Extratumoral Macrophages Promote Tumor and Vascular Growth in an Orthotopic Rat Prostate Tumor Model

Abstract

Tumor-associated macrophages are involved in angiogenesis and tumor progression, but their role and specific site of action in prostate cancer remain unknown. To explore this, Dunning R-3327 AT-1 rat prostate tumor cells were injected into the prostate of syngenic and immunocompetent Copenhagen rats and analyzed at different time points for vascular proliferation and macrophage density. Endothelial proliferation increased with tumor size both in the tumor and importantly also in the extratumoral normal prostate tissue. Macrophages accumulated in the tumor and in the extratumoral normal prostate tissue and were most abundant in the invasive zone. Moreover, only extratumoral macrophages showed strong positive associations with tumor size and extratumoral vascular proliferation. Treatment with clodronate-encapsulated liposomes reduced the monocyte/macrophage infiltration and resulted in a significant inhibition of tumor growth. This was accompanied by a suppressed proliferation in microvessels and in the extratumoral prostate tissue also in arterioles and venules. The AT-1 tumors produced, as examined by RT² Profiler PCR arrays, numerous factors promoting monocyte recruitment, angiogenesis, and tissue remodeling. Several, namely, chemokine (C-C) ligand 2, fibroblast growth factor 2, matrix metalloproteinase 9, interleukin 1β, interferon γ, and transforming growth factor β, were highly upregulated by the tumor in vivo compared with tumor cells in vitro, suggesting macrophages as a plausible source. In conclusion, we here show the importance of extratumoral monocytes/macrophages for prostate tumor growth, angiogenesis, and extratumoral arteriogenesis. Our findings identify tumor-associated macrophages and several chemotactic and angiogenic factors as potential targets for prostate cancer therapy.

Introduction

The tumor microenvironment contains many different cell types, such as fibroblasts, myofibroblasts, adipocytes, blood, and lymph vessels, and a wide range of leukocytes that can all influence tumor progression. Tumor-associated macrophages (TAMs) are the major component of the leukocyte infiltrate, and increased intratumoral macrophage density correlates with poor prognosis in more than 80% of published studies [1]. Numerous factors recruit peripheral blood monocytes from the circulation providing the tumor with a constant supply of TAMs, and accumulating evidence suggests that tumors exploit macrophages for their own benefit. For example, TAMs can promote angiogenesis by secreting proangiogenic factors and facilitate invasion by releasing matrix-degrading enzymes [1–5].

Studies of macrophage infiltration to human prostate cancer, however, show diverse results with both positive and negative association to clinical outcome and cancer progression [6,7]. In a subcutaneous human prostate xenograft model in immune-compromised mice, inhibition of a potent monocyte recruitment factor, monocyte chemoattractant protein 1/chemokine (C-C) ligand 2 (MCP-1/CCL-2), results in reduced macrophage infiltration, angiogenesis, and tumor growth [8], but as CCL-2 also influences tumor and endothelial cells directly, the effects could be multiple.

To elucidate this further, and to mimic the natural tumor environment in patients, we have used a rat prostate tumor model where a small number of Dunning R-3327 AT-1 rat prostate tumor cells are injected into the ventral prostate of immunocompetent and syngenic rats [9]. The tumor cells rapidly establish tumors located within a normal prostate and host environment. Our previous study showed that vascular density increase with tumor growth both in the tumor and importantly also in the surrounding normal prostate [9]. This is of significance because extratumoral vessels secure blood supply to and drainage from the tumor and manipulation of these vessels impaired tumor growth [9,10]. To reduce macrophage infiltration to the tumor and its surroundings, we now used intraperitoneal injections of clodronate (dichloromethylene-bisphosphonate)-encapsulated liposomes designed to target phagocytic cells. After internalization and release of free clodronate, macrophages die through apoptosis [11]. Tumors from clodronate-liposome-treated rats were compared with controls with regard to tumor weight, intratumoral and extratumoral macrophage content, and vascularity. In addition, the angiogenic and chemotactic profiles of the AT-1 orthotopic tumor were examined.

Here, we show the importance of macrophages to stimulate extratumoral vascular growth and thereby create a microenvironment necessary for subsequent tumor growth. Furthermore, we demonstrate that the tumor expresses multiple factors that could stimulate monocyte recruitment and angiogenesis and could induce tissue remodeling. Our findings identify TAMs and several chemotactic and angiogenic factors as potential therapeutic targets for prostate cancer.

Materials and Methods

Orthotopic Implantation of Dunning R-3327 Rat AT-1 Tumor Cells

Rat prostate AT-1 tumor cells (kindly donated by Dr J.T. Isaacs, John Hopkins, Baltimore, MD) were grown in culture as previously described [12].

For morphologic analysis, AT-1 cells (2 x 10³ cells in 50 µl of RPMI 1640) were carefully injected into one lobe of the ventral prostate of adult Copenhagen rats (Charles River, Sulzfeld, Germany) as previously described [9]. Rats were killed at 7 (n = 6), 10 (n = 16), and 14 days (n = 11) after tumor cell injection. At sacrifice, the animals were injected with bromodeoxyuridine (BrdU, 50 mg/kg; Sigma-Aldrich, Oslo, Norway), and the prostate tissue was removed, weighed, and prepared as described earlier [9,13].

For RNA preparation, AT-1 cells were injected in the same way (n = 11). Animals injected with vehicle (50 µl of RPMI, n = 10) were used as controls. After 10 days, the animals were killed and the VP lobes were quickly removed, frozen in liquid nitrogen, and stored at −80°C. All of the animal work was approved by the local ethical committee for animal research.

