.

37. Nishimura K, Ting HJ, Harada Y, Tokizane T, Nonomura N,Kang HY, Chang HC, Yeh S, Miyamoto H, Shin M, Aozasa K,Okuyama A, Chang C. Modulation of androgen receptortransactivation by gelsolin: A newly identified androgenreceptor coregulator. Cancer Res 2003;63:4888–4894.

38. Burchardt T, Burchardt M, Chen MW, Cao Y, de la Taille A,ShabsighA,HayekO,Dorai T, ButtyanR. Transdifferentiationofprostate cancer cells to a neuroendocrine cell phenotype in vitroand in vivo. J Urol 1999;162:1800–1805.

NeuroendocrineCells PromoteMetastasis of AdenocarcinomaCells 545