stromal cells promote angiogenesis and growth of human ... · carcinoma cells is fundamentally...

[CANCER RESEARCH 62, 3298–3307, June 1, 2002]

Stromal Cells Promote Angiogenesis and Growth of Human Prostate Tumors in aDifferential Reactive Stroma (DRS) Xenograft Model1

Jennifer A. Tuxhorn, Stephanie J. McAlhany, Truong D. Dang, Gustavo E. Ayala, and David R. Rowley2

Departments of Molecular and Cellular Biology [J. A. T., S. J. M., T. D. D., D. R. R.] and Pathology [G. E. A.], Baylor College of Medicine, Houston, Texas 77030

ABSTRACT

Reactive stroma has been reported in many cancers, including breast,colon, and prostate. Although changes in stromal cell phenotype andextracellular matrix have been reported, specific mechanisms of howreactive stroma affects tumor progression are not understood. To addressthe role of stromal cells in differential regulation of tumor incidence,growth rate, and angiogenesis, LNCaP xenograft tumors were constructedin nude mice with five different human prostate stromal cell lines as wellas GeneSwitch-3T3 cells engineered to express lacZ under mifepristoneregulation. Alone, LNCaP prostate carcinoma cells were essentially non-tumorigenic, whereas combinations of LNCaP cells with three differenthuman prostate stromal cell lines (L/S tumors) resulted in a tumor inci-dence (50–63%) similar to that of control LNCaP plus Matrigel (L/M)tumors over a 9-week period. In contrast, LNCaP combinations with twoother human prostate stromal cell lines were nontumorigenic, illustratingthat stromal cell effects are differential. L/S tumors exhibited well-devel-oped blood vessels at 2 weeks, whereas control L/M tumors were avascularat 2 weeks and exhibited blood lakes in lieu of extensive vessels at latertime points. Xenografts constructed under three-way conditions (LNCaP,Matrigel, and stromal cells; L/M/S tumors) exhibited a 100% tumorincidence and showed rapid blood vessel formation as early as day 7 withmature vessels formed by day 10. L/M/S tumors exhibited a 10.3-foldincrease in microvessel density, and the corresponding hemoglobin:tumorweight ratio was increased 2-fold relative to L/M control tumors at day 10.L/M/S tumor wet weight and volume increased by 1.6- and 2.4-fold,respectively, by day 21, compared with control L/M tumors. L/M/S tu-mors made with LNCaP cells plus GeneSwitch-3T3-pGene/lacZ stromalcells showed similar results. Mifepristone-regulated gene expression wasobserved in stromal cells immediately adjacent to clusters of carcinomacells and in vessel walls in a mural cell (pericyte) position. This studyshows that regulation of angiogenesis is one mechanism through whichstromal cells affect LNCaP tumor incidence and growth rate. This regu-lation may be mediated through direct recruitment and interactions ofstromal cells with endothelial cells. Furthermore, this study describes forthe first time a model system with regulated transgene expression in thestromal compartment of an experimental carcinoma. These findings pointto the stromal compartment as a potential source of new prognosticmarkers and therapeutic targets and show the utility of the carcinoma-stromal xenograft model system in dissecting specific mechanisms ofreactive stroma.

INTRODUCTION

In many human cancers, the stromal microenvironment adjacent tocarcinoma cells is fundamentally different from the stroma of normaladult tissues. Studies of human breast, colon, and prostate cancerspecimens have identified activated stromal cell phenotypes, modifiedECM3 composition, and increased microvessel density in the reactive

stroma of these tumors (1–4).4 Moreover, the reactive stroma asso-ciated with these cancers exhibits biological markers consistent withstroma at sites of wound repair (5). Many of the biological processesin wound repair, including stromal cell activation to the myofibroblastphenotype, stimulation of collagen type I deposition, and induction ofangiogenesis, are observed in reactive stroma during cancer progres-sion. Because remodeling of the stromal compartment in wound repairis associated with promoting new tissue growth (formation of granu-lation tissue and re-epithelialization; Ref. 6), it follows that reactivestroma in cancer might be expected to promote tumorigenesis (7, 8).Accordingly, it is important to better understand specific mechanismsand signaling pathways in reactive stroma of cancer to exploit thiscompartment for better prognostics and potential therapeutics.

Recent studies have addressed the role of stroma in promoting tumor-igenesis using experimental carcinoma models. Elenbaas et al. (9) haveshown that combining transformed human mammary epithelium (derivedfrom normal glands) with normal human mammary fibroblasts elevatedthe incidence of tumor formation to 100% as compared with a 52%incidence in the control (epithelium only). Moreover, these studiesshowed that coinjection of normal fibroblasts with transformed mam-mary epithelium reduced the tumor latency period from 52 to 20 days.Similarly, another study showed that the coinoculation of carcinoma-associated fibroblasts with initiated, but not tumorigenic, prostate epithe-lium resulted in increased tumor incidence and growth (10). Together,these data show that when all other parameters are constant, the stromalmicroenvironment in cancer plays a critical role in determining theincidence and rate of tumorigenesis.

The specific mechanisms and pathways through which reactivestroma regulates carcinoma tumorigenesis are not fully understood. Itis known that in human prostate cancer, markers of reactive stroma areobserved quite early in tumor development. The activation of stromalcells to the myofibroblast phenotype and stimulated expression ofspecific ECM components were observed in the stroma immediatelyadjacent to precancerous PIN lesions in human surgical specimens.4

These stromal regions adjacent to PIN have also been reported toexhibit elevated microvessel density relative to normal prostatestroma (11–13). Accordingly, reactive stroma is likely to functionearly in tumorigenesis, possibly during precancerous stages.

It is clear that the ability to model reactive stroma in an animalsystem would aid in advancing our understanding of specific mech-anisms (14). Currently, there is no promoter with which to target geneexpression exclusively to the stromal compartment of a specific tissuein transgenic animals. Accordingly, xenografts of human tumors inmice remain the method of choice. Xenografts of LNCaP humanprostate carcinoma cells have been used to address tumorigenesis ofan androgen sensitive prostate carcinoma in many previous studies(15–17). The purpose of the present study was to assess the differen-tial effects of several human prostate stromal cell lines initiated by ourlaboratory on early tumorigenesis events in LNCaP xenografts in a

Received 12/21/01; accepted 3/27/02.The costs of publication of this article were defrayed in part by the payment of page

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

1 This work was supported by NIH Grants RO1-CA58093, RO1-DK45909, Special-ized Program of Research Excellence CA58204, and UO1-CA84296. S. J. M. is a HudsonScholar of the Baylor College of Medicine Medical Scientist Training Program.

2 To whom requests for reprints should be addressed, at Department of Molecular andCellular Biology, Baylor College of Medicine, Houston, TX 77030. Phone: (713) 798-6220; Fax: (713) 790-1275; E-mail: [email protected].

3 The abbreviations used are: ECM, extracellular matrix; PIN, prostatic intraepithelialneoplasia; DRS, differential reactive stroma; sm �-actin, smooth muscle �-actin; �-gal,

�-galactosidase; X-gal, 5-bromo-4-chloro-3-indoyl-�-D-galactopyranoside; DAB, diami-nobenzidine tetrahydrochloride; L/M, LNCaP/Matrigel (two-way xenografts); L/S,LNCaP/stromal cell (two-way xenografts); L/M/S, LNCaP/Matrigel/stromal cell (three-way xenografts).

4 J. A. Tuxhorn, G. E. Ayala, V. C. Smith, T. D. Dang, and D. R. Rowley. Reactivestroma in human prostate cancer: induction of myofibroblast phenotype and ECM remod-eling, submitted for publication.