Factor VIII, CD68, and BrdU

Sections, 5-µm thick, were immunostained using primary antibodies against CD68 (AbD Serotec, Oxford, UK), factor VIII (Dako, Stockholm, Sweden), and BrdU (Dako) as described earlier [13–15]. The percentage of factor VIII-stained blood vessels, CD68-positive macrophages, and BrdU-positive proliferating tumor cells and the number of BrdU-labeled endothelial cells per vascular profile were measured as described earlier [13,15]. Moreover, sections were doublestained with antibodies against BrdU and CD68. Sections were first incubated with the primary antibody for BrdU overnight, followed by a 30-minute incubation with secondary antibody Envision AP Mouse (Dako). The sections were developed using Permanent Red (Dako). The following day, the same sections were incubated with an antibody against CD68, followed by secondary antibody Envision HRP Mouse for 30 minutes. The slides were then developed using diaminobenzidine (Dako).

Apoptosis

Apoptotic tumor cells were identified using a TUNEL assay according to the protocol provided by the manufacturer (Roche Diagnostics, Bromma, Sweden). The percentage of apoptotic cells was determined in 2000 cells of each tumor examined.

Interleukin 1β and Matrix Metalloproteinase 9 Immunohistochemistry

After microwave heating of paraffin sections in 0.01 M citrate buffer pH 6.0 or target 1 antigen retrieval solution (Dako), respectively, the sections were incubated with antibodies against interleukin 1β (IL-1β diluted at 1:1000; Catalog no. AF-501-NA; R&D Systems, Minneapolis, MN) or matrix metalloproteinase 9 (MMP-9 diluted at 1:100; Catalog no. Sc-6840; Santa Cruz Biotechnology, Santa Cruz, CA) overnight. The sections were processed using a Vectastain, Elite ABC kit (Vector, Burlingame, CA) or an LSAB kit (Dako), respectively.

Stereology

The volume density of tumor tissue was determined on hematoxylineosin-stained sections as previously described [9,13,15]. Total tumor weight was then estimated by multiplying the volume density with prostate weight.

Targeting TAMs Using Clodronate-Liposomes

Macrophage infiltration to the tumor was suppressed by eliminating circulating phagocytic cells by intraperitoneal (i.p.) injections of dichloromethylene-bisphosphonate (clodronate) liposomes. clodronate was a gift from Roche Diagnostics GmbH (Mannheim, Germany) and was incorporated into liposomes as previously described [11]. Clodronate-liposomes (1 ml/100-g body weight, n = 8) were administered every second day starting 4 days before AT-1 tumor cell injection. Equal phosphate-buffered saline (PBS) injections (n = 8) were used as appropriate controls in recommendations [11]. AT-1 tumor cells were injected into the ventral prostate of Copenhagen rats as described; animals were killed 10 days later and tissues were prepared for morphologic analysis as described.

RNA Preparation, RT² Profiler PCR Arrays, and Real-time Quantitative Reverse Transcription-Polymerase Chain Reaction

RNA, from tumors, ventral prostate tissues, and AT-1 cells, was extracted using the TRIzol method (Invitrogen, Stockholm, Sweden) following manufacturer's instructions.

For polymerase chain reaction (PCR) array studies, AT-1 tumor tissue (n = 11) was compared with normal ventral prostate from sham-injected animals (n = 10) and with AT-1 cells in vitro (n = 3 different cell batches). Total RNA, isolated from the individual animals in each group, was pooled together andDNase-treated (DNAse I; Sigma-Aldrich) to remove contaminating DNA. Complementary DNA was synthesized on 1 µg of total DNase-treated RNA using RT² Profiler PCR array first strand kit C-02 (SABiosciences, by way of MedProbe, Lund, Sweden) according to protocol. RT² Profiler PCR arrays, rat angiogenesis (Catalog no. APRN-024A; SABiosciences), and rat chemokines (Catalog no. PARN-022A; SABiosciences), were performed according to the manufacturer's instructions using the ABI Prism 7900 HT instrument (Applied Biosystems, Foster City, CA) and ABI Prism 7900 SDS software 2.1. Acquired data were analyzed with PCR array data analysis template downloaded from the SuperArray Web site (www.sabioscience.com) and normalized to the expression level of housekeeping control genes. The specificity of the SYBR Green assay was confirmed by melting point analysis.

Quantitative reverse transcription-PCR (qRT-PCR) was used to verify data from the PCR array. Total RNA from individual animal or AT-1 cells was reverse-transcribed using Superscript II (Invitrogen) in a 10-µl reaction according to protocol.

Quantification of mRNA levels was performed by real-time qRT-PCR using the LightCycler SYBR Green I technology (Roche Diagnostics, Bromma, Sweden). Reverse transcription-PCR was performed in a 10-µl reaction volume, using 0.25 µM primers, 2 to 4 mM MgCl₂, and 2 µl of complementary DNA according to protocol.

The PCR conditions were optimized for each set of primers (Table 1), and melting curve analysis was performed to confirm specificity. Negative controls were run in parallel. Data were analyzed using Light-Cycler analysis software 3.5.3 (Roche Diagnostics, Bromma, Sweden).

Table 1.

Oligonucleotides Used as Primers in the Real-time qRT-PCR.

