Endocrine-Related Cancer (2008) 15 91–100

Arachidonic acid modulates the crosstalkbetween prostate carcinoma and bonestromal cells

Adriano Angelucci1, Stefania Garofalo1, Silvia Speca1, Antonella Bovadilla1,Giovanni Luca Gravina2, Paola Muzi2, Carlo Vicentini 3 and Mauro Bologna1,2

1Department of Basic and Applied Biology, University of L’Aquila, 67100 L’Aquila, Italy2Department of Experimental Medicine, University of L’Aquila, 67100 L’Aquila, Italy3Department of Surgery, University of L’Aquila, 67100 L’Aquila, Italy

(Correspondence should be addressed to A Angelucci; Email: adriano.angelucci@gmail.com)

Abstract

Diets high in n-6 fatty acids are associated with an increased risk of bone metastasis from prostatecarces (PCa). The molecular mechanism underlying this phenomenon is largely unknown.Arachidonic acid (AA) and its precursor linoleic acid can be metabolized to produce pro-inflammatorycytokines that act as autocrine and paracrine regulators of cancer behavior. We and other authorshave previously reported that factors released by PCa cells excite an aberrant response in bonemarrow stromal cells (BMSCs). We planned to study how AA may modulate in vitro the interactionbetweenPCacells and humanBMSCs.First,weobserved that AA isa potent mitogenic factor for PCacells through the production of both 5-lipoxygenase (5-LOX) and cyclooxygenase-2 (COX-2)metabolites. While 5-LOX controls cell survival through the regulation of the Bcl-2/Bax ratio, COX-2activity stimulates the release of transforming growth factor-a (TGF-a) and pro-inflammatorycytokines. The blockade of COX-2 activity through a specific inhibitor is sufficient to repressAA-induced gene transcription. The over-expression of transforming growth factor -a (TGF-a),tumour necrosis factor-a (TNF-a) and interleukin-1 b (IL-1b) by AA-primed PCa cells resultedparticularly effective in modifying cell behavior of cultured human BMSCs. In fact, we observed anincrement in the cell number of BMSCs, due prevalently to the action of TGF-a, the number ofosteoblasts, and the production of receptor activator for nuclear factor k B ligand (RANKL), eventsmainly controlledby inflammatorycytokines.Thesefindingsprovidea possible molecular mechanismby which dietary n-6 fatty acids accumulating in bone marrow may influence the formation of PCa-derived metastatic lesions and indicate new molecular targets for the therapy of metastatic PCa.

Endocrine-Related Cancer (2008) 15 91–100

Introduction

Increasing evidence associates dietary behavior to

cancer. One of the most consistent findings in this

matter suggests that long-chain polyunsaturated fatty

acids may play an important role in cancer develop-

ment and progression. In particular, epidemiological

and animal model studies have indicated that the ratio

of long chain n-6 to n-3 fatty acids may be associated

with the risk of cancer (Terry et al. 2003). Arachidonic

acid (AA) and its precursor, linoleic acid, can directly

stimulate in vitro growth of several types of cancer

cells, including human prostate cancer (PCa) cells.

Moreover, a strong association between n-6 fatty acids

Endocrine-Related Cancer (2008) 15 91–100

1351–0088/08/015–091 q 2008 Society for Endocrinology Printed in Great

content in the diet and the risk of metastatic PCa exists

(Wynder et al. 1994, Wang et al. 1995, Rose 1997).

AA is present in cellmembranes as phospholipids, and

it is released by cytosolic phospholipase A2 in response

to a number of stimuli, including activation of

membrane-bound receptors. However, a strict relation-

ship between the amount of available AA in the

membrane and the diet exists. In fact, as demonstrated

by preclinical studies, intake of dietary n-6 fatty acid

regulates the levels of AA not only in serum but also in

membrane phospholipids (Kobayashi et al. 2006). AA is

metabolized via either the cyclooxygenase (COX) or

lipoxygenase (LOX) pathway producing prostaglandins

Britain

DOI: 10.1677/ERC-07-0100

Online version via http://www.endocrinology-journals.org

A Angelucci et al.: Arachidonic acid and bone metastasis

and hydroperoxyeicosatetraenoic acids (HETEs)

respectively. The development of tumor in humans and

experimental animals is consistently linked to aberrant

AA metabolism through the COX and LOX pathways

(Furstenberger et al. 2006). Aberrant AA metabolism

in tumor cells seems to be dependent on the action of

COX-2 and specific LOX (Cuendet & Pezzuto 2000),

and these enzymes are over-expressed in PCa in a strict

association with tumor grade (Madaan et al. 2000).