3298

Research. on March 3, 2020. © 2002 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

nude mouse model system. Specifically, tumor incidence, tumorgrowth, and angiogenesis were evaluated under different xenograftconditions. We report here that LNCaP tumor incidence and rate oftumor growth was dependent on the stromal cell conditions used, andthat tumor-promoting stromal cells were associated with early andelevated angiogenesis. Furthermore, these studies show the develop-ment of a modular DRS LNCaP xenograft model system that exhibitsa 100% tumor incidence, tumor homogeneity, early angiogenesis, andregulated transgene expression in the stromal cell population of thetumor.

MATERIALS AND METHODS

Cell Lines. LNCaP human prostate carcinoma cells were purchased fromAmerican Type Culture Collection (Manassas, VA) and cultured in RPMI1640 (Life Technologies, Inc., Rockville, MD) supplemented with 10% fetalbovine serum (Hyclone, Logan, UT), 100 units/ml penicillin, and 100 �g/mlstreptomycin (Sigma Chemical Co., St. Louis, MO).

Human prostate stromal cell lines were established according to our proto-col published previously (18, 19). Briefly, fresh tissue cores were obtainedfrom the Baylor College of Medicine Prostate Specialized Program of Re-search Excellence Pathology and Tissue Microarray Core. Tissues originatedfrom organ donors or patients undergoing radical prostatectomy. Unless noted,tissue cores were isolated from regions of normal prostate, based on his-topathological criteria. The cores were diced into 1-mm cubes, washed withHBSS, and put into 24-well tissue culture plates containing Bfs medium:DMEM (Life Technologies, Inc., Rockville, MD) supplemented with 5% fetalbovine serum, 5% Nu Serum (Collaborative Research, Bedford, MA), 0.5�g/ml testosterone, 5 �g/ml insulin, 100 units/ml penicillin, and 100 �g/mlstreptomycin (Sigma, St. Louis, MO). The explants were incubated at 37°Cwith 5% CO2, and culture medium was changed every 48 h. Stromal cellsmigrated out of the tissue and attached to the culture dish. After the cellsreached confluence, the remaining tissue was removed, and the cells werepassaged using standard trypsin/EDTA procedures. The following stromal celllines were generated: HTS-40C (transition zone of a 40-year-old), HPS-40F(peripheral zone of a 40-year-old), HPS-44D (peripheral zone of a 44-year-old), HTS-2T (transition zone of a 16-year-old organ donor), and HPS-TZ1A(transition zone of benign prostatic hyperplasia specimen). Standard immuno-cytochemistry procedures were used to evaluate cell phenotype. Cytokeratinexpression was not detected with the AE1/AE3 pooled monoclonal antibody(Boehringer Mannheim, Indianapolis, IN), and all cell lines were 100% vi-mentin positive (sc-7557; Santa Cruz Biotechnology, Santa Cruz, CA). Ex-pression of sm �-actin (clone 1A4; Sigma) varied among stromal cell lines,from no expression (HPS-TZ1A) to approximately 50% of the population(HTS-40C) to nearly 100% (HTS-2T). Cultures at passages 5–15 were used forexperiments.

The GeneSwitch System (Invitrogen, Carlsbad, CA), a mifepristone-regulated expression system for mammalian cells, was used to create a stromalcell line with regulated expression of the lacZ gene. The GeneSwitch-3T3fibroblast line, which expresses the GeneSwitch regulatory protein from thepSwitch vector, was cultured in DMEM supplemented with 10% fetal bovineserum and 50 �g/ml hygromycin B (Invitrogen, Carlsbad, CA). This cell linewas transfected with the pGENE/V5-His/lacZ vector (Invitrogen) usingFuGENE 6 transfection reagent according to manufacturer’s recommendations(Roche, Indianapolis, IN). Stably transfected cells were selected with 300�g/ml Zeocin (Invitrogen). Zeocin-resistant clones were treated with 100 pM

mifepristone (Sigma) and screened for �-gal expression using standard in vitroX-gal staining procedures. The GeneSwitch-3T3-pGene/lacZ cell line used inthis study represents a pooled population of stably transfected cells.

Animals. Athymic NCr-nu/nu male homozygous nude mice, 6–8 weeks ofage, were purchased from Charles River Laboratories (Wilmington, MA). Allexperiments were conducted in accordance with the NIH Guide for the Careand Use of Laboratory Animals.

Preparation of LNCaP Xenografts. Xenograft tumors were generatedunder different stromal conditions: one-way xenografts (LNCaP cells, stromalcells, or Matrigel ECM only); two-way xenografts (LNCaP � Matrigel,LNCaP � stromal cells, or stromal cells � Matrigel); and three-way xe-nografts (LNCaP cells � stromal cells � Matrigel). LNCaP cells and stromal

cells were harvested from culture plates by trypsinization, resuspended ingrowth medium, and counted with a hemacytometer. Cells were used imme-diately or frozen in aliquots for modular xenograft experiments. For frozenaliquots, cells were pelleted at �950 � g for 50 s in a clinical centrifuge (IECmodel CL with rotor 809, setting 5), resuspended in filtered freezing medium(culture medium � 10% DMSO), and transferred to cryogenic vials (CorningInc., Corning, NY). Cells were frozen using a StrataCooler Cryo preservationmodule (Stratagene, La Jolla, CA) and stored at �80°C. Matrigel ECM(Becton Dickinson, Bedford, MA) was stored in 0.5-ml aliquots at �20°C andthen thawed on ice at 4°C for 3 h before use.

For modular three-way xenograft experiments, frozen aliquots of cells(16 � 106 LNCaP cells and 4 � 106 stromal cells) were thawed in a 37°Cwater bath for 1–2 min, transferred to separate 15-ml conical tubes, washedwith 10 ml of culture medium, and pelleted as described above. LNCaP cellswere resuspended in 10 ml of culture medium, whereas stromal cells wereresuspended in 3 ml of culture medium. Then, cells were combined in one tubeand pelleted again. The supernatant was aspirated to the 300-�l mark, and cellswere gently resuspended in the remaining medium. Before combining withMatrigel, cells were incubated on ice for 5 min. A prechilled pipette was usedto transfer cells to the tube containing 0.5 ml of Matrigel (on ice). Cells weremixed with the Matrigel and then drawn into a prechilled pipette 1-ml syringewith a 20-gauge needle. Using a 25-gauge needle, 100 �l of cell suspension(2 � 106 LNCaP cells and 0.5 � 106 stromal cells, 4:1 ratio) were injected s.c.in each lateral flank. Three animals (6 total injection sites) were used for eachexperiment set.

For LNCaP/stromal cell two-way xenograft experiments, 16 � 106 LNCaPcells and 8 � 106 stromal cells were washed, combined, and resuspended asdescribed above. The cells were mixed with 500 �l of PBS instead of Matrigel,and 100 �l of cell suspension (2 � 106 LNCaP cells and 1 � 106 stromal cells,2:1 ratio) were injected s.c. in each lateral flank. For LNCaP/Matrigel two-wayxenograft experiments, 16 � 106 LNCaP cells were washed twice with 10 mlof culture medium, resuspended in 300 �l of medium, incubated on ice, mixedwith 0.5 ml of Matrigel, and injected s.c., as described above. For one-wayxenograft experiments, 16 � 106 LNCaP cells or 8 � 106 stromal cells werewashed, resuspended in 300 �l of medium, mixed with 500 �l of PBS, andinjected s.c. as described above.

To induce expression of the lacZ reporter gene in GeneSwitch-3T3-pGene/lacZ cells, animals received mifepristone (RU 486; Sigma) at 0.3 mg/kgadministered as 100-�l i.p. injections. Animal dosage of mifepristone wasbased on protocols shown previously to induce gene expression in vivo (20,21). Treatment began at the time of experimental setup and was repeated every48 h. Mifepristone was dissolved in ethanol (2 mg/ml) and then diluted insesame oil (Sigma) to the appropriate concentration. For vehicle control,animals were treated with ethanol diluted in sesame oil.