Primer	Sequence (5′-3′)	Product Size (bp)	Annealing Temperature (°C)	Accession No.
ANG-1	Sense - GTC GGA GAT GGC CCA G	113	65	NM_053546
	Antisense - CTG TGA GCT TTC TGG TC
ANG-2	Sense - CAG CCA ACC AGG AAG TGA TT	76	62	XM_344544
	Antisense - GAG CAT CTG GGA ACA CTT GCA
CCL-2	Sense - CTC TTC CTC CAC CAC T	149	60	NM_031530
	Antisense - TTG GGA TCA TCT TGC CAG
CSF-1	Sense - CCC TCG GAC ATT GGA T	161	60	NM_023981
	Antisense - GGT AGT GGT GGA CGT T
CXCL1	Sense - GAG GCT TGC CTT GAC C	175	60	NM_030845
	Antisense - AGG ACC CTC AAT AGA AAT CG
CXCL9	Sense - ATC CCC TAG ACG GTT G	147	60	NM_145672
	Antisense - AGT GAC GTG TTT ACT CTG
EGF	Sense - CCG TCG TAC GAT GGG T	159	60	NM_012842
	Antisense - CCT CTG CCC GTA GTC A
EREG	Sense - CCG GTT CCT ACT CAG C	167	60	NM_021689
	Antisense - TGA CGG AAT CAC GGT T
FGF-2	Sense - CAC TTA CCG GTC ACG GAA AT	80	60	NM_019305
	Antisense - CCG TTT TGG ACT CGA GTT TA
Interferon γ	Sense - AGG ACG GTA ACA CGA AA	117	60	NM_138880
	Antisense - GGT GCG ATT CGA TGA C
IL-1β	Sense - AAA AAT GCC TCG TGC TGT CT	127	60	NM_031512
	Antisense - GGG ATT TTG TCG TTG CTT GT
IL-6	Sense - GGA GTG GCT AAG GAC C	164	60	NM_012589
	Antisense - AGA TAC CCA TCG ACA GG
MMP-3	Sense - CTT CGA TGC AGT CAG C	214	60	NM_133523
	Antisense - ATG ACC TCG GAT AGC C
MMP-9	Sense - TTC GAC GCT GAC AAG AAG TG	156	60	NM_031055
	Antisense - AGG GGA GTC CTC GTG GTA GT
PTGS1	Sense - GAC TAC GGT GTC GAG	131	60	NM_017043
	Antisense - CGC ATT TCT CGG GAC T
Serpinb5	Sense - ATT CCA TCG AGG TGC C	186	60	NM_057108
	Antisense - AGA GCC CTG CAA CTA C
TGFα1	Sense - GGC TAC CAT GCC AAC TTC TGC CT	199	67	X52498
	Antisense - TGT TGG ACA GCT GCT CCA CCT TG
TIMP-3	Sense - GGC CTC AAT TAC CGC T	157	60	NM_012886
	Antisense - GGA TGC AGG CGT AGT G
VEGF-A	Sense - CGA AGT GGT GAA GTT CAT GGA TGT	74	60	M32977
	Antisense - TGG AAG ATG TCC ACC AGG GTC

Hypoxia Treatment of AT-1 Cells In Vitro

AT-1 cells were grown at 37°C for 6 and 24 hours in a hypoxic incubator (1% O₂, 5% CO₂, 94% N₂; Billups-Rothenberg, San Diego, CA) or in normoxia (21% O₂, 5% CO₂, 74% N₂). Total RNA was prepared as described above, and hypoxic AT-1 cells were compared with normoxic controls using the PCR arrays described.

Statistical Analysis

Values are presented as mean ± SD. Mann-Whitney U test was used for comparison between groups. The Spearman's rho test was used for the correlation studies. P < .05 was considered significant. Statistical analysis was performed using the statistical software SPSS 14.0 (SPSS Inc., Chicago, IL).

Results

Macrophage Density and Vascular Proliferation in the AT-1 Rat Prostate Tumor Model

When a small number of AT-1 tumor cells were injected to one of the ventral prostate lobes of syngenic and immunocompetent rats, it resulted in rapid tumor growth. The average tumor diameter was 2, 5, and 9 mm at days 7, 10, and 14, respectively. In other words, the tumor diameter grew with 1 mm/day, and the tumor-invasive front expanded with 0.5 mm/day into the surrounding normal prostate tissue. Blood vessels, stroma cells, and the extracellular matrix in the normal tissue adjacent to the expanding tumor thus lay within the tumor the following day.

We have previously shown that vascular density increase with tumor growth in the tumor and also in the surrounding normal prostate tissue [9]. To understand the mechanisms behind the increased vascularity and to study if there was an association with macrophages, we analyzed endothelial cell proliferation (BrdU incorporation into endothelial cells) and monocyte/macrophage content (CD68-positive cells) in the tumor and in the extratumoral normal prostate tissue.

The tumor contained a high number of CD68-positive macrophages particularly at the invasive zone, here defined as the 0.5-mm tumor surface region and the 0.5-mm-wide zone of normal tissue most adjacent to the tumor (Figure 1A). Both average intratumoral macrophage density and endothelial cell proliferation were higher than in the extratumoral normal tissue (Table 2). Vascular BrdU labeling index in the tumor correlated with tumor size (r_s = 0.62, P < .01; Table 2). Conversely, the average intratumoral macrophage density inversely correlated to both tumor size (r_s = -0.57, P < .05) and intratumoral vascular density (r_s = -0.54, P < .05; Table 2).

Figure 1.