Although different LOXs may exert opposite effects on

PCa cells, recent studies have focused on 5-LOX

activity as a key regulator of tumor aggressiveness in

PCa cell lines (Ghosh & Myers 1998a, Moretti et al.

2004, Hassan & Carraway 2006). Moreover, 5-LOX has

been considered an attractive target, together with

COX-2, in antitumoral therapy, in cancer types

possessing a basal high capacity in metabolizing AA,

like PCa (Gugliucci et al. 2002).

Most patients with advanced PCa develop bone

metastases and suffer from long-term skeletal

morbidity, involving pain, pathological fractures,

and spinal cord compression, with a very significant

impact on the quality of life. The preferential

spreading of PCa cells to bone may be explained

by the fact that bone marrow represents a favorable

microenvironment for the growth of cancer cells with

a specific genotype. At the same time, PCa cells

mimic in some instances bone cells perturbing the

normal process of bone remodeling and providing a

niche in which they can grow and expand. Clinical

evidence of bone metastasis from PCa demonstrates

the existence of an intense crosstalk between cancer

and bone cells resulting in mixed bone lesions

controlled by an aberrant activation of osteoblasts

and osteoclasts (Vessella & Corey 2006). Thus, the

molecular interaction between bone and tumor cells

appears to be a critical step in the growth of

metastatic cells in the bone. To date, an effective

anti-metastatic targeted therapy for Pca has not been

found, and new data able to clarify the underlying

molecular mechanisms are needed.

In this report, we analyze the effect of exogenous

AA on metastastic PCa cells. In particular, we evaluate

in vitro the possible role of AA in determining the

formation of metastasis in bone microenvironment.

Our data support a dual pro-metastatic effect of an

AA-enriched environment: 1) assuring a high proli-

ferative rate in tumor cells and 2) determining through

the tumor metabolism an abnormal response in bone

stromal cells. In vivo evidence supporting this

hypothesis and justifying new therapeutic strategies

is needed.

92

Materials and methods

Cell culture

Human prostate cancer cell lines, PC3 and LNCaP, and

breast cancer cell lines, MDA-MB-231 and MCF-7,

were originally obtained by ATCC (Rockville, MD,

USA) and maintained in DMEM (except LNCaP cells

cultured in RPMI) supplemented with 10% fetal bovine

serum, glutamine, and penicillin–streptomycin

(Sigma). Human stromal cells were isolated from

surgical human bone specimens obtained by routine

clinical practice with the informed consent of the

patients as described previously (Ciapetti et al. 2006).

For the evaluation of osteoblasts content, bone stromal

cells were fixed in 4% formaldehyde in 0.1 phosphate

buffer, pH 7.2, and then incubated for 30 min with a

solution of 2% Fast violet B salt (Sigma) and 4%

Naphthol AS-MX phosphate alkaline solution (Sigma).

In order to evaluate cell proliferation, cells were plated

at a density of 104 cells/cm2, at various days of culture,

and were recovered with a solution of trypsin/EDTA;

washed twice with ice-cold PBS; and counted in

a Burker slide. Conditioned medium (CM) was

obtained culturing sub-confluent cells for 48 h in a

medium without serum. At the end of incubation, the

CM was recovered and centrifuged. The amount of

residual AA in the CM was evaluated in a pivotal

experiment using tritiated AA and measuring the

radioactivity in the culture supernatant. After 48 h of

culture, the residual AA was lower than 10 nM.

Colony growth assay

We evaluated PC3 cells capacity for growth at clonal

density, plating cells at densities of 10 cells/cm2 in

10% fetal bovine serum-supplemented DMEM. After 2

weeks of culture, adherent cells were fixed with cold

methanol for 10 min, washed with PBS/BSA, and air-

dried for 5 min. Adherent cells were stained with 0.5%

crystal violet for 15 min at room temperature.

The stained colonies were photomicrographed and

analyzed by number and size with the public domain

software ImageJ (by Wayne Rasband, http://rsb.info.

nih.gov/ij/).

Real-time PCR

Total RNA was extracted from the cultured cells

using Genelute Mammalian Total RNA kit (Sigma),

and 1 mg RNA was used to synthesize cDNA

(SuperScript III Platinum Kit, Invitrogen). Real-time

PCR analysis was performed using Stratagene

MX3000P personal Q-PCR (Stratagene, La Jolla,

CA, USA) and SuperScript III Platinum Kit

www.endocrinology-journals.org

Endocrine-Related Cancer (2008) 15 91–100

(Invitrogen). For all genes, 5 ml cDNA were used

except for GAPDH amplification that was performed

using 2.5 ml cDNA. Primers were designed using the

online tool Primer3 (http://frodo.wi.mit.edu/cgi-bin/

primer3/primer3_www.cgi), and the sequences were as

shown in Table 1. Mean threshold cycle (Ct) values

were determined by Stratagene software using three

distinct amplification curves for each gene. Relative

expression of the target gene was estimated using the

formula: relative expressionZ2KDCt , where DCtZCt

(target gene)–Ct (GAPDH).