Processing of Tumors. Animals were monitored daily, and tumors wereharvested at time points ranging from 4 days to 9 weeks, depending on theparameter being evaluated. Tumor volume was not allowed to exceed �150mm3. Tumors visible with the naked eye were surgically removed and dis-sected away from surrounding skin and fascia. Tumors were measured in threedimensions with calipers, and tumor volume (mm3) was calculated with theformula V � 0.52 � length x width x height (22). Tumors were then trans-ferred to preweighed tubes on ice. Tubes were reweighed, and tumor weight(g) was determined. Tissues were fixed in 4% paraformaldehyde overnight at4°C and then washed 3� with PBS and processed for histology. Tissues wereembedded in paraffin, and 5-�m sections were cut and mounted onto ProbeOnPlus slides (Fisher Scientific, Pittsburgh, PA). Sections were stained with H&Eand Masson’s Trichrome (Sigma) for histological analysis.

Immunohistochemistry and in Situ Hybridization. Antibodies usedwere: sm �-actin, mouse monoclonal 1A4 (Sigma); antimouse CD31/PECAM-1, rat monoclonal MEC13.3 (BD PharMingen, San Diego, CA);biotin-conjugated Universal Secondary (Research Genetics, Huntsville, AL)used for sm-�-actin; and biotin-conjugated goat antirat IgG (BD PharMingen)used for CD31. Specificity of each primary antibody has been evaluated andpublished previously. No significant staining was observed if sections wereincubated with secondary antibody only.

Immunostaining was performed with the MicroProbe Staining System(FisherBiotech, Pittsburgh, PA). Reagents formulated for use with capillaryaction systems were purchased from Research Genetics (Huntsville, AL) andused according to the manufacturer’s protocol. Tissues were deparaffinized

3299

REACTIVE STROMA AND PROSTATE TUMORIGENESIS



using Auto Dewaxer and cleared with Auto Alcohol. Brigati’s Iodine and AutoPrep were used to improve tissue antigenicity. Then, sections were washedwith Universal Buffer. For sm �-actin staining, sections were incubated in 20�g/ml goat-antimouse Fab fragment (Jackson ImmunoResearch, West Grove,PA) for 30 min at 37°C to block host immunoglobulins. For CD31 staining,antigen retrieval was required; tissues were incubated in 0.1% trypsin (Zymed,South San Francisco, CA) for 10 min at 37°C. After these treatments, sectionswere washed with Universal Buffer, incubated in Protein Blocker, and washedagain. Antibodies were diluted in Primary Antibody Diluent and used at thefollowing conditions: sm �-actin, 1:200 for 8 min at 50°C; and CD31, 1:50overnight at 4°C. Tissues were washed with Universal Buffer and incubated insecondary antibody: Universal Secondary undiluted for 4 min at 50°C; andantirat IgG 1:100 for 45 min at 37°C. Tissues were washed in Universal Buffer,treated with Auto Blocker to inhibit endogenous peroxidase activity, andwashed again. For detection, sections were incubated in streptavidin horserad-ish peroxidase (sm �-actin) or RTU VectaStain Elite ABC reagent (VectorLaboratories, Burlingame, CA; CD31), washed in Universal Buffer and thenincubated in stable diaminobenzidine tetrahydrochloride two times for 3 mineach at 50°C. Tissues were counterstained with Auto Hematoxylin for 30 s.

In situ hybridization to specifically label human cells was performed withthe MicroProbe Staining System according to a procedure published previ-ously (23). All reagents were purchased from Research Genetics and usedaccording to the manufacturer’s protocol. Tissue sections were deparaffinizedin xylene and then rehydrated in 100, 95, and 70% ethanol, followed bydouble-distilled H2O. Slides were incubated in Auto Blocker for 1 min at roomtemperature and washed three times with Universal Buffer. Tissues weretreated with pepsin for 1 min at 105°C, and then Alu1/Alu2 probe was added(prediluted and used at 1 �g/ml). Tissues were incubated in Alu1/Alu2 probefor 5 min at 105°C and then at 45°C for 1.5 h. Sections were washed with PostHybe Wash for 5 min at 45°C and incubated in streptavidin horseradishperoxidase for 10 min at 50°C. After incubation in Probe Lock for 10 s at roomtemperature, tissues were treated with stable diaminobenzidine tetrahydrochlo-ride, rinsed with Auto Wash, and counterstained with Auto Hematoxylin asdescribed above.

Microvessel Density Analysis. Analysis was performed according tostandard procedures (24). Tissue sections were stained for CD31 as describedabove. Sections were scanned at �200, and areas with the highest vasculardensity were identified. Vessels in three high-power fields (�400) werecounted by two independent observers, one of whom was blinded to experi-mental conditions. The average vessel count was determined for each speci-men. Six tumors from each condition were analyzed, and counts were com-bined for statistical analysis by unpaired t test.

Hemoglobin Assay. Hemoglobin content was measured using Drabkin’smethod (25). Whole tumors were homogenized in 50 �l of double-distilledH2O using disposable pellet pestles for microtubes (Fisher Scientific). Homo-genates were incubated in 0.5 ml of Drabkin’s Solution (Sigma) for 15 min atroom temperature. Samples were centrifuged to pellet cell debris. The super-natants were transferred to cuvettes, and the absorbance was measured at 540nm. Drabkin’s Solution was used as a blank. The absorption, which is pro-portional to the total hemoglobin concentration, was divided by tumor weight.Six tumors from each condition were assayed, and measurements were com-bined for statistical analysis by unpaired t test.

Whole-Mount �-gal Staining. Whole tumors were stained according tostandard procedures. Briefly, tumors were harvested at day 10 and fixed with2% paraformaldehyde/0.2% glutaraldehyde (Polysciences, Inc., Warrington,PA) in PBS for 30 min at 4°C on a platform rocker. Tissues were washed threetimes for 30 min each at room temperature with rocking in �-gal wash buffer[0.1 M sodium phosphate buffer (pH 7.4), 2 mM MgCl2, 0.01% sodiumdeoxycholate, and 0.02% NP40]. Then, tumors were incubated at 37°C for 3 hin �-gal staining solution [�-gal wash buffer � 5 mM potassium ferrocyanide,5 mM potassium ferricyanide, and 1 mg/ml X-gal (Promega, Madison, WI)].Tissues were washed three times for 30 min each in PBS at room temperaturewith rocking. Tumors were postfixed in 4% paraformaldehyde overnight at4°C and washed three times with PBS. Whole tumors were photographed witha Leica GZ6 stereomicroscope, and then tissues were embedded in paraffin andsectioned. Thin sections were counterstained with nuclear fast red.

Statistical Analysis. Incidence of tumor formation between two experi-mental conditions was analyzed by Fisher’s exact test, whereas multipleconditions were compared with the �2 test. Tumor weight, tumor volume,

microvessel density, and hemoglobin content were each evaluated with anunpaired t test. Analysis was performed with GraphPad Prism for Macintoshversion 3.0 (GraphPad Software, San Diego, CA). P � 0.05 was consideredstatistically significant.