(A) Section from the ventral prostate in an animal injected with AT-1 cells 10 days earlier stained for macrophages (CD68, brown; original magnification, x40), microvessels and a larger vessel stained for BrdU (brown; original magnification, x400), and a small venule double-stained for proliferation (BrdU, red) and macrophages (CD68, brown) (original magnification, x400). Note the intense staining of CD68 in the invasive zone and that BrdU labeling is present in endothelial and in vascular smooth muscle cells. Note also the close contact between proliferating vascular cells and macrophages. (B) Tumor weight, macrophage densities (percentage of CD68-positive cells), vascular densities (percentage of factor VIII-positive cells), and vascular proliferation (BrdU labeling index) were significantly inhibited in clodronate-liposome-treated rats (n = 8) compared with PBS controls (n = 8). Values are means ± SD. *P < .05. **P < .01. (C) Tumors from clodronate-liposome-treated rats and PBS controls stained for tumor apoptosis (TUNEL, brown, arrowheads; original magnification, x400), macrophages (CD68, brown; original magnification, x40), blood vessels (factor VIII, red; original magnification, x100), and proliferation (BrdU, red; original magnification, x200). Blood vessels (arrowheads) showed decreased BrdU staining in clodronate-liposome-treated animals compared with controls, whereas no difference in tumor proliferation was noted. (D) AT-1 tumor-invasive zone stained for MMP-9 (brown; original magnification, x200) and IL-1β (brown; original magnification, x400).

Table 2.

Quantitative Analysis of Intratumoral and Extratumoral Vascular Density, Vascular Proliferation, and Macrophage Density.

Days	Vascular Density Factor VIII (%)	Vascular Proliferation BrdU (Labeling Index)		Macrophage Density CD68 (%)
		Total	Large Vessels	Total	Invasive Zone
Intratumoral
7	2.5 (0.29)	25.6 (10.5)	ND	5.0 (0.67)	8.1 (1.0)
10	3.6 (0.82)*	35.9 (12.0)	ND	4.8 (0.62)	8.0 (1.5)
14	4.6 (1.49)*	46.4 (10.8)†	ND	3.5 (1.40)	8.9 (1.8)
Extratumoral
7	1.2 (0.27)	5.3 (3.5)	2.5 (6.1)	0.62 (0.18)	3.7 (0.79)
10	1.8 (0.32)*	21.2 (7.5)*	7.1 (6.1)	0.74 (0.16)	3.7 (1.20)
14	2.8 (0.85)*	26.8 (6.3)*	17.1 (4.9)*	1.40 (0.24)*	4.1 (0.94)

Groups of 5 to 16 rats at each time point.

Values are presented as means (SD).

ND indicates not determined.

^*P < .01 when compared with day 7.

^†P < .05 when compared with day 7.

In the extratumoral normal prostate tissue, the fraction of BrdU-positive endothelial cells also correlated with extratumoral vessel density (r_s = 0.69, P < .01) demonstrating a production of new blood vessels (Table 2). Proliferation of endothelial cells in the normal prostate tissue also correlated with tumor size (r_s = 0.66, P < .01). Interestingly, opposite to macrophage content in the tumor, average extratumoral macrophage density correlated with tumor size (r_s = 0.84, P < .01), extratumoral vascular density (r_s = 0.52, P < .05), and extratumoral vascular proliferation (r_s = 0.89, P < .01). Double staining of macrophages (CD68) and endothelial proliferation (BrdU) in the extratumoral normal prostate tissue revealed that macrophages often were in close contact with proliferating endothelial cells in capillaries and with vascular smooth muscle cells in arterioles and veins (Figure 1A). We therefore quantified BrdU labeling index in the smooth muscle cell layer in arterioles and venules. Proliferation in such vessels correlated to both tumor size (r_s = 0.86, P < .01) and extratumoral macrophage density (r_s = 0.76, P < .01; Table 2). This shows that vascular growth was not restricted to capillaries and that macrophages probably are important for growth of all types of extratumoral blood vessels.

As macrophage density was particularly high at the invasive zone and low in the central tumor parts, we specifically quantified macrophage density in the 0.5-mm tumor-invasive region and found that it remained high and constant during the period examined (Table 2). The macrophage-rich border zone subsequently increased with tumor growth and progressed into the normal prostate. Macrophage density in the 0.5-mm invasive zone of normal prostate tissue was also particularly high already at 7 days and then remained high during tumor growth (Table 2). In addition, we also noted that the invasive zone had the highest vascular BrdU labeling index (data not shown). The majorities of blood vessels were thus produced and enlarged in the macrophage-rich border zone and then subsequently incorporated in the expanding tumor.

Reduction of Macrophages with Clodronate-Liposomes Inhibited Tumor Growth, Angiogenesis, and Arteriogenesis

To determine the importance of macrophages for AT-1 prostate tumor growth, rats were depleted of circulating phagocytic cells by administration of clodronate-liposomes. Peritoneal injections before and during tumor growth were administrated to reduce the recruitment of macrophages to the injected tumor cells.

In clodronate-liposome-treated rats, AT-1 tumor size was 71% smaller (P < .01) compared with untreated controls (Figure 1B). Tumor cell apoptosis (TUNEL), which was similar in the invasive zone and more central tumor areas, was significantly increased in the clodronate-liposome-treated animals compared with controls (0.6 ± 0.2% vs 0.3 ± 0.2%, P < .01; Figure 1C). Increased apoptosis was seen both in central and in peripheral parts of the tumors. There was, however, no increase in apoptosis in the normal ventral prostate tissue either close to or more remote from the tumor (data not shown). Tumor cell proliferation was unaffected by clodronate-liposome treatment (data not shown).