Western blotting

Total cell lysates were obtained by resuspending the

cells in a buffer containing 1% Triton, 0.1% SDS, 2 mM

CaCl2, and 100 mg/ml phenylmethyl sulfonyl fluoride.

Protein content was determined using the Protein Assay

Kit 2 (Bio-Rad Laboratory). Eighty micrograms of

proteins were electrophoresed in 10% SDS–polyacryla-

mide gel and then electrotransferred to a nitrocellulose

membrane (Schleicher&Schuell, Dassel, Germany). The

membrane was incubated with 1 mg/ml primary antibody

and then with specific horseradish peroxidase-conjugated

secondary antibodies. Protein bandswere visualized using

a chemiluminescent detection system (Amersham Bios-

ciences). Primary and secondary antibodies were pur-

chased from SantaCruz Biotechnology (Santa Cruz, CA,

USA) except blocking antibodies against tumour necrosis

factor-a (TNF-a) and interleukin-1b (IL-1b) which werepurchased from Calbiochem (Darmstadt, Germany).

Statistical analysis

Results are expressed as meansGS.D. for at least three

distinct experiments. Demonstration of significant

Table 1 Sequences of primers used in the real-time PCR

amplifications

Gene Primer sequences

EGF 5 0-TCACCTCAGGGAAGATGACC-30

5 0-CAGTTCCCACCACTTCAGGT-3 0

TGF-a 5 0-TTTGTGGGCCTTCAAAACTC-30

5 0-AACTGCTGCACACACCTCAC-3 0

TNF-a 5 0-TCCTTCAGACACCCTCAACC-30

5 0-AGGCCCCAGTTTGAATTCTT-30

IL-1b 5 0-GGGCCTCAAGGAAAAGAATC-3 0

5 0-TTCTGCTTGAGAGGTGCTGA-30

RANKL 5 0-CCTTTTGCTCATCTCACTATT-30

5 0-AATGTTGGCATACAGGTAATA-30

GAPDH 5 0-GGCCTCCAAGGAGTAAGACC-3 0

5 0-AGGGGTCTACATGGCAACTG-30

www.endocrinology-journals.org

differences among means was performed by Student’s

t analysis considering 0.01 as the threshold value of P.

All statistical analyses were performed using Kaleida-

graph 3.6 (Synergy Software, Reading, PA, USA).

Figure 1 Effects of exogenous arachidonic acid on PC3 cells.(a) PC3 cells were treated for 48 (black bars) and 72 h (whitebars) with increasing doses of arachidonic acid (mM), and thecell number was recorded and expressed as percentage withrespect to control (100%). (b) PC3 cells were treated with 10 mMarachidonic acid (AA), an inhibitor of 5-lipoxygenase (MK886,1 mM), an inhibitor of cyclooxygenase-2 (MK663, 10 mM), andcombinations of these substances. Cell counts were recordedat 48 (black bars) and 72 h (white bars). (c) Protein expressiondetected by western blot of Bcl-2, Bax, and actin in the presenceof arachidonic acid (AA, 10 and 50 mM) and AA plus 1 mMMK886. In the histograms on the right, relative densitometricvalues for each band are reported.

93

A Angelucci et al.: Arachidonic acid and bone metastasis

Results

Exogenous AAmodulates PCa cells’ proliferative

status

Adose-dependent cell response toAA at 48 and 72 hwas

evaluated (Fig. 1a). AA determined a strong induction in

cell proliferation evident from 48 h and lasting at least

until 72 h. Treatments with concentrations higher than

10 mM AA were progressively cytotoxic. In order to

analyze themetabolic pathways involved inAA-induced

proliferation, we used two chemical specific inhibitors

Figure 2 (a) Relative mRNA expression by q-PCR in PC3 cellsarachidonic acid for 12 h alone or in the presence of the inhibitor of 5-(MK663, 10 mM). Relative gene expression was calculated assigninrepresents the mean (GS.D.) of three different experiments. (*P!0.western blot of COX-2, TNF-a and actin in the presence or absence231, LNCaP, and MCF-7 cells.

94

for 5-LOX (MK886) and COX-2 (MK663). MK663

selectively inhibits COX-2 with an IC50 value of about

1 mM for COX-2 and a selectivity ratio (COX-1/COX-2

IC50) for the inhibition of COX-2 of 106 (Riendeau et al.