RESULTS

Stromal Cells Differentially Promote LNCaP Tumorigenesis.To investigate the role of the stromal microenvironment in prostatecancer, we examined the effect of human prostate stromal cell lines onLNCaP tumorigenesis under several xenograft conditions. LNCaPhuman prostate carcinoma cells were combined with stromal compo-nents and then injected s.c. in athymic nude mice. Stromal compo-nents included different human prostate stromal cell lines and stromalcell-derived ECM (Matrigel; Refs. 26 and 27). For one-way controlxenografts, LNCaP cells or human prostate stromal cells were indi-vidually injected with buffer. For two-way xenografts, LNCaP cellswere injected with either Matrigel as a control or with prostate stromalcells in buffer. For three-way xenografts, LNCaP cells were injectedwith both Matrigel and prostate stromal cells in buffer. Five differenthuman prostate stromal cell lines were evaluated. As described in“Materials and Methods,” the human prostate stromal cell lines werecytokeratin negative and vimentin positive, with varied expression ofsm �-actin. Thus, the stromal cell lines represented a mix of fibro-blasts and myofibroblasts, which are the cell phenotypes observed inhuman prostate cancer reactive stroma.4

For tumor incidence studies, the growth period ranged from 2 to 9weeks, and tumors were harvested at different time points for histo-logical analysis. A tumor was defined as a palpable mass that con-tained nodules of LNCaP carcinoma cells upon histological examina-tion. Inoculation of nude mice with LNCaP cells alone (one-waycontrol xenografts) resulted in a low incidence of tumor formation(8%; Fig. 1, Column a), which is consistent with previous reports (28,29). After 9 weeks, the human prostate stromal cell lines did not formpalpable masses when injected alone (Fig. 1, Columns b–d) or wheninjected with Matrigel (data not shown). Thus, under one-way xe-nograft conditions, the LNCaP carcinoma cells were only weaklytumorigenic, whereas the human prostate stromal cells were nontu-morigenic.

Similar to previous studies that established standard LNCaP xe-nograft procedures (15, 30), coinjection of LNCaP cells with Matrigel(Fig. 1, Column e) increased tumor incidence to 63% (P � 0.001,Fisher’s exact test). Throughout this report, standard LNCaP/Matrigelxenografts (L/M) were used as a positive control by which to examinethe effect of stromal cells on LNCaP tumorigenesis. Coinjection of

Fig. 1. Stromal conditions regulate incidence of tumor formation in LNCaP xenografts.Columns: a, LNCaP only (n � 24); b, HTS-40C only (n � 5); c, HPS-40F only (n � 5);d, HTS-2T only (n � 5); e, LNCaP � Matrigel (n � 24); f, LNCaP � HTS-40C (n � 24);g, LNCaP � HPS-40F (n � 12); h, LNCaP � HPS-TZ1A (n � 12); i, LNCaP � HTS-2T(n � 12); j, LNCaP � HPS-44D (n � 6); k, LNCaP � Matrigel � HTS-40C (n � 16);l, LNCaP � Matrigel � HTS-2T (n � 24); m, LNCaP � Matrigel � GeneSwitch-3T3-pGene (n � 6); n, LNCaP � Matrigel � GeneSwitch-3T3-pGene/lacZ (n � 18).

3300




human prostate stromal cells with LNCaP cells in two-way xenograftexperiments (L/S) differentially regulated the incidence of LNCaPtumor formation. Three independent prostate stromal cell lines (HTS-40C, HPS-40F, and HPS-TZ1A) promoted growth of LNCaP tumorsin at least 50% of injection sites (Fig. 1, Columns f–h), whereas twoother stromal cell lines (HTS-2T and HPS-44D) did not supportLNCaP tumorigenesis (0% tumor incidence at 9 weeks after inocula-tion; Fig. 1, Columns i–j). Compared with injection of LNCaP cellsalone, the increase in tumor incidence attributable to coinjection ofLNCaP cells with HTS-40C, HPS-40F, or HPS-TZ1A stromal cellswas statistically significant (P � 0.003, 0.009, 0.003, respectively,Fisher’s exact test). Moreover, there was no significant difference inthe incidence of LNCaP tumor formation with these stromal cell linesas compared with Matrigel. In addition, there were no obvious dif-

ferences in the rate of tumor growth between L/M and tumorigenicL/S xenografts.

Combining LNCaP cells with both Matrigel and human prostatestromal cells (three-way L/M/S xenografts) further elevated LNCaPtumor incidence to 100% (Fig. 1, Columns k–l). Relative to tumorincidence in two-way xenografts, the increased tumor incidence inthree-way conditions was statistically significant (P � 0.001, �2 test).Interestingly, HTS-2T normal human prostate stromal cells promotedLNCaP tumor growth in three-way xenografts but not in two-wayxenografts (P � 0.001, Fisher’s exact test). These data show that thestromal microenvironment, including both the stromal cells and theECM, is a critical regulator of LNCaP tumorigenesis.

Analysis of Two-way Xenograft Tumors. Histological examina-tion revealed dramatic differences in the morphology of control 2-way

Fig. 2. Histological analysis and CD31 immu-nostaining of two-way L/M and L/S xenograft tu-mors. A, control L/M tumor harvested at 2 weeksand stained with Masson’s Trichrome. At this timepoint, L/M tumors contained small LNCaP clustersembedded in Matrigel (light blue areas). Asterisk,invasion of mouse host stromal cells at the tumorperiphery. B, CD31 immunostaining of controlL/M tumors detected few (if any) vessels at earlytime points (2 weeks). C, 2-week L/S (HTS-40C)tumor stained with Masson’s Trichrome. Early L/Stumors consisted of large nodules of LNCaP cellssurrounded by a collagen-rich stroma (blue areas).Similar results were observed in L/S tumors gen-erated with HPS-40F and HPS-TZ1A cells. D,CD31-positive vessels were observed throughoutthe stromal compartment of 2-week L/S tumors. E,late-stage (6 weeks) L/M control tumors stainedwith H&E contained sheets of LNCaP cells butwere hemorrhagic. Asterisk, pools of RBCs, alsoknown as “blood lakes.” F, CD31 immunostainingof L/M tumors at 6 weeks. Blood lakes (asterisk) inlate-stage L/M tumors were not lined by CD31-positive cells, although a few CD31-positive ves-sels were observed (arrow). G, late-stage (6 weeks)L/S tumors stained with H&E resembled poorlydifferentiated carcinoma, with little stroma remain-ing. Blood vessels (arrows) were present in thinstrips of tumor stroma; blood lakes were not ob-served. H, sm �-actin immunostaining of L/S tu-mors at 6 weeks. Numerous blood vessels weredetected in late L/S tumors. Vessels were CD31positive (not shown) and were surrounded by sm�-actin-positive mural cells. Bar, 50 �m.

3301




LNCaP/Matrigel (L/M) tumors as compared with two-way LNCaP/stromal cell (L/S) tumors. At early time points (2 weeks), control L/Mtumors consisted of small clusters of LNCaP cells embedded inMatrigel (Fig. 2A). Invasion of mouse host stromal cells into theMatrigel was also observed (Fig. 2A, asterisk). Similar to observationsby Gao et al. (31), the stromal cells did not form a collar of smoothmuscle around the clusters of carcinoma cells. In contrast to L/Mtumors, the histological appearance of early L/S tumors was moretypical of the epithelial-stromal architecture observed in human pros-tate carcinoma. Masson’s trichrome staining showed that nodules ofLNCaP cells were surrounded by a stroma rich in collagen fibers andblood vessels (Fig. 2C). Thus, the stroma in two-way L/S tumorsexhibited features more consistent with human prostate cancer reac-tive stroma, which we have described previously (4).4 Additionally,the LNCaP nodules were more extensive in two-way L/S tumorsrelative to two-way control L/M tumors. These data suggest thathuman prostate stromal cells may confer a growth/survival advantageto LNCaP cells and thus enhance LNCaP tumorigenesis.

At later time points (6 weeks), both control L/M tumors and L/Stumors resembled poorly differentiated carcinomas composed ofsheets of LNCaP cells with little intervening stroma (Fig. 2, E and G).However, although connective tissue septa in L/S tumors containedblood vessels (Fig. 2G, arrows), control L/M tumors appeared to lacknormal vasculature. Rather, the L/M tumors were hemorrhagic, andlarge pools of RBCs, previously termed “blood lakes” (32), wereobserved frequently (Fig. 2E, asterisk).