To verify macrophage reduction, macrophage density (CD68) was analyzed in tumors and extratumoral prostate tissues at day 10 after tumor implantation. In clodronate-liposome-treated animals, macrophage density was 40% lower (P < .05) in the tumor and 35% lower (P < .05) in the extratumoral prostate tissue compared with controls (Figure 1, B and C). The liver has previously been shown to be depleted of macrophages by i.v. injections of clodronate-liposomes [16], and here macrophage density was decreased with 62% (P < .01).

To investigate whether macrophage reduction affected vascularization, tumor and extratumoral tissues were analyzed for vascular density (factor VIII expression) and vascular proliferation (BrdU-positive endothelial cells). Both vascular density (Figure 1, B and C) and endothelial cell proliferation (Figure 1, B and C) were significantly inhibited in the tumor and in adjacent prostate tissue in clodronate-liposome-treated animals compared with controls. In addition, macrophage reduction resulted in suppressed proliferation of vascular smooth muscle cells (BrdU labeling index) in large vessels compared with controls (3.0 ± 2.3 vs 8.4 ± 2.3, P < .01).

Characterization of Angiogenic Factors and Chemokines Expressed by the Orthotopic AT-1 Tumor

To investigate what factors the AT-1 tumor express, which could stimulate angiogenesis and monocyte/macrophage recruitment, we used a rat angiogenesis and a rat chemokine PCR array. The orthotopic tumor in vivo was compared with normal prostate control tissue and with AT-1 tumor cells in vitro.

The results showed that the tumor produced numerous angiogenic factors: 28 of 75 examined factors were upregulated and 17 were downregulated in the tumor tissue compared with normal prostate controls (Table 3). In addition, 18 factors were verified with qRT-PCR and were all significantly changed compared with normal prostate tissue (Table 3). Proangiogenic factors, such as fibroblast growth factor 2 (FGF-2), angiopoietins 1 and 2 (ANG-1 and ANG-2), transforming growth factor β (TGFβ), and vascular endothelial growth factor (VEGF), were mainly elevated, whereas antiangiogenic factors, such as Serine proteinase inhibitor b5 (Serpin-b5; also called Maspin), tissue inhibitor of metalloproteinase 3 (TIMP-3), thrombospondin 4, and brain-specific angiogenesis inhibitor 1 were decreased. Matrix metalloproteinases 3, 9, and 19, central for extracellular matrix degradation, were also highly increased in the tumor (Table 3).

Table 3.

Angiogenic Factors Altered in the Orthotopic AT-1 Rat Prostate Tumor Compared with Normal Rat Prostate Tissue and with AT-1 Tumor Cells In Vitro Analyzed with PCR array and qRT-PCR.

Accession No.	Symbol	Description	Fold Up- or Down-regulation
			Superarray	qRT-PCR
Angiogenic Factors Upregulated in the Orthotopic AT-1 Tumor Compared with Normal Control Tissue
XM_344544	Ang-2	Angiopoietin 2	6.2	4.1 (1.0)*
NM_053546	Ang-1	Angiopoietin 1	14.9	9.4 (3.1)*
NM_031530	CCL2/MCP-1	Chemokine (C-C) ligand 2	15.6	7.1 (2.7)*
XM_241632	Col18a1	Collagen XVIII alpha 1	32.3
NM_022266	CTGRP	Connetive tissue growth factor	4.7
NM_030845	CXCL1/Gro1	Chemokine (C-X-C) ligand 1	51.6	4.4 (1.8)*
NM_145672	CXCL9/Mig	Chemokine (C-X-C) ligand 9	8.4	5.6 (7.8)*
XM_234903	Efna2	Ephrin A2	3.4
NM_021689	Ereg	Epiregulin	164.5	3.3 (0.4)*
NM_019305	Fgf-2/bFGF	Fibroblast growth factor 2	23.1	5.3 (1.1)*
NM_019143	Fn-1	Fibronectin 1	32.8
NM_024359	Hif1α	Hypoxia inducible factor 1 alpha	3.3
NM_012797	ID125A	Inhibitor of DNA binding 1	3.3
NM_138880	IFNγ	Interferon gamma	6.1	4.5 (1.8)*
NM_031512	IL1β	Interleukin 1 beta	24.0	13.8 (3.2)*
NM_012589	IL6	Interleukin 6	82.3	4.2 (1.2)*
XM_235707	Itga5	Integrin alpha 5	4.4
XM_222317	Mmp19	Matrix metalloproteinase 19	22.2
NM_133523	Mmp3	Matrix metalloproteinase 3	545.0	19.5 (7.1)*
NM_031055	Mmp9	Matrix metalloproteinase 9	32.2	8.9 (2.3)*
NM_012613	Npr1	Natriuretic peptide receptor 1	3.1
NM_145098	Nrp1	Neuropilin 1	15.9
NM_030869	Nrp2	Neuropilin 2	11.4
NM_013085	UPAM	Plasminogen activator, urokinase	11.2
NM_017043	Ptgs1/Cox1	Prostaglandin endoperoxide synthase 1	23.8	547.3 (509.3)*
NM_021578	TGFβ	Transforming growth factor beta 1	22.5	18.5 (3.2)*
NM_053819	Timp-1	Tissue inhibitor of metalloproteinase 1	5.1
NM_031836	VEGF	Vascular endothelial growth factor A	3.7	5.5 (1.4)*
Angiogenic Factors Downregulated in the Orthotopic AT-1 Tumor Compared with Normal Control Tissue
NM_031012	APN	Alanyl aminopeptidase	-7.9
XM_343260	Bai1	Brain-specific angiogenesis inhibitor 1	-8.2
XM_343607	Col4a3	Collagen IV alpha 3	-21.3
NM_053903	Efna5/Lerk7	Ephrin A5	-29.6
NM_012842	Egf	Epidermal growth factor	-106.2	-33.9 (14.2)*
NM_023090	Hif2α	Endothelial PAS domain protein 1	-3.5
NM_021867	Fgf16	Fibroblast growth factor 16	-5.0
NM_053429	Fgfr3	Fibroblast growth factor receptor 3	-29.1
NM_173838	Fzd5	Frizzled homolog 5	-3.9
NM_013076	OB/obese	Leptin	-10.2
NM_030859	Mdk	Midkine	-4.8
NM_031054	Mmp2	Matrix metalloproteinase 2	-4.1
XM_574314	Plg/Ab1-346	Plasminogen	-26.9
NM_057108	Serpinb5	Serine proteinase inhibitor 5	-98.3	-9.3 (4.1)*
XM_220811	Tbx4	T-box 4	-11.5
XM_342172	Thbs4	Thrombospondin 4	-4.2
NM_012886	Timp3	Tissue inhibitor of metalloproteinase 3	-6.1	-2.9 (1.5)*
Angiogenic Factors Upregulated in the AT-1 Tumor In Vivo Compared with the AT-1 Cells In Vitro
XM_344544	Ang-2	Angiopoietin 2	22.6	76.6 (17.9)†
NM_031530	CCL2/MCP-1	Chemokine (C-C) ligand 2	55.6	16.7 (6.3) †
XM_241632	Col18a1	Collagen XVIII alpha 1	5.9
NM_145672	CXCL9/Mig	Chemokine (C-X-C) ligand 9	98.8	525.8 (729.8)†
XM_234903	Efna2	Ephrin A2	26.5
NM_019305	Fgf-2/bFGF	Fibroblast growth factor 2	9.2	5.3 (1.1)†
NM_138880	IFNγ	Interferon gamma	6.1	419.9 (164.4)†
NM_031512	IL1β	Interleukin 1 beta	501.5	93.2 (21.9)†
NM_031055	Mmp9	Matrix metalloproteinase 9	37.5	17.8 (4.7)†
NM_012613	Npr1	Natriuretic peptide receptor 1	92.8
NM_145098	Nrp1	Neuropilin 1	64.4
NM_021578	TGFβ	Transforming growth factor beta 1	7.3	2.9 (0.5)†