2001). MK886 is a 5-LOX ‘translocation inhibitor’ with

an IC50 value of 0.02 mM (Parkar et al. 1990). Since

MK886 has also been demonstrated to decrease PC3 cell

proliferation independently from 5-LOX inhibition, we

used it at a concentration (1 mM) that did not have a

significant cytotoxic effect on PC3 cell (Fig. 1b). 5-LOX

inhibitor but not MK663 was able to annul the

treated with arachidonic acid. PC3 cells were incubated withlipoxygenase (MK886, 1 mM) or the inhibitor of cyclooxygenase-2g the reference value of 1 to the untreated cells. Each value01 treated versus control). (b) Protein expression detected byof 10 mM arachidonic acid in total cell lysates of PC3, MDA-MB-

www.endocrinology-journals.org

Figure 3 Autocrine mitogenic effect of arachidonic acid-primedPC3 cells. (a) Conditioned media from PC3 cells untreated ortreated with AA (AA CM) or with AA plus MK663 (AACMK663)for 12 h were added to PC3 cells previously starved for 24 h.PC3 cells were cultured also in the presence of AA CM and aninhibitor of EGFR (ZD1839, 10 mM). Cells were treated for 48 hand then counted. (b) PC3 cells cultured at low density weretreated for 14 days with conditioned media from PC3 cellsuntreated or treated with AA for 12 h (AA CM) with or without10 mM ZD1839. Histograms show the number of colonies in theordinate and the size (mm) of colonies in the abscissae.(*P!0.01 treated versus control).

Endocrine-Related Cancer (2008) 15 91–100

proliferative effect of AA both at 48 and 72 h. When the

protein expression of Bcl-2 and Bax is analyzed, we

observed a thigh linkage between their relative presence

and the concentration of AA (Fig. 1c). While 10 mMAA

determined an anti-apoptotic equilibrium, upregulating

Bcl-2 and downregulating Bax, 50 mM AA had an

opposite effect, stimulating a pro-apoptotic status.

Moreover, MK886, but not MK663 (data not shown),

counteracted the upregulation of Bcl-2 in the presence of

AA (Fig. 1c).

Gene transcription induction by AA

PC3 cells were cultured in the presence of 10 mMAA for

12 h, and total mRNA was isolated and screened in real-

time PCR with a panel of primers specific for

inflammatory cytokines. AA stimulated a significant

upregulation in TNF-a and IL-1b mRNA (Fig. 2a). In

order to evaluate the existence of an autocrine effect based

upon the secretion of growth factors, we evaluated the

expression levels of mRNAs for epidermal growth factor

receptor (EGFR) ligands, EGF and transforming growth

factor-a (TGF-a). TGF-a gene transcription was stimu-

lated by AA treatment with an increase of about twofold

with respect to the value in the untreated cells (Fig. 2a).On

thecontrary,EGFmRNAdidnotdemonstrate a significant

variation with respect to control values. The concomitant

addition of MK663, but not MK886, was able to block

completely the gene transcription effect of AA. At the

same time, the addition of the inhibitors didnot change in a

significant manner the basal values of expression of the

examined genes. We analyzed the protein expression

level of COX-2 and TNF-a in highly osteotropic PC3 and

MDA-MD-231 cells and in low aggressive, non-meta-

static LNCaP and MCF-7 cells (Fig. 2b). COX-2

was detectable in all the cell lysates but PC3 and MDA-

MD-231cells expressed it at the highest level.Only inPC3

cells AA stimulated after 24 h a detectable upregulation of

COX-2. In basal culture conditions, TNF-a was evident

only in PC3 cells. The addition of AA determined the

upregulation of TNF-a, both in PC3 and MDA-MB-231

cells but not in other cell lines.

Autocrine effects of AA-induced TGF-a

The autocrine role of factors released uponAA treatment

was evaluated using the CM from PC3 cells primed for

48 h with 10 mM AA. The CM was added to PC3 cells

cultured in serum-free medium, and the cell number was

evaluated after 72 h (Fig. 3a). The CM from AA-treated

PC3 cells determined an increment of about 30% in

the cell number with respect to untreated cells. When we

used the CM from PC3 cells treated with AA and

MK663, we did not observe any significant effect.

www.endocrinology-journals.org

The proliferative stimulus induced by the CM was

blocked by EGFR inhibitor ZD1839, suggesting that

TGF-amay play a leading role in the autocrine growth of

PC3 cells. The same effects were detected at a greater

extent when we treated PC3 cells cultured at low density

(Fig. 3b). In this experimental situation, the CM from

AA-primed PC3 cells determined a significant increase

in the number and extension of visible PC3 cell colonies

after 14 days. Blocking EGFR activation determined

a significant reduction in the growth of cell colonies both

in basal culture condition and in the presence of AACM.