To examine the vasculature in two-way xenograft tumors, weperformed immunohistochemistry with an antibody to CD31 (PE-CAM-1), an endothelial cell marker. CD31 immunostaining wasrarely detected in control L/M tumors at 2 weeks (Fig. 2B), which isconsistent with the lack of blood vessels observed by H&E or Mas-son’s Trichrome stain. In contrast, numerous CD31-positive vesselswere present in the stromal compartment of early (2 weeks) L/Stumors (Fig. 2D). Moreover, sm �-actin-positive mural cells (peri-cytes or smooth muscle cells) surrounded these blood vessels, indi-cating that vessels were mature (data not shown; Ref. 33). Althoughsome CD31-positive vessels were observed in late-stage L/M tumors(Fig. 2F, arrow), most RBCs appeared to be in blood lakes, whichwere CD31 negative (Fig. 2F, asterisk). Late-stage L/S tumors werewell vascularized, and vessels were positive for both CD31 (data notshown) and sm �-actin (Fig. 2H). Analysis of two-way xenografttumors suggests that stromal cells influence LNCaP tumor develop-ment and vascularization, particularly during early tumorigenesis.

Analysis of Three-Way Xenograft Tumors. Comparison of two-way control L/M tumors with three-way L/M/S tumors allowed amore direct evaluation of the role of stromal cells in LNCaP tumor-igenesis. As shown above, coinjection of stromal cells with LNCaPcells and Matrigel significantly increased tumor incidence relative totwo-way conditions (Fig. 1). Moreover, three-way L/M/S tumorsexhibited a 47% increase in wet weight compared with two-way L/Mtumors at day 10 (Fig. 3A; P � 0.009, unpaired t test). A significantdifference in tumor volume was not observed at day 10 (Fig. 3B);however, the tumors appeared very different. Control L/M tumors atday 10 were translucent, whereas three-way L/M/S tumors wereopaque with blood vessels visible on the tumor surface (Fig. 4A).Histological examination of three-way tumors at day 10 revealed largeclusters of LNCaP cells with adjacent stromal cells and numerousblood vessels embedded in Matrigel (Fig. 4, B and C). In contrast,L/M tumors consisted of small clusters of LNCaP cells with littlevascularization, similar to tumors harvested at 2 weeks (Fig. 2, A andB). In situ hybridization with a human-specific probe showed thatthree-way L/M/S tumor stroma was composed of human prostatestromal cells (Fig. 4D, arrows) mixed with mouse host stromal cells.

The stromal cells were frequently adjacent to clusters of LNCaP cells,similar to the histological architecture of human prostate cancer.These stromal cells were sm �-actin positive, which is consistent withthe myofibroblast phenotype (Fig. 5B, arrowheads). In addition, stain-ing with Masson’s trichrome showed that stromal cells adjacent tocarcinoma cells and blood vessels were producing collagen matrix(Fig. 4C, arrow).

By day 21, three-way L/M/S tumors consisted of sheets of LNCaPcells with intervening stroma, typical of poorly differentiated humanprostatic carcinoma (Fig. 4E). The wet weight (Fig. 3A) of three-wayL/M/S tumors was significantly higher than two-way control L/Mtumors at the day 21 time point (64% increase; P � 0.001, unpairedt test). Moreover, at this same time point, the three-way tumorsexhibited a 2.4-fold increase in tumor volume compared with two-wayL/M tumors (Fig. 3B; P � 0.001, unpaired t test). Histologicalanalysis of L/M tumors at day 21 revealed considerable heterogeneity.Sheets or large clusters of LNCaP cells were observed in some regions(Fig. 4G), whereas other areas of the two-way L/M tumors exhibitedsmall clusters of LNCaP cells, similar to day 10 (Fig. 4F). It was notedthat regions with increased LNCaP cell density contained bloodvessels (Fig. 4G, arrow), although some blood lakes had also formed(Fig. 4G, asterisk). Neither the average wet weight (Fig. 3A) nor thevolume (Fig. 3B) of two-way control L/M tumors exhibited statisti-cally significant differences between days 10 and 21 (P � 0.2 and 0.8,respectively, unpaired t test). Together, these data show that stromalcells increase LNCaP tumor incidence and accelerate the rate ofLNCaP tumor growth.

Angiogenesis in Three-Way L/M/S Tumors. To investigate thevasculature in early three-way L/M/S tumors, we examined tissuesharvested 4, 7, and 10 days after injection. Blood vessels were notobserved in H&E-stained tissue sections at day 4. However, vesselswere typically observed in day 7 tumors, and tumors were well

Fig. 3. Stromal cells in three-way L/M/S xenografts promote LNCaP tumorigenesis. A,mean tumor weight of control L/M (n � 6) and three-way L/M/S (n � 6) tumors at day10; mean tumor weight of L/M (n � 14) and L/M/S (n � 14) DRS tumors at day 21. Bars,SE. B, mean tumor volume of control L/M (n � 4) and three-way L/M/S (n � 4) tumorsat day 10; mean tumor volume of L/M (n � 14) and L/M/S (n � 10) tumors at day 21.Asterisk, statistically significant increase in weight or volume of three-way L/M/S tumorscompared with two-way L/M tumors at the same time point (P � 0.01). All three-waytumors were generated with HTS-2T cells. Bars, SE.

3302




vascularized by day 10 with no evidence of blood lakes. Vessels werelined with CD31-positive endothelial cells (Fig. 5A) and surroundedby sm �-actin-positive mural cells (Fig. 5B, arrows). Thus, vascular-ization of early three-way L/M/S tumors was attributable to thedevelopment of mature blood vessels, not blood lakes.

Microvessel density and hemoglobin content were measured to

quantitate the vascularity in control L/M tumors and in L/M/S tumors.Microvessel density analysis was performed on CD31-stained tissuesections. The average microvessel count per �400 field in three-waytumors at day 10 was 31 � 5 (Fig. 6A, Column b). In contrast, theaverage microvessel count of early two-way control L/M tumors was3 � 2 (Fig. 6A, Column a). The 10.3-fold increase in vascularization

Fig. 4. Analysis of three-way L/M/S xenograft tumors. A, gross morphology of two-way L/M and three-way L/M/S tumors harvested at 10 days. Bar, 1 mm. B, H&E-stained tissuesections from L/M/S tumors harvested at day 10 showed large clusters of LNCaP cells with adjacent stromal cells and blood vessels. Similar results were observed with HTS-40C andHTS-2T human prostate stromal cells. C, L/M/S tumor (day 10) stained with Masson’s Trichrome. Stromal cells adjacent to LNCaP clusters and blood vessels appeared to secrete newcollagen (dark blue; arrow) within the Matrigel ECM (light blue). D, in situ hybridization of three-way L/M/S tumors (day 10) with a human-specific probe (Alu1/Alu2). LNCaP cellsand human stromal cells (arrows) were labeled with this technique. Note that human stromal cells were typically adjacent to LNCaP clusters. Mouse host stromal cells were alsoobserved (blue counterstain). E, H&E-stained three-way L/M/S tumors at day 21 exhibited large clusters or sheets of LNCaP cells with some intervening stroma, similar to poorlydifferentiated carcinoma. Blood vessels (arrows), but not blood lakes, were observed in the tumor stroma. F and G, H&E-stained tissue sections from two-way L/M tumors at day 21were heterogeneous. Some areas resembled day 10 tumors (F), whereas other regions consisted of large clusters or sheets of LNCaP cells (G). Regions with increased LNCaP cell density(G) contained blood vessels (arrow) and blood lakes (asterisk). Bar (B–G), 50 �m.

3303




of three-way L/M/S tumors was statistically significant (P � 0.001,unpaired t test). In addition, two-way L/M tumors and three-way DRStumors harvested at day 10 were assayed for hemoglobin content byDrabkin’s method. The concentration of hemoglobin, normalized totumor weight, in three-way L/M/S tumors was 2-fold higher thanin control L/M tumors (Fig. 6B; P � 0.004, unpaired t test). Thesedata show that the accelerated LNCaP tumorigenesis observed inthree-way xenografts with stromal cells, relative to two-way xe-nografts with Matrigel only, correlates with elevated microvesseldensity and hemoglobin content in early tumors. Moreover, thesedata suggest that stromal cells enhance LNCaP tumorigenesis bypromoting angiogenesis.

Regulated Gene Expression in Stromal Cells of Three-WayL/M/S Tumors. To assess the utility of the three-way L/M/S modelfor regulated stromal gene expression and to determine the localiza-tion of stromal cells relative to carcinoma cells and developing bloodvessels, we used mouse 3T3 fibroblasts stably transfected with thepSwitch and pGene/lacZ vectors. The incidence of tumor formationwhen LNCaP cells were coinjected with Matrigel and Geneswitch-3T3-pGene/lacZ cells was 100% (Fig. 1, Column n). In addition,tumor size, morphology, and vascularity at day 10 were similar tothree-way tumors generated with LNCaP cells and human prostatestromal cells. Animals were treated with vehicle control or withmifepristone (RU 486) to induce expression of the lacZ reporter gene.

No statistically significant differences in weight or volume wereobserved between vehicle control and mifepristone-treated tumors(data not shown). Tumors harvested from animals treated with vehiclecontrol did not exhibit detectable �-gal activity in whole tissues (Fig.7A) or in tissue sections (Fig. 7C). In contrast, mifepristone inducedexpression of the lacZ gene in LNCaP/3T3-pGene/lacZ tumors, asshown by the high level of X-gal staining (Fig. 7B). Examination oftissue sections revealed that �-gal activity was restricted to stromalcells (Fig. 7, D and E). Moreover, positive staining stromal cells werefrequently adjacent to clusters of LNCaP cells (Fig. 7D, arrows),which is consistent with the tissue architecture of other three-wayL/M/S tumors (Fig. 4D). Significantly, stromal cells expressing �-galwere also observed surrounding blood vessels in a mural cell (peri-cyte) position in the vessel wall (Fig. 7E, arrows). To our knowledge,this is the first study to show regulated stromal cell gene expression inan in vivo cancer model. Furthermore, these data suggest a directinteraction between inoculated stromal cells and the developing vas-culature.

DISCUSSION

Studies of human breast, colon, and prostate cancers suggest that astromal response occurs as a component of carcinoma progression(1–4).4 This reactive stroma exhibits activated stromal cell pheno-types and expresses different ECM components as compared withnormal tissue stroma. The specific mechanisms of the stromal re-sponse and how reactive stroma affects cancer progression are un-known. However, because reactive stroma exhibits characteristicstypified by wound repair stroma, it has been suggested that cancer-associated stroma is tumor promoting (4, 5, 7). Data presented in thisstudy confirm that the stromal microenvironment promotes and dif-ferentially regulates tumorigenesis of a human prostate cancer xe-nograft. Furthermore, our data suggest that one mechanism throughwhich stromal cells affect the incidence and rate of tumorigenesis isby stimulating early angiogenesis.

Our data demonstrate that the stromal effect on LNCaP tumorigen-esis is differential, in that not all human prostate stromal cell lineswere tumor promoting in two-way combinations. Moreover, undertwo-way xenograft conditions, the incidence of tumor formation didnot exceed 63%. The results of the three-way experiments show thata more complete stromal microenvironment was required to reachmaximal tumor incidence, accelerated tumor growth, and more ho-

Fig. 5. Vasculature in three-way L/M/S xenograft tumors at day 10. A, CD31-immunostaining of early three-way tumors revealed numerous blood vessels distributed throughoutthe tumor stroma. B, sm �-actin immunostaining showed that CD31-positive vessels were surrounded by mural cells (arrows) and, therefore, were mature. Stromal cells adjacent toLNCaP clusters were also sm �-actin positive (arrowheads), consistent with the myofibroblast phenotype. Bar, 50 �m.

Fig. 6. Stromal cells in three-way L/M/S tumors promote angiogenesis. A, microvesseldensity analysis of two-way L/M (n � 6) and three-way L/M/S (n � 6) tumors at day 10.Three high power fields (�400) of the highest vascular density were examined for eachtumor, and all counts were combined for statistical analysis. Bars, SE. B, mean hemo-globin content of whole tumors at day 10 (L/M, n � 6; L/M/S, n � 6). Asterisk,statistically significant increase (P � 0.01). All three-way tumors were generated withHTS-2T cells. Bars, SE.

3304




mogeneous tumors. The three-way combinations represented a morecomplete tumor model in that all three critical components are sup-plied: tumor cells, stromal cells, and a stromal cell-derived matrix. Ofmost significance, this three-way mix results in a tumor with stimu-lated angiogenesis during early tumor formation.

It is clear that the appropriate rate and extent of angiogenesis is acritical component of cancer progression (34–36). The mechanismsthrough which stromal cells promote angiogenesis in our model arenot yet understood; however, stromal cells may affect angiogenesis atseveral levels. Stromal cells may stimulate host endothelial cell mi-gration into the xenograft as well as promote the subsequent formationof endothelial tubes. Indeed, growth factors such as vascular endo-thelial growth factor and fibroblast growth factor-2, known to beimportant for endothelial migration and tube formation, are secretedby stromal cells (37, 38). Moreover, deposition of specific ECMcomponents by stromal cells may regulate endothelial cell response toangiogenic factors (39). It has also been shown that stromal cellsdirectly promote vessel maturation by cell-cell interactions. Endothe-lial cells recruit adjacent stromal cells as mural cells destined tobecome pericytes or smooth muscle cells (33). This endothelial cell-mural cell interaction functions to stabilize the developing vessel (33).It has been shown that nascent endothelial cell tubes are unstable andregress in the absence of perivascular stromal cells if vascular endo-

thelial growth factor is withdrawn (40). Accordingly, in addition tostimulating the process of angiogenesis, stromal cells are an integralcomponent of the mature vessel wall and are required for vesselfunction. Indeed, our data with lacZ-positive stromal cells show aclose association of these cells with endothelial tubes, in a mural cellposition. It is likely that the human stromal cells are also incorporatedinto vessel walls, although they were not readily observed in tissuesections stained with a human-specific probe. The chromosome in situhybridization technique does not label stromal cells with the sameintensity as epithelial cells, and stromal cells in vessel walls areattenuated in cross section. Thus, chromosome in situ hybridizationmay not be sufficient to identify human stromal cells in the vesselwall.

Nonstromal L/M control tumors at early time points showed essen-tially no vessel development and exhibited blood lakes at later timepoints. The blood lakes were not lined by CD31-positive endothelialcells or sm �-actin-positive mural cells. This is consistent with datareported by Wilson and Sinha (41), where LNCaP-Matrigel tumorsexhibited “large dilated blood vessels” that were negative for vonWillebrand factor, an endothelial marker. Our results suggest that thesubsequent formation of blood lakes in two-way L/M tumors func-tions to compensate for lack of adequate vascularization attributableto the absence of stromal cells in early tumors. In contrast, our

Fig. 7. Regulated expression of lacZ in stromal cells of three-way L/M/S tumors. A and B, tumors generated with LNCaP cells, Matrigel, and GeneSwitch-3T3-pGene/lacZ cellswere assayed for �-gal activity by whole-mount X-gal staining at day 10. X-gal staining was observed in tumors harvested from animals treated with mifepristone (B), but not in tumorsfrom animals treated with vehicle control (A). Bar, 1 mm. C, �-gal activity was not detected in tissue sections of vehicle control tumors. D and E, tissue sections frommifepristone-treated tumors. �-gal activity was restricted to stromal cells. lacZ-positive cells were observed adjacent to LNCaP clusters (D, arrows) and surrounding blood vessels ina mural cell (pericyte) position (E, arrows). Bar (C–E), 50 �m.

3305




three-way L/M/S tumors exhibited normal-appearing vessels as earlyas day 7 and were well vascularized with mature vessels by day 10after inoculation. Accordingly, the pattern and type of vascularizationin two-way and three-way LNCaP xenograft tumors appear to bedependent on the presence of stromal cells.