For Superarray comparisons, AT-1 tumors (n = 11) were pooled into one sample and compared with pooled normal prostate tissue (n = 10). Quantitative RT-PCR was performed on tumor (n = 11) or normal prostate tissue (n = 10) from individual animals.

Values are presented as mean (SD).

^*P < .01.

^†P < .05.

To determine whether some of these factors could be produced by macrophages, we compared the orthotopic AT-1 tumor at day 10 with AT-1 cells in vitro. Of the 28 angiogenesis factors upregulated compared with normal prostate control tissue, 12 were highly increased in vivo, suggesting that macrophages or other nonmalignant cells could contribute to their production (Table 3). Several of these factors have been shown to be produced by macrophages (for reviews, see [5,17]), and staining of MMP-9 and IL-1β in the tumors showed that both these proteins were located in macrophage-like cells, particularly at the invasive border. Tumor cells, however, were unstained (Figure 1D).

The most important chemokines for monocyte recruitment to tumors seem to be CCL-2, colony-stimulating factor 1 (CSF-1), and VEGF (for review, see [1]). The chemokine array revealed that the AT-1 tumor expressed both CCL-2 and CSF-1 and several other chemokines in addition (Table 4). mRNA expressions of CCL-2, CSF-1, and VEGF were also verified with qRT-PCR and were all significantly increased in vivo compared with normal prostate control tissue (Figure 2). Both CSF-1 and VEGF were highly expressed by AT-1 tumor cells in vitro compared with normal prostate control tissue, whereas CCL-2 had low expression in vitro (Figure 2). Vascular endothelial growth factor and CCL-2 had significantly increased mRNA expressions in the AT-1 tumor in vivo compared with AT-1 cells in vitro.

Table 4.

Chemokines Altered in the Orthotopic AT-1 Rat Prostate Tumor Compared with Normal Rat Prostate Tissue Analyzed with PCR array.

Accession No.	Symbol	Description	Fold Up- or Down-regulation
Chemokines Upregulated in the Orthotopic AT-1 Tumor Compared with Normal Control Tissue
NM_012513	Bdnf	Brain-derived neurotrophic factor	12.2
NM_053303	Blr1/NLR	Burkitt lymphoma receptor 1	3.3
XM_213425	Ccl12	Chemokine (C-C) ligand 12	8.1
NM_031530	Ccl2	Chemokine (C-C) ligand 2	4.1
NM_020542	Ccr1	Chemokine (C-C) receptor 1	3.3
XM_226200	Cmtm3/Cklfsf3	CKLF-like MARVEL	4.1
NM_023981	Csf1	Colony-stimulating factor 1	3.5
NM_133534	Cx3cr1	Chemokine (C-X3-C) receptor 1	6.3
NM_030845	Cxcl1/Gro1	Chemokine (C-X-C) ligand 1	7.9
NM_053647	Cxcl2/Mip-2	Chemokine (C-X-C) ligand 2	3.7
NM_022214	Cxcl5	Chemokine (C-X-C) ligand 5	18.5
NM_153721	Cxcl7/Nap-2	Chemokine (C-X-C) ligand 7	7.5
NM_022205	Cxcr4	Chemokine (C-X-C) receptor 4	6.8
NM_0010037	Grp77/C5L2	G protein-coupled receptor 77	5.9
NM_019165	IL18	Interleukin 18	4.5
NM_017183	IL8rβ	Interleukin 8 receptor beta	4.5
XM_344130	Lnhβb	Inhibin beta-B	3.1
NM_019340	Rgs3	Regulator of G-protein signaling 3	3.3
Chemokines Downregulated in the Orthotopic AT-1 Tumor Compared with Normal Control Tissue
NM_013107	Bmp6	Bone morphogenetic protein 6	-4.3
XM_342421	C5	Similar to complement 5	-3.2
NM_078621	Ccbp2/D6	Chemokine binding protein 2	-8.3
XM_342824	Ccl19	Chemokine (C-C) ligand 19	-3.3
NM_019233	Ccl20	Chemokine (C-C) ligand 20	-3.7
XM_236658	Ccrl2	Chemokine (C-C) receptor-like 2	-6.7
NM_053352	Cmkor1	Chemokine orphan receptor 1	-4.0
NM_134455	Cx3cl1	Chemokine (C-X3-C) ligand 1	-36.5
NM_053415	Cxcr3	Chemokine (C-X-C) receptor 3	-12.4
NM_012590	Inha	Inhibin alpha	-3.1
NM_031054	Mmp2	Matrix metalloproteinase 2	-5.7
NM_033098	Tapbp/Tapasin	TAP binding protein	-4.4

For Superarray comparisons, AT-1 tumors (n = 11) were pooled into one sample and compared with pooled normal prostate tissue (n = 10).

Figure 2.

Monocyte chemoattractant protein 1 (CCL-2), CSF-1, and VEGF mRNA expressions in AT-1 prostate tumor cells in vitro (n = 3 different batches) and AT-1 prostate tumors in vivo (n = 11) quantified with qRT-PCR and expressed in relation to levels in normal rat prostate tissue (n = 10). Values are means ± SD: *P < .05, **P < .01.

Various studies have shown that hypoxia can upregulate both angiogenic factors and chemokines (for review, see [18]). Because large parts of our AT-1 tumors are hypoxic [9], we examined whether hypoxia could explain why some factors are increased in AT-1 tumors in vivo compared with AT-1 cells in vitro. To test this, AT-1 cells in vitro were incubated up to 24 hours in hypoxia and compared with normoxic controls using the same arrays. The results showed no major differences in angiogenic or chemotactic profiles between hypoxic and normoxic AT-1 cells in vitro (data not shown) compared with the major changes seen when comparing AT-1 tumors in vivo versus AT-1 cells in vitro. For example, CSF-1, VEGF, ANG-2, CCL-2, IL-1β, TGFβ, and MMP-9 expressions were only marginally affected, and FGF-2 decreased by hypoxia in vitro.

Discussion

In this study, we observed that macrophages were present in high numbers in rat prostate tumors, particularly at the invasive zone but importantly also in the extratumoral normal prostate tissue. In addition, we showed that the tumor expressed numerous factors that stimulate monocyte recruitment, angiogenesis, and tissue remodeling. Furthermore, a reduction of monocytes/macrophages to the tumor and its surroundings with clodronate-liposomes resulted in inhibited tumor growth and repressed intratumoral and extratumoral angiogenesis and arteriogenesis.

Studies of macrophage infiltration to human prostate cancer have shown diverse results with both positive and negative associations to clinical outcome and cancer progression [6,7]. Interestingly, TAMs in different tumor compartments apparently have opposing effects on prostate cancer progression [6,7]. Two different polarization states have been described for macrophages: the M1 phenotype has antitumoral effects, whereas the M2 macrophage promotes angiogenesis, tumor growth, and metastasis [19]. Tumor-associated macrophages generally exhibit an M2-like phenotype, although a mixed M1 and M2 phenotype is also described [19]. In this prostate cancer model, macrophages in the extratumoral tissue seemed particularly important for vessel growth, whereas the role of macrophages inside the tumor was more unclear.We noted that both macrophage density and vascular proliferation were highest in the tumor-invasive zone where macrophages could promote invasion, tissue remodeling, and angiogenesis. Already existing vessels in the extratumoral tissue may also promote tumor vascularization through co-option, a process where existing vessels are surrounded by tumor cells and used to supply the tumor [20]. Furthermore, macrophages infiltrating the extratumoral tissue were often seen in close contact with proliferating endothelial cells in capillaries and with smooth muscle cells in arterioles and venules. Reduced macrophage infiltration resulted in repressed proliferation in capillaries and in large vessels, and macrophages therefore seem to have a central role in both angiogenesis and arteriogenesis in this tumor model. Almost all studies on TAMs explore their roles within tumors, and their roles in the surrounding nonmalignant tissue and particularly for tumor- related arteriogenesis have surprisingly not been examined [17]. However, as macrophages play an important role for the arteriogenesis and collateral artery growth in ischemia [21,22], we suggest that they are probably also important for the expansion of larger vessels in the normal tissue surrounding tumors. Previous studies have shown that inhibiting the macrophage chemoattractant CCL-2 in a subcutaneous prostate xenograft model in nude mice reduced macrophage influx, intratumoral angiogenesis, and tumor growth [8]. In a subcutaneous rat prostate tumor model, treatment with an antiangiogenic agent also inhibited macrophage function and impaired tumor growth [23]. Agents used in the earlier studies have, beside the effects onmacrophages, direct effects on both tumor and endothelial cells. To reduce macrophage infiltration, we used intraperitoneal injections of clodronate-liposomes, which are phagocytosed by, and induce apoptosis in circulating phagocytic cells [11]. This method has been used to deplete macrophages and thereby to reduce angiogenesis and tumor growth in a variety of tumor models [24–26]. In some models such as glioma, depletion of macrophages, however, resulted in increased tumor growth, although a small reduction in vessel density was observed [27]. Because the concentration of free clodronate in the prostate is unknown in this experimental setup, some of the effects seen might be due to direct effects of clodronate on tumor and endothelial cells. Direct effects are, however, most likely minor because clodronate-liposomes cannot cross-vascular barriers, and free clodronate released from dead phagocytic cells have a very short half-life in the circulation [11]. Monocytes/macrophages that have phagocytosed clodronate-liposomes could also be inhibited in their normal functions. Circulating monocytes have been shown to be depleted, but as new monocytes are constantly entering the circulation from the bonemarrow, a continuous supply of clodronate-liposomes is needed to inhibit macrophage recruitment [28]. This probably explains why tumors from clodronate-liposome-treated animals contained CD68-positive cells at day 10 (2 days after the last liposome injection). Notably, although the reduction in macrophage densities in tumor and extratumoral tissues was modest, we observed a significant inhibition in tumor growth, suggesting that only a moderate decline in macrophage infiltration is sufficient. Together, our report and a previous report [8,23] strongly suggest the importance of macrophages for prostate cancer growth and angiogenesis. The specific effects of macrophage depletion during various phases of prostate tumor development and metastasis, however, need to be examined further.

Prostate tumor growth is angiogenesis-dependent, and the orthotopic rat prostate tumors expressed numerous factors important for angiogenesis. Several of these factors are apparently produced by tumor cells, whereas others are produced by host cells. Some of the factors produced, for example, VEGF, TGFβ, and ANG-2, have already been reported to be of major significance for angiogenesis in prostate cancer [29,30]. Macrophages could be the source of many of the angiogenic factors because expressions of these factors, for example, CCL-2, IL-1β, FGF-2, TGFβ, and MMP-9, were highly increased in the tumor in vivo compared with the tumor cells in vitro. To support this, MMP-9 and IL-1β staining was found in macrophages in the invasive zone. Matrix metalloproteinases, and MMP-9 in particular, are important for tissue remodeling and facilitate tumor growth, migration, invasion, and angiogenesis (for review, see [31,32]). Matrix metalloproteinase 9 could thus be a key factor produced by macrophages in our tumor model. Interleukin 1β has been shown to alter androgen receptor function [33] and promote tumor invasiveness and angiogenesis in prostate, breast, and melanoma tumor models [34]. Macrophage accumulation in the normal tissue adjacent to tumors could thus explain why this tissue shows an altered early response to castration [9]. It is also possible that factors upregulated in tumors in vivo versus in vitro are produced by tumor cells. Hypoxia (chronic or intermittent) and other environmental factors in the prostate in vivo may alter their expression profile. We could, however, not observe any major differences between cells exposed to hypoxia for 24 hours and controls at the mRNA level. Although this finding does not exclude the possibility that the local environment may alter expression in tumor cells, it suggests that the most likely reason to the large differences in expression profiles between tumors in vivo and tumor cells in vitro is expression in host cells.

Our clodronate-liposome-treated animals (examined 2 days after last injection) were unfortunately not suitable to prove macrophage versus tumor cell expressions of these factors because the reduction in tumor macrophage content was incomplete.

If macrophages are important for prostate tumor growth, it is important to elucidate the factors responsible for the macrophage accumulation. Chemokine (C-C) ligand 2, CSF-1, and VEGF are all important for monocyte recruitment to tumors [1] and, together with other chemokines, were highly expressed by the AT-1 tumor in vivo. Colony-stimulating factor 1 and VEGF were also expressed in vitro suggesting that AT-1 cells attract macrophages by secreting these factors, and that macrophages arriving to the tumor enhance this process by secreting CCL-2 and VEGF. This suggests that inhibition of several chemotactic factors simultaneously might be needed to reduce macrophage infiltration.

In summary, macrophages accumulating in the normal tissue surrounding tumors could be one important source of angiogenic and arteriogenic factors in prostate cancer and decreasing macrophages could be used as an antivascular/antitumor treatment. Decreasing macrophages could also inhibit tissue remodeling and tumor invasion into the surrounding normal tissue by reducing the levels of MMP-9 and IL-1β. Nevertheless, combinatorial therapies directed against both tumor and nonmalignant cells and neutralizing many of the individual factors involved are probably needed to get a pronounced and sustained effect on tumor growth. Further studies are needed to test this hypothesis.

Abbreviations

BrdU

bromodeoxyuridine

CCL-2/MCP-1

chemokine (C-C) ligand 2/monocyte chemoattractant protein 1

CSF-1

colony-stimulating factor 1

FGF-2

fibroblast growth factor 2

IL-1β

interleukin 1β

MMP-9

matrix metalloproteinase 9

TGFβ

transforming growth factor β

TAMs

tumor-associated macrophages

VEGF

vascular endothelial growth factor