Angiogenesis with pericyte abnormalities in a transgenic model of prostate carcinoma

Authors

  • Michael G. Ozawa B.S.,

    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    Search for more papers by this author
    • The first two authors contributed equally to the current article.

  • Virginia J. Yao Ph.D.,

    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    Search for more papers by this author
    • The first two authors contributed equally to the current article.

  • Yvan H. Chanthery B.A.,

    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    Search for more papers by this author
  • Patricia Troncoso M.D.,

    1. Department of Pathology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Akiyoshi Uemura M.D., Ph.D.,

    1. Stem Cell Research Group, Riken Center for Developmental Biology, Chuo-ku, Kobe, Japan
    Search for more papers by this author
  • Amanda S. Varner Ph.D.,

    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    Search for more papers by this author
  • Ian M. Kasman,

    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    Search for more papers by this author
  • Renata Pasqualini Ph.D.,

    1. Department of Genitourinary Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Department of Cancer Biology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Wadih Arap M.D., Ph.D.,

    1. Department of Genitourinary Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Department of Cancer Biology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Donald M. McDonald M.D., Ph.D.

    Corresponding author
    1. Department of Anatomy, University of California at San Francisco, San Francisco, California
    2. Cardiovascular Research Institute and Comprehensive Cancer Center, University of California at San Francisco, San Francisco, California
    • Cardiovascular Research Institute and Comprehensive Cancer Center, Room S-1363, University of California, 513 Parnassus Avenue, San Francisco, CA 94143-0130
    Search for more papers by this author
    • Fax (415) 476-4845


Abstract

BACKGROUND

Previous studies of the TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) model vasculature suggest that, as tumors develop, vessels invade the glandular epithelium. However, changes in the vasculature are difficult to study in conventional thin tissue sections. The authors used a new approach to characterize morphologic and architectural changes of blood vessels and pericytes during tumor development in TRAMP mice.

METHODS

Eighty-micron cryostat sections of normal prostate and three histopathologic stages of TRAMP tumor sections, classified by epithelial cell E-cadherin immunoreactivity, were immunostained with vascular endothelial cell and pericyte receptor antibodies and evaluated by confocal microscopy.

RESULTS

In the normal mouse prostate, capillaries were most abundant in the fibromuscular tunica between the epithelium and smooth muscle of the ductules. In the prostatic intraepithelial neoplasia (PIN) stage, vessels accompanied epithelial cell protrusions into the ductule lumen but remained in the connective tissue at the basal side of the epithelium. Well differentiated tissues had extensive angiogenesis with five times the normal mean vascularity outside ductules. Vessels were of variable diameter, were associated with an increased number of pericytes, and some had endothelial sprouts. Angiogenic blood vessels from poorly differentiated adenocarcinomas were tortuous, variable in caliber, and lacked the normal hierarchy. Pericytes on these vessels had an abnormal phenotype manifested by α-smooth muscle actin expression and loose association with endothelial cells. Angiogenesis and loss of vascular hierarchy were also found in human prostate carcinoma.

CONCLUSIONS

Vascular abnormalities, which begin at the PIN stage and intensify in well differentiated and poorly differentiated tumors, may be useful readouts for early detection and treatment assessment in prostate carcinoma. Cancer 2005. © 2005 American Cancer Society.

Prostate carcinoma is the fourth most common cause of cancer-related deaths in developed countries.1 Diagnosis and scoring with the Gleason grading system based on hematoxylin and eosin (H & E) stained sections is the clinical standard.2, 3 However, there is need for more sensitive or informative assessment criteria. Changes in the vasculature may provide novel readouts. Unlike normal blood vessels, tumor blood vessels are leaky, tortuous, and variable in diameter.4 Also, pericytes of tumor vessels are irregularly shaped, loosely attached, and may extend processes to neighboring blood vessels.5, 6

Increased expression of proangiogenic factors such as vascular endothelial growth factor (VEGF), angiopoietin-1, epidermal growth factor, and basic fibroblast growth factor (FGF) has been implicated in the pathogenesis of prostate carcinoma.7–10 Indeed, proangiogenic factors stimulate the migration and proliferation of vascular endothelial cells to form new vascular networks.11 Platelet-derived growth factor (PDGF) signaling between endothelial cells and pericytes (mural cells) is believed to stabilize new blood vessels.12–14 Although Gleason scores have been combined with microvascular density in prostate carcinoma, with the belief that tumor growth and vascular proliferation are interdependent, this is a controversial prognostic index for disease recurrence.15–17

The TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) is a prostate carcinoma model in which the cellular characteristics of the vasculature in different histopathologic stages can be systematically studied.18 The androgen-regulated minimal rat probasin promoter of the TRAMP transgene activates the SV40 early genes to produce spontaneous prostate-specific tumors. Stages of tumor development in [C57BL/6 × FVB/n] F1 TRAMP mice have been established by histologic evaluation of H & E-stained sections.18–20 Prostatic intraepithelial neoplasia (PIN) occurs in the ventral and dorsolateral lobes from 8 weeks of age, the well differentiated (WD) stage appears by 12 weeks of age, and poorly differentiated (PD) adenocarcinomas arise in 24-week-old TRAMP animals. Evidence for angiogenesis is consistent with increased expression of angiogenic factors such as VEGF mRNA,21 FGF-2, and FGFR1.22 Conversely, microvessel density is reduced in TRAMP tumors after treatment with SU5416, a tyrosine kinase inhibitor,23, 24 and with adenoviral expression of soluble recombinant VEGFR-2-Fc in spontaneous murine and human prostate carcinoma xenograft tumors.25

Results from previous studies of the normal murine prostate suggest that most blood vessels are located in the loose connective tissue around ductules. In TRAMP tumors, the vasculature is believed to invade the ductule epithelium.22, 26 Activation of an angiogenic switch, involving the expression of proangiogenic factors, during the PIN stage leads to robust angiogenesis in the WD and PD stages.

In the current study, we sought to characterize architectural and cellular changes in blood vessels of the normal prostate that lead to vascular abnormalities in the PIN, WD, and PD stages of TRAMP tumor development. Using confocal microscopy to evaluate thick tissue sections stained by immunohistochemistry, we determined the location and morphology of blood vessels in relation to prostate ductule epithelial cells and changes in vascular endothelial cells and pericytes. Studies of the normal mouse prostate revealed an extensive vascular network exists within the stromal compartment between the epithelium and smooth muscle of each ductule. Stage-specific changes were found in the vasculature beginning at the PIN stage, and vascular abnormalities became progressively more severe in the WD and PD stages. Vasculature abnormalities were also found in human prostate carcinoma.

MATERIALS AND METHODS

Mouse and Human Prostate

C57BL/6 TRAMP mice heterozygous for the rat probasin-Tag transgene were obtained from Dr. N. M. Greenberg (Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA) and crossed to nontransgenic FVB/n female mice (Charles River, Wilmington, MA) to obtain [C57BL/6 × FVB/n] F1 transgenic and nontransgenic males. Animals were genotyped as described previously.27 All experimental procedures were approved by the University of California San Francisco Institutional Animal Care and Use Committee. Human prostate samples were obtained from the tumor bank of the Specialized Program of Research Excellence Program in Prostate Cancer at the University of Texas M. D. Anderson Cancer Center (MDACC; Houston, TX).

Preparation of Tissue Sections

Mice that had been anesthetized with Avertin® (2,2,2-tribromoethanol; 0.015–0.017 mg/g, injected intraperitoneally) (Sigma-Aldrich, St. Louis, MO)28 of approximately 11–25 weeks of age were fixed by systemic perfusion with 1% paraformaldehyde in phosphate-buffered saline (PBS) (pH of 7.4) for 3 minutes at 120 mm Hg. The inferior vena cava was cut as an outlet. Prostates were removed, incubated in fixative for 1–3 hours at 4 °C, infiltrated with 30% sucrose in PBS/0.01% thimerosol overnight at 4 °C, and frozen in optimal cutting temperature (O.C.T) compound. (Sakura, Torrance, CA) at −80 °C. Human prostate specimens were immersed in 4% paraformaldehyde in PBS for 2–3 hours at 4 °C, embedded in 30% sucrose in PBS containing 0.01% thimerosol overnight at 4 °C, and frozen in O.C.T.

Immunohistochemistry

All steps were performed at room temperature unless stated otherwise. Eighty-micron cryostat sections were air-dried, and then rinsed twice with PBS and once with PBS containing 0.3% Triton X-100 (PBST). Sections were blocked in 5% normal goat serum (Jackson ImmunoResearch, West Grove, PA) in PBST for 1 hour. Sections were incubated for 12–16 hours in PBST and 1% normal goat serum containing combinations of the following primary antibodies: Armenian hamster monoclonal anti-mouse CD31 (1:500; Chemicon, Temecula, CA), mouse monoclonal Cy3-conjugated alpha-smooth muscle actin (α-SMA; 1:2000, Sigma-Aldrich), rat monoclonal anti-mouse platelet–derived growth factor receptor-beta (PDGFR-β [1:2000; clone APB5]), and rat monoclonal anti-mouse E-cadherin (1:5000; Zymed, South San Francisco, CA). Sections were rinsed with PBST and incubated in 0.22-μm filtered PBST for 4–6 hours containing appropriate combinations of the following secondary antibodies: goat fluorescein isothiocyanate (FITC)-conjugated anti-Armenian hamster immunoglobulin G (IgG) (1:200; Jackson ImmunoResearch), goat Cy3-conjugated anti-rat IgG, or goat Cy5-conjugated anti-rat IgG (1:400; Jackson ImmunoResearch). Sixty-micron serial cryostat sections of normal and neoplastic human prostate were immunostained with mouse monoclonal anti-human CD34 (1:2000; Clone QBend 10, DakoCytomation, Carpenteria, CA) in PBST containing 1% normal goat serum for 12–14 hours. Sections were rinsed and incubated in 0.22-μm filtered PBST containing goat FITC-conjugated anti-mouse secondary antibodies (1:400; Jackson ImmunoResearch) for 3–4 hours. Rinsed sections were mounted in Vectashield (Vector Laboratories Inc., Burlingame, CA).