Precancerous PIN lesions in the human prostate gland inducephenotypic changes in the immediate stroma, characterized by acti-vation of stromal cells and stimulation of matrix remodeling (42).4

Increased microvessel density in the stroma adjacent to PIN lesionshas also been reported (11, 12). This suggests that early events inepithelial transformation act to induce reactive stroma, which in turnstimulates angiogenesis required for further progression of the pre-cancerous lesion. Indeed, the preponderance of data now indicate thatangiogenic switches occur early in the premalignant stages of mostcarcinomas (43, 44). Our data suggest that the induction of a reactivestroma in developing cancer with the resulting angiogenic switch is animportant and probably required step in successful tumorigenesis.

Although our studies have addressed angiogenesis, stromal cells arelikely to provide paracrine-acting growth factors as well as structuraland regulatory ECM components that directly affect LNCaP cellsurvival and growth. Accordingly, the stromal microenvironmentlikely regulates tumorigenesis through several mechanisms. Indeed,coculture of prostate carcinoma-associated fibroblasts with trans-formed prostate epithelial cells resulted in increased proliferation andreduced apoptosis of the epithelial cells (10). Again, these observa-tions suggest that the three-way xenograft approach is a more com-plete tumor model.

We have termed the experimental approach reported here as theDRS xenograft model. The model has been useful to address the roleof the stromal microenvironment in tumorigenesis. Furthermore, wehave extended the utility of the model by showing the regulatedexpression of a transgene in the stromal cells of an experimentalcarcinoma xenograft. The DRS model system will allow for the directtesting of stromal cell gene products in regulating carcinoma tumor-igenesis. This approach will be valuable in dissecting specific mech-anisms of stromal-epithelial interactions in prostate cancer. Of partic-ular value was the finding that this system was efficient in regulationof transgene expression. Expression was stimulated in response tomifepristone; however, it was undetectable in vehicle control-treatedanimals. In addition, the model is based on the use of frozen cellaliquots (epithelial and stromal cells) for establishing the xenograft.The use of frozen cell aliquots as opposed to passaging cells fromgrowing cultures greatly increased throughput and homogeneity ofxenograft experiments. Combined with the use of engineered stromalcells, this provides for a modular approach. We anticipate being ableto combine carcinoma cells with differentially transfected stromalcells, all derived from an initial pSwitch transfectant cell line. Theadvancement of regulated stromal gene expression in the DRS xe-nograft tumor model is significant because promoters do not currentlyexist for targeting gene expression to the stromal compartment of atransgenic animal in a tissue-specific manner. Moreover, regulatableexpression with the GeneSwitch system will allow for expression of atransgene at any time point during tumorigenesis. One potential caveatof the DRS model in its current form is the use of 3T3 cells as anengineered stromal cell source. For the specific questions asked in thisstudy, use of 3T3 cells has been suitable for examination of earlyevents; however, 3T3 cells are tumorigenic and can form sarcomaswith time in nude mice. Accordingly, we are establishing GeneSwitchtransfectants of prostate stromal cell lines derived from C57/BL6 micefor use in further studies with human carcinoma cells in nude miceand with TRAMP carcinoma cells (45) in syngeneic C57/BL6 hosts.

The studies presented here represent an advancement in the under-standing of stroma-regulated events in early tumorigenesis, particu-

larly in the regulation of angiogenesis. Furthermore, these studieshave led to the development of a modular and rapid DRS xenograftmodel system, which will be useful in evaluating the biologicalfunction of differential gene expression in the stromal compartment ofcancer. It is becoming more clear that the stromal compartment isimportant to tumor development and progression. It follows that thedevelopment of prognostic markers and therapeutic targets in thestromal compartment is a distinct possibility in the future.

ACKNOWLEDGMENTS

We thank Liz Hopkins for histological preparation of tissues and Dr. StevenRessler for thoughtful discussions.

REFERENCES

1. Ronnov-Jessen, L., Petersen, O. W., and Bissell, M. J. Cellular changes involved inconversion of normal to malignant breast: importance of the stromal reaction. Physiol.Rev., 76: 69–125, 1996.

2. Noel, A., and Foidart, J. M. The role of stroma in breast carcinoma growth in vivo.J. Mammary Gland Biol. Neoplasia, 3: 215–225, 1998.

3. Martin, M., Pujuguet, P., and Martin, F. Role of stromal myofibroblasts infiltratingcolon cancer in tumor invasion. Pathol. Res. Pract., 192: 712–717, 1996.

4. Tuxhorn, J. A., Ayala, G. E., and Rowley, D. R. Reactive stroma in prostate cancerprogression. J. Urol., 166: 2472–2483, 2001.

5. Dvorak, H. F. Tumors: wounds that do not heal. Similarities between tumor stromageneration and wound healing. N. Engl. J. Med., 315: 1650–1659, 1986.

6. Clark, R. A. F. Wound repair. In: R. A. F. Clark (ed.), The Molecular and CellularBiology of Wound Repair, pp. 3–50. New York: Plenum Publishing Corp., 1996.

7. Sieweke, M. H., and Bissell, M. J. The tumor-promoting effect of wounding: a possiblerole for TGF-�-induced stromal alterations. Crit. Rev. Oncog., 5: 297–311, 1994.

8. Rowley, D. R. What might a stromal response mean to prostate cancer progression?Cancer Metastasis Rev., 17: 411–419, 1998.

9. Elenbaas, B., Spirio, L., Koerner, F., Fleming, M. D., Zimonjic, D. B., Donaher, J. L.,Popescu, N. C., Hahn, W. C., and Weinberg, R. A. Human breast cancer cellsgenerated by oncogenic transformation of primary mammary epithelial cells. GenesDev., 15: 50–65, 2001.

10. Olumi, A. F., Grossfeld, G. D., Hayward, S. W., Carroll, P. R., Tlsty, T. D., andCunha, G. R. Carcinoma-associated fibroblasts direct tumor progression of initiatedhuman prostatic epithelium. Cancer Res., 59: 5002–5011, 1999.

11. Brawer, M. K., Deering, R. E., Brown, M., Preston, S. D., and Bigler, S. A. Predictorsof pathologic stage in prostatic carcinoma. The role of neovascularity. Cancer (Phila.),73: 678–687, 1994.

12. Sinha, A. A., Gleason, D. F., Staley, N. A., Wilson, M. J., Sameni, M., and Sloane,B. F. Cathepsin B in angiogenesis of human prostate: an immunohistochemical andimmunoelectron microscopic analysis. Anat. Rec., 241: 353–362, 1995.

13. Huss, W. J., Hanrahan, C. F., Barrios, R. J., Simons, J. W., and Greenberg, N. M.Angiogenesis and prostate cancer: identification of a molecular progression switch.Cancer Res., 61: 2736–2743, 2001.

14. Matrisian, L. M., Cunha, G. R., and Mohla, S. Epithelial-stromal interactions and tumorprogression: meeting summary and future directions. Cancer Res., 61: 3844–3846, 2001.

15. Lim, D. J., Liu, X. L., Sutkowski, D. M., Braun, E. J., Lee, C., and Kozlowski, J. M.Growth of an androgen-sensitive human prostate cancer cell line, LNCaP, in nudemice. Prostate, 22: 109–118, 1993.

16. Stewart, R. J., Panigrahy, D., Flynn, E., and Folkman, J. Vascular endothelial growthfactor expression and tumor angiogenesis are regulated by androgens in hormoneresponsive human prostate carcinoma: evidence for androgen dependent destabiliza-tion of vascular endothelial growth factor transcripts. J. Urol., 165: 688–693, 2001.

17. Navone, N. M., Logothetis, C. J., von Eschenbach, A. C., and Troncoso, P. Modelsystems of prostate cancer: uses and limitations. Cancer Metastasis Rev., 17: 361–371, 1998.