95

A Angelucci et al.: Arachidonic acid and bone metastasis

Paracrine effects on human BMSCs

We cultured stromal cells from human bone marrow

(BMSCs) with the CM from AA-treated PC3 cells and

evaluated BMSC cell number with respect to control

after 72 h of culture (Fig. 4a). The CM from

AA-treated PC3 cells determined an increase in the

number of BMSCs of w30% with respect to untreated

BMSCs. The CM from AACMK663-treated PC3 cells

did not determine any significant difference with

respect to control cells. The increase in proliferation

was abolished by the addition of the inhibitor ZD1839,

suggesting a leading role for EGFR transduction axis.

At the same time, we evaluated the presence of

osteoblasts through the detection of alkaline phospha-

tase (ALP) activity. In the BMSC population cultured

in 15% fetal bovine serum-supplemented medium plus

one-fourth of PC3 CM we detected !5% of ALP-

positive cells. This percentage significantly grew in the

presence of CM from AA-treated PC3 cells, but not

Figure 4 The paracrine effect of arachidonic acid-primed PC3 on hPC3 cells untreated or treated with 10 mM AA (AA CM) or with AABMSCs cultured at sub-confluent density. BMSCs were cultured atreated for 72 h and then counted. Three different experiments werimages on the right show exemplificative optical fields for each exp(b) Conditioned media from PC3 cells untreated or treated with 10 m12 h were added to BMSCs. The involvement of EGFR activation wZD1839. Cells were cultured for 72 h and then stained for the preseand percentages with respect to the total number of BMSCs were cavalues (GS.D.) are reported. The images on the right show exemplpositive for ALP activity. (*P!0.01 treated versus control).

96

with the CM from AACMK663-treated PC3 cells,

reaching 10% of the total BMSC population (Fig. 4b).

As demonstrated by the addition of ZD1839, the

formation of osteoblasts is only partially dependent on

EGFR functioning (Fig. 4b).

The role of TNF-a and IL-1b secreted by PC3 cells

on BMSCs

In order to verify the effect ofTNF-a and IL-1bpresent inPC3CMonosteoblast differentiation,we treatedBMSCs

with the CM from AA-treated PC3 cells in the presence

of the blocking antibody against TNF-a or IL-1b. When

the expression ofALP activity inBMSCs is analyzed, we

observed that the antibody anti-TNF-a completely

blocked the differentiative effect of AA CM, reverting

the percentage of osteoblasts to expectedvalues (Fig. 5a).

On the contrary, the blocking antibody against IL-1bwasnot able to reduce significantly the differentiative effect

of AA. As demonstrated in our previous data, PC3CM is

uman bone stromal cells (BMSCs). (a) Conditioned media fromplus 10 mM MK663 (AACMK663 CM) for 12 h were added tolso in the presence of AA CM plus 10 mM ZD1839. Cells weree performed and mean values (GS.D.) are reported. Theerimental group as visualized by contrast phase microscopy.M AA (AA CM) or AA plus 10 mM MK663 (AACMK663 CM) for

as tested culturing BMSCs in the presence of AA CM and 10 mMnce of alkaline phosphatase (ALP). Positive cells were countedlculated. Three different experiments were performed and meanificative field for each experimental group. Dark cells are those

www.endocrinology-journals.org

Figure 5 The effect of TNF-a and IL-1b produced by PC3 cellson BMSCs. (a) Conditioned media from PC3 cells untreatedor treated with 10 mM AA for 12 h (AA CM) in the presence orabsence of blocking antibody against TNF-a or IL-1b(10 mg/ml) were added to BMSCs cultured at sub-confluentdensity. Cells were cultured for 72 h and stained for thepresence of alkaline phosphatase (ALP). The images showexemplificative optical fields showing the reduction in ALP-positive cells in the presence of the antibody anti-TNF-a.(b) Relative mRNA expression levels for RANKL by q-PCR inBMSCs. Conditioned media from PC3 cells untreated ortreated with 10 mM AA (AA CM) or with AA plus 10 mM MK663(AACMK663 CM) for 12 h were added to BMSCs cultured atsub-confluent density. AA CM was administered also incombination with the blocking antibody anti-TNF-a or IL-1b(10 mg/ml) or with 10 mM ZD1839. Relative gene expressionwas calculated assigning the value of 1 to the untreated cells.The values represent the mean (GS.D.) of three differentexperiments. (*P!0.01 treated versus control).

Endocrine-Related Cancer (2008) 15 91–100

www.endocrinology-journals.org

able to induce the receptor activator for nuclear factorkBligand (RANKL) expression in BMSCs (Angelucci et al.

2006). We observed that treating PC3 cells with AA

determined the release of factors that are able to enhance

the gene transcription of RANKL in BMSCs (Fig. 5b).

The effect of AA CMwas only partially counteracted by

the EGFR inhibitor ZD1839, while the concomitant

addition of antibody anti-TNF-a or anti-IL-1bdetermined a drastic reduction in RANKL mRNA

expression with respect to AA CM-treated BMSCs

(Fig. 5b).

Discussion

Several authors have previously demonstrated that

tumor cells over-express AA metabolizing enzymes,

and this feature may establish a bridge between the

dietary fat intake and the progression of cancer. In

particular, dietary n-6/n-3 fatty acid ratio seems to

control not only the membrane fatty acid composition

but also the growth of PCa xenografts in preclinical

models (Kobayashi et al. 2006). Therefore, we may

propose that PCa cells over-producing LOX and COX

enzymes are able to take an important advantage for

their growth from the presence of elevated concen-

trations of AA in their microenvironment. Our study

and a plethora of data from other authors indicate that

both LOX and COX pathways have to be considered

important in the definition of AA-induced tumor

aggressive behavior. The most evident effect of AA

on PCa cells is a dose-dependent induction of cell

proliferation. Products responsible for mitogenic effect

of AA have included eicosanoids as well as several

LOXmetabolites. This mitogenesis is blocked if partial

metabolism of AA is interrupted by the use of MK886,

the inhibitor of 5-LOX (Ghosh & Myers 1998b).

Moreover, the levels of 5-HETE, the metabolite

produced by 5-LOX, are strictly correlated in PC3

cells with DNA synthesis (Hassan & Carraway 2006).

Interestingly, the same authors indicated that 5-HETE

is a downstream effector in EGFR signaling. Our data

confirm that 5-LOX has a leading role in the control of

PCa cell proliferation in the normal culture conditions;

in fact the inhibitor MK886 is able to suppress the

mitogenic effect induced by AA completely. This

phenomenon has a molecular counterpart in the

relative expression levels of Bcl-2, elevated in the

presence of AA, and Bax, downregulated by AA.

The addition of MK886, but not of the COX-2 inhibitor

MK663, was able to restore the basal levels of Bcl-2,

thereby controlling the proliferative capacity of cells.

Epidemiological data indicate that high-fat diets,

especially those rich in saturated n-6 fatty acids, can

97

Figure 6 Proposed diagram of arachidonic acid (AA)-stimulated cross talk between PC3 cells and bone stromal cells. AA ismetabolized in PC3 cells through 5-lipoxygenase (5-LOX) and cyclooxygenase-2 (COX-2). The first pathway leads to cellsurvival/proliferative events controlling mitochondrion integrity, and the second pathway leads to gene transcription of TGF-a andpro-inflammatory cytokines, including TNF-a and IL-1b. TGF-a controls the proliferation and migration of PC3 cells in an autocrinemanner and bone stromal cells in a paracrine manner through binding to EGFR. TNF-a released by PC3 cells determines theformation of osteoblasts and the production of RANKL, events associated with the concomitant digestion of mineralized bone matrixby osteoclasts. IL-1b stimulates the production of RANKL by BMSCs. Osteoclast differentiation has also been described as directlydependent on the presence of pro-inflammatory cytokines.

A Angelucci et al.: Arachidonic acid and bone metastasis

affect bone health, especially during development and

in aged individuals. One of the proposed mechanisms

involves the generation of active fatty acid metabolites

in bone marrow stromal cells, which are able to direct

the proliferation and differentiation of these cells

(Nuttall et al. 1998). Interestingly, a direct correlation

between reduced bone density and increase in marrow

fat seems to exist, a condition frequently observed in

advanced age (Burkhardt et al. 1987). This evidence

suggests that fat intake may influence the metastatic

tumor cell growth in bone marrow and, in particular, in

aged individuals. In agreement with this hypothesis are

data indicating that PCa cells interact with adipocytes

in vitro with a resultant increase in their proliferation

(Tokuda et al. 2003), and that they localize to lipid-rich

regions in bone marrow metastases. In particular,

Brown et al. (2006) have demonstrated that AA

promotes invasion in PC3 cells and may represent

one of the key molecules that direct the PCa cells to the

bone marrow. If we consider that EGFR activation is

implicated in PCa cell migration and invasion, some of

the features indicated by Brown may be explained

through the AA-induced autocrine release of TGF-a.EGFR activation may play a pleiotropic role in bone

metastasis formation. In fact, as demonstrated pre-

viously, factors released under the control of EGFR

activation are able to modulate bone remodeling

through the induction of RANKL in human osteoblasts

98

(Angelucci et al. 2006). This evidence together with

the data presented here suggests that AA and EGFR

may share downstream common molecular events

mediated by the transcription of the same genes.

Moreover, since several authors have demonstrated

that AA is a secondary messenger released upon EGFR

activation (Hassan & Carraway 2006), we may

hypothesize the existence of an autocrine loop

maintaining EGFR continuously activated.

The CM from AA-primed PC3 cells stimulates

in vitro a multifaceted response in human BMSCs

(Fig. 6). TGF-a through the activation of EGFR

promotes the proliferation of BMSCs, but its action is

not sufficient to explain the other effects on BMSCs. On

the contrary, TNF-a and IL-1b play a central role in

modulating the formation of osteoblasts and the

production of RANKL. Because IL-1b was able to

stimulate the production of RANKL but not the

formation of osteoblasts, we hypothesize that the two

phenomena are controlled by different molecular

mechanisms. For this reason, the TNF-a/IL-1b ratio in

a complex contest in which several other factors,

usually in a net of reciprocal interactions, can modulate

bone remodeling (see (Yoneda & Hiraga 2005)), may

establish the outcome of bone lesion, lytic, or sclerotic.

TNF-a is amultifunctional cytokine defined as a leading

factor in chronic inflammatory osteolysis is able

to stimulate osteoclastogenesis (Kitaura et al. 2004).

www.endocrinology-journals.org

Endocrine-Related Cancer (2008) 15 91–100

Both stromal cells and osteoclast precursors are TNF-atargets, but stromal cells seem to make greater

contribution in TNF-a-induced osteoclastogenesis

(Kitaura et al. 2005). In our hypothesis, the BMSCs

are fundamental in determining bone remodeling

directed by PCa cells. In fact, BMSCs are both able to

release osteoclastogenic factors, i.e., RANKL, and able

to differentiate in osteoblasts, thereby controlling

potentially both resorption and deposition of miner-

alized bone matrix (Fig. 6). In agreement with this

hypothesis is the evidence that bone metastasis, also in

the presence of a predominant type of lesion, develops

through the aberrant activation of both osteoblasts and

osteoclasts. Because our study was conducted in vitro

using human cells and simulating a microenvironment

rich in AA, it is possible that results obtained through

the use of murine experimental models may differ from

ours (Gamradt et al. 2005).

The proposed model may contribute to elucidate the

largely obscure role of AA in PCa cancer progression.

A diet rich in n-6 fatty acids can significantly alter the

timing and extent of clinical outcome in PCa bone

metastasis modulating the interaction between PCa

cells and BMSCs.

Acknowledgements

This study was entirely supported by funding from the

University of L’Aquila. The authors declare that there is

no conflict of interest that would prejudice the

impartiality of the reported research.

References

Angelucci A, Gravina GL, Rucci N, Millimaggi D, Festuccia

C, Muzi P, Teti A, Vicentini C & Bologna M 2006

Suppression of EGF-R signaling reduces the incidence of

prostate cancer metastasis in nude mice. Endocrine-

Related Cancer 13 197–210.

Brown MD, Hart CA, Gazi E, Bagley S & Clarke NW 2006

Promotion of prostatic metastatic migration towards

human bone marrow stoma by Omega 6 and its

inhibition by Omega 3 PUFAs. British Journal of Cancer

94 842–853.

Burkhardt R, Kettner G, BohmW, Schmidmeier M, Schlag R,

Frisch B, Mallmann B, Eisenmenger W & Gilg T 1987

Changes in trabecular bone, hematopoiesis and bone

marrow vessels in aplastic anemia, primary osteoporosis,

and old age: a comparative histomorphometric study. Bone

8 157–164.

Ciapetti G, Ambrosio L, Marletta G, Baldini N & Giunti A

2006 Human bone marrow stromal cells: in vitro

expansion and differentiation for bone engineering.

Biomaterials 27 6150–6160.

www.endocrinology-journals.org

Cuendet M & Pezzuto JM 2000 The role of cyclooxygenase

and lipoxygenase in cancer chemoprevention. Drug

Metabolism and Drug Interactions 17 109–157.

Furstenberger G, Krieg P, Muller-Decker K & Habenicht AJ

2006 What are cyclooxygenases and lipoxygenases doing

in the driver’s seat of carcinogenesis? International

Journal of Cancer 119 2247–2254.

Gamradt SC, Feeley BT, Liu NQ, Roostaeian J, Lin YQ, Zhu

LX, Sharma S, Dubinett SM & Lieberman JR 2005 The

effect of cyclooxygenase-2 (COX-2) inhibition on human

prostate cancer induced osteoblastic and osteolytic lesions

in bone. Anticancer Research 25 107–115.

Ghosh J & Myers CE Jr 1998a Arachidonic acid metabolism

and cancer of the prostate. Nutrition 14 48–49.

Ghosh J & Myers CE 1998b Inhibition of arachidonate

5-lipoxygenase triggers massive apoptosis in human

prostate cancer cells. PNAS 95 13182–13187.

Gugliucci A, Ranzato L, Scorrano L, Colonna R, Petronilli V,

Cusan C, Prato M, Mancini M, Pagano F & Bernardi P

2002 Mitochondria are direct targets of the lipoxygenase

inhibitor MK886 A strategy for cell killing by combined

treatment with MK886 and cyclooxygenase inhibitors.

Journal of Biological Chemistry 277 31789–31795.

Hassan S & Carraway RE 2006 Involvement of arachidonic

acid metabolism and EGF receptor in neurotensin-

induced prostate cancer PC3 cell growth. Regulatory

Peptides 133 105–114.

KitauraH, SandsMS,AyaK, Zhou P, HirayamaT,Uthgenannt

B, Wei S, Takeshita S, Novack DV, Silva MJ et al. 2004

Marrow stromal cells and osteoclast precursors differen-

tially contribute to TNF-alpha-induced osteoclastogenesis

in vivo. Journal of Immunology 173 4838–4846.

Kitaura H, Zhou P, Kim HJ, Novack DV, Ross FP &

Teitelbaum SL 2005 M-CSF mediates TNF-induced

inflammatory osteolysis. Journal of Clinical Investigation

115 3418–3427.

Kobayashi N, Barnard RJ, Henning SM, Elashoff D, Reddy

ST, Cohen P, Leung P, Hong-Gonzalez J, Freedland SJ,

Said J et al. 2006 Effect of altering dietary omega-

6/omega-3 fatty acid ratios on prostate cancer membrane

composition, cyclooxygenase-2, and prostaglandin E2.

Clinical Cancer Research 12 4662–4670.

Madaan S, Abel PD, Chaudhary KS, Hewitt R, Stott MA,

Stamp GW & Lalani EN 2000 Cytoplasmic induction and

over-expression of cyclooxygenase-2 in human prostate

cancer: implications for prevention and treatment.

BJU International 86 736–741.

Moretti RM, Montagnani Marelli M, Sala A, Motta M &

Limonta P 2004 Activation of the orphan nuclear receptor

RORalpha counteracts the proliferative effect of fatty acids

on prostate cancer cells: crucial role of 5-lipoxygenase.

International Journal of Cancer 112 87–93.

Nuttall ME, Patton AJ, Olivera DL, Nadeau DP & Gowen M

1998 Human trabecular bone cells are able to express both

osteoblastic and adipocytic phenotype: implications for

osteopenic disorders. Journal of Bone and Mineral

Research 13 371–382.

99

A Angelucci et al.: Arachidonic acid and bone metastasis

Parkar BA, McCormick ME & Foster SJ 1990 Leukotrienes

do not regulate interleukin 1 production by activated

macrophages. Biochemical and Biophysical Research

Communications 169 422–429.

Riendeau D, Percival MD, Brideau C, Charleson S, Dube D,

Ethier D, Falgueyret JP, Friesen RW, Gordon R, Greig G

et al. 2001 Etoricoxib (MK-0663): preclinical profile and

comparison with other agents that selectively inhibit

cyclooxygenase-2. Journal of Pharmacological and

Experimental Therapeutics 296 558–566.

Rose DP 1997 Effects of dietary fatty acids on breast and

prostate cancers: evidence from in vitro experiments and

animal studies. American Journal of Clinical Nutrition 66

1513S–1522S.

Terry PD, Rohan TE & Wolk A 2003 Intakes of fish and

marine fatty acids and the risks of cancers of the

breast and prostate and of other hormone-related

cancers: a review of the epidemiologic evidence.

American Journal of Clinical Nutrition 77 532–543.

100

Tokuda Y, Satoh Y, Fujiyama C, Toda S, Sugihara H &

Masaki Z 2003 Prostate cancer cell growth is modulated

by adipocyte–cancer cell interaction. BJU International

91 716–720.

Vessella RL & Corey E 2006 Targeting factors involved in

bone remodeling as treatment strategies in prostate

cancer bone metastasis. Clinical Cancer Research 12

6285s–6290s.

Wang Y, Corr JG, Thaler HT, Tao Y, Fair WR & Heston WD

1995 Decreased growth of established human prostate

LNCaP tumors in nude mice fed a low-fat diet. Journal of

National Cancer Institute 87 1456–1462.

Wynder EL, Rose DP & Cohen LA 1994 Nutrition and

prostate cancer: a proposal for dietary intervention.

Nutrition and Cancer 22 1–10.

Yoneda T & Hiraga T 2005 Crosstalk between cancer cells

and bone microenvironment in bone metastasis.

Biochemical and Biophysical Research Communications

328 679–687.

www.endocrinology-journals.org