H & E Staining

Prostate specimens from perfusion-fixed TRAMP mice were embedded in paraffin, sectioned at 5 μm, and stained with H & E (Biopathology Sciences Medical, South San Francisco, CA). Five-micron serial frozen sections of normal human prostate and prostate carcinoma were stained with H & E in the Department of Pathology at MDACC.

Imaging

Digital bright-field images were acquired with either a Zeiss Axiophot microscope (Zeiss, Thornwood, NY) with Plan-Apochromat objectives fitted with a Sony 3CCD 755 camera (Sony Corp. New York, NY) and interfaced with Scion Image 1.62C software or a Nikon microscope (Nikon, Melville, NY) fitted with Plan-Apo objectives interfaced with Olympus image-capturing software. Fluorescence images were acquired with an Olympus IX70 inverted fluorescence microscope fitted with an Olympus camera using the Magnafire software (Olympus, Irving, TX). Confocal images were acquired with a Zeiss LSM 510 laser scanning confocal microscope with krypton-argon and helium-neon lasers and analyzed with the LSM 510 software (version 3.2 (Carl Zeiss, Jena, Germany).

Blood Vessel Area Density Measurements

Blood vessels in prostate tissue sections from nontransgenic littermates and TRAMP mice with PIN, WD, and PD were identified by immunostaining with the Armenian hamster CD31 antibody. Digital fluorescence images from each prostate section were captured with standardized shutter and gain settings using an externally cooled, three-chip CCD camera (CoolCam; SciMeasure Analytical Systems, Atlanta, GA) fitted to a Zeiss Axiophot fluorescence microscope using a 10× Fluar objective. Blood vessel area densities of ductules were determined as the percentage of CD31 pixels in ductules or tumor of the total number of CD31 pixels in the entire field of view using the ImageJ software package (available at URL: http://rsb.info.nih.gov/ij [accessed April 2003]). Mean ± standard errors were calculated from three images from each mouse (three to four mice per group).

RESULTS

Comparison of H & E Staining and E-Cadherin Immunofluorescence

Evaluation of the normal prostate and TRAMP tissue sections by H & E staining and E-cadherin immunoreactivity of thin sections established the loss of epithelial E-cadherin expression with tumor progression using bright-field microscopy.19, 29, 30 In our studies, we sought to evaluate the vascular changes in the normal prostate and TRAMP tissue sections within the context of these defined histologic parameters with fluorescence microscopy by using E-cadherin as our reference.29 To do this, we first needed to verify the correlation between E-cadherin immunofluorescence of the ductule epithelium and the H & E histologic evaluations of these tissue sections (Fig. 1).

Figure 1.

E-cadherin immunoreactivity defines TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) model histopathologic stages. (A) Ductules in the normal dorsal lobe are lined with columnar epithelial cells (arrows) with scattered regions of stratified epithelial cells (yellow arrowhead). (B) In prostatic intraepithelial neoplasia (PIN) and (C) well differentiated (WD) stage tissue specimens, epithelial proliferation and invagination are conspicuous (arrows). (A and B) The loose connective tissue (asterisks) between ductules in the normal prostate and PIN stage does not stain with hematoxylin and eosin and contains a number of blood vessels (black arrowheads). (D) In poorly differentiated (PD) adenocarcinoma, a single ductule (arrow) is surrounded by tumor cells (arrowhead). (E) In the normal dorsal lobe, E-cadherin immunoreactivity (red) delineates epithelial cell junctions (arrow), whereas the loose connective tissue is E-cadherin negative (arrowhead). (F) In PIN, E-cadherin immunoreactivity reveals moderate invagination of the ductule epithelium (arrows). (G) Thickened and invaginated epithelium of variably sized ductules are characteristic of WD stage tissue specimens (arrows). (H) In PD adenocarcinoma, E-cadherin immunoreactivity is limited to epithelial cells of an entrapped collapsed ductule (arrow). Scale bar = 60 μm (A–D) and 55 μm (E–H). [The scale bar in panel H applies to all panels.]