18. Rowley, D. R. Characterization of a fetal urogenital sinus mesenchymal cell line U4F:secretion of a negative growth regulatory activity. In Vitro Cell. Dev. Biol., 28A:29–38, 1992.

19. Gerdes, M. J., Dang, T. D., Lu, B., Larsen, M., McBride, L., and Rowley, D. R.Androgen-regulated proliferation and gene transcription in a prostate smooth musclecell line (PS-1). Endocrinology, 137: 864–872, 1996.

20. Abruzzese, R. V., Godin, D., Mehta, V., Perrard, J. L., French, M., Nelson, W.,Howell, G., Coleman, M., O’Malley, B. W., and Nordstrom, J. L. Ligand-dependentregulation of vascular endothelial growth factor and erythropoietin expression by aplasmid-based autoinducible GeneSwitch system. Mol. Ther., 2: 276–287, 2000.

21. Burcin, M. M., Schiedner, G., Kochanek, S., Tsai, S. Y., and O’Malley, B. W.Adenovirus-mediated regulatable target gene expression in vivo. Proc. Natl. Acad.Sci. USA, 96: 355–360, 1999.

22. Janik, P., Briand, P., and Hartmann, N. R. The effect of estrone-progesteronetreatment on cell proliferation kinetics of hormone-dependent GR mouse mammarytumors. Cancer Res., 35: 3698–3704, 1975.

23. Gao, J., Tombal, B., and Isaacs, J. T. Rapid in situ hybridization technique fordetecting malignant mouse cell contamination in human xenograft tissue from nudemice and in vitro cultures from such xenografts. Prostate, 39: 67–70, 1999.

3306




24. Fox, S. B. Microscopic assessment of angiogenesis in tumors. In: J. C. Murray (ed.),Methods in Molecular Medicine: Angiogenesis Protocols, Vol. 46, pp. 29–46. To-towa, NJ: Humana Press, Inc., 2001.

25. Drabkin, D. S., and Austin, J. H. Spectrophotometric constants for common hemoglobinderivatives in human, dog, and rabbit blood. J. Biol. Chem., 98: 719–733, 1932.

26. Kleinman, H. K., McGarvey, M. L., Hassell, J. R., Star, V. L., Cannon, F. B., Laurie,G. W., and Martin, G. R. Basement membrane complexes with biological activity.Biochemistry, 25: 312–318, 1986.

27. Swarm, R. L. Transplantation of a murine chondrosarcoma in mice of different inbredstrains. J. Natl. Cancer Inst. (Bethesda), 31: 953–975, 1963.

28. Gleave, M., Hsieh, J. T., Gao, C. A., von Eschenbach, A. C., and Chung, L. W.Acceleration of human prostate cancer growth in vivo by factors produced by prostateand bone fibroblasts. Cancer Res., 51: 3753–3761, 1991.

29. Olumi, A. F., Dazin, P., and Tlsty, T. D. A novel coculture technique demonstratesthat normal human prostatic fibroblasts contribute to tumor formation of LNCaP cellsby retarding cell death. Cancer Res., 58: 4525–4530, 1998.

30. Pretlow, T. G., Delmoro, C. M., Dilley, G. G., Spadafora, C. G., and Pretlow, T. P.Transplantation of human prostatic carcinoma into nude mice in Matrigel. CancerRes., 51: 3814–3817, 1991.

31. Gao, J., Arnold, J. T., and Isaacs, J. T. Conversion from a paracrine to an autocrinemechanism of androgen-stimulated growth during malignant transformation of pros-tatic epithelial cells. Cancer Res., 61: 5038–5044, 2001.

32. Hashizume, H., Baluk, P., Morikawa, S., McLean, J. W., Thurston, G., Roberge, S.,Jain, R. K., and McDonald, D. M. Openings between defective endothelial cellsexplain tumor vessel leakiness. Am. J. Pathol., 156: 1363–1380, 2000.

33. Darland, D. C., and D’Amore, P. A. Blood vessel maturation: vascular developmentcomes of age. J. Clin. Investig., 103: 157–158, 1999.

34. Folkman, J., and Shing, Y. Angiogenesis. J. Biol. Chem., 267: 10931–10934, 1992.35. Joseph, I. B., and Isaacs, J. T. Angiogenesis and prostate cancer. In: L. W. Chung,

W. B. Isaacs, and J. W. Simons (eds.), Prostate Cancer: Biology, Genetics, and theNew Therapeutics, pp. 279–306. Totowa, NJ: Humana Press, Inc., 2001.

36. Liekens, S., De Clercq, E., and Neyts, J. Angiogenesis: regulators and clinicalapplications. Biochem. Pharmacol., 61: 253–270, 2001.

37. Giri, D., Ropiquet, F., and Ittmann, M. Alterations in expression of basic fibroblastgrowth factor (FGF) 2 and its receptor FGFR-1 in human prostate cancer. Clin.Cancer Res., 5: 1063–1071, 1999.

38. Levine, A. C., Liu, X. H., Greenberg, P. D., Eliashvili, M., Schiff, J. D., Aaronson,S. A., Holland, J. F., and Kirschenbaum, A. Androgens induce the expression ofvascular endothelial growth factor in human fetal prostatic fibroblasts. Endocrinol-ogy, 139: 4672–4678, 1998.

39. Eliceiri, B. P., and Cheresh, D. A. Adhesion events in angiogenesis. Curr. Opin. CellBiol., 13: 563–568, 2001.

40. Benjamin, L. E., Golijanin, D., Itin, A., Pode, D., and Keshet, E. Selective ablation ofimmature blood vessels in established human tumors follows vascular endothelialgrowth factor withdrawal. J. Clin. Investig., 103: 159–165, 1999.

41. Wilson, M. J., and Sinha, A. A. Human prostate tumor angiogenesis in nude mice:metalloprotease and plasminogen activator activities during tumor growth and neo-vascularization of subcutaneously injected Matrigel impregnated with human prostatetumor cells. Anat. Rec., 249: 63–73, 1997.

42. Xue, Y., Smedts, F., Latijnhouwers, M. A., Ruijter, E. T., Aalders, T. W., de laRosette, J. J., Debruyne, F. M., and Schalken, J. A. Tenascin-C expression in prostaticintraepithelial neoplasia (PIN): a marker of progression? Anticancer Res., 18: 2679–2684, 1998.

43. Hanahan, D., and Folkman, J. Patterns and emerging mechanisms of the angiogenicswitch during tumorigenesis. Cell, 86: 353–364, 1996.

44. Ali, I. U., Senger, D. R., and Smith, L. E. Angiogenesis as a potential biomarker inprostate cancer chemoprevention trials. Urology, 57: 143–147, 2001.

45. Foster, B. A., Gingrich, J. R., Kwon, E. D., Madias, C., and Greenberg, N. M.Characterization of prostatic epithelial cell lines derived from transgenic adeno-carcinoma of the mouse prostate (TRAMP) model. Cancer Res., 57: 3325–3330,1997.

3307




2002;62:3298-3307. Cancer Res Jennifer A. Tuxhorn, Stephanie J. McAlhany, Truong D. Dang, et al. Xenograft ModelProstate Tumors in a Differential Reactive Stroma (DRS) Stromal Cells Promote Angiogenesis and Growth of Human

Updated version

http://cancerres.aacrjournals.org/content/62/11/3298

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/62/11/3298.full#ref-list-1

This article cites 43 articles, 14 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/62/11/3298.full#related-urls

This article has been cited by 39 HighWire-hosted articles. Access the articles 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/62/11/3298To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/content/62/11/3298.full#ref-list-1

http://cancerres.aacrjournals.org/content/62/11/3298.full#related-urls

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

mailto:[email protected]



stromal cells promote angiogenesis and growth of human ... · carcinoma cells is fundamentally...

Documents