Histologic features of prostate specimens from nontransgenic littermates and TRAMP mice were similar to those reported (Fig. 1A–D).19, 20 The histologic descriptions of the mouse prostate specimen in the current study reflect the recently established nomenclature for genetically engineered mouse models of human prostate carcinoma.20 In the normal dorsal lobe, each ductule was lined with simple columnar epithelial cells (Fig. 1A, arrows) with occasional regions of stratified epithelium (Fig. 1A, yellow arrowhead). In comparison, the luminal epithelium proliferated and multiple invaginations were present in the ductules of the dorsal lobe in the PIN (Fig. 1B, arrows) and WD stage tissue sections (Fig. 1C, arrows). In the PD stage, undifferentiated cells predominated (Fig. 1D, arrowhead), and few ductules were present (Fig. 1D, arrow). The loose connective tissue surrounding each normal and PIN stage ductule (Fig. 1A,B, asterisk) contained blood vessels (black arrowheads), but unlike the human prostate, contained few H & E-positive stained cells.20

In the normal dorsal lobe, E-cadherin immunoreactivity was strong at the epithelial cell junctions in each ductule (Fig. 1E, arrow) but was absent in the loose connective tissue between ductules (Fig. 1E, arrowhead). The pattern of E-cadherin immunoreactivity in PIN stage lesions was similar to that found in normal tissue sections except luminal invaginations of the thickened epithelium were more numerous (Fig. 1F, arrows). In the WD stage, epithelial cell proliferation and invagination were prominent such that ductules had smaller lumens and were variable in size and shape (Fig. 1G, arrows). In PD tumors (Fig. 1H), the surrounding tumor cells did not express E-cadherin; collapsed ductules containing E-cadherin immunoreactive epithelial cells were scattered throughout (Fig. 1H, arrow).

Vascular Changes in PIN, WD, and PD Stages

The vasculature in the dorsolateral lobe prostates from nontransgenic littermates and TRAMP mice was compared (Fig. 2). In the normal prostate, a network of capillaries was localized within the ductule stroma (Fig. 2A–C, arrows), which comprises a thin fibromuscular tunica of α-SMA immunoreactive bland spindle cells (Fig. 2B, arrowhead) interspersed with collagen.20 Normal stromal capillaries neither entered nor crossed the epithelium within each ductule and did not have α-SMA immunoreactive pericytes, whereas the surrounding bland spindle cells were α-SMA immunoreactive (Fig. 2B,C, arrows). Larger diameter venules and arterioles were found in the loose connective tissue around the ductules (Fig. 2B,C, asterisk), which is consistent with the H & E-stained thin section (see Fig. 1A, black arrowheads).

Figure 2.

Normal prostate and TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) vascular architecture and morphology. (A–C) In the dorsolateral lobe of the normal prostate, capillaries (CD31, green, arrows) are located between the basal surface of the epithelium (E-cadherin, red) and the luminal surface of alpha-smooth muscle actin (α-SMA) immunoreactive cells (blue, arrowhead) of each ductule. Larger blood vessels are located in the loose connective tissue outside ductules (asterisks). (D–F) In prostatic intraepithelial neoplasia (PIN), some blood vessels (arrows) extend within the stroma of the epithelial cell invaginations (red) but do not enter the lumen whereas other vessels surround the perimeter of the ductules (arrowheads). (E) In the absence of E-cadherin immunostaining, PIN stage blood vessels show a regular organization and morphology (arrows). (D–F) The distribution and hierarchy of peripheral capillaries (arrowheads) in ductules or larger blood vessels in the surrounding loose connective tissue (asterisks) outside the perimeter of α-SMA immunoreactive spindle cells (blue) are unchanged and have a normal distribution (asterisks). (B, C, E, and F) Ductule capillaries are not associated with α-SMA immunoreactive pericytes. (G) Abnormally shaped, tortuous blood vessels containing endothelial sprouts (arrows) predominate between the ductules in the well differentiated stage. Blood vessel density is increased within ductules and in the loose connective tissue surrounding each ductule. Some capillaries are adjacent to the lumen of the ductule (left arrowhead), whereas other capillaries appear disorganized (right arrowhead). (H) In the poorly differentiated stage, vessels of irregular caliber and morphology contain several endothelial sprouts (arrows) and surround the epithelial cells of a collapsed ductule. Scale bar = 15 μm (A, D, G, H), 25 μm (B , E), 55 μm (C), and 35 μm (F). [The scale bar in panel H applies to all panels.]

In the PIN stage, epithelial cell proliferation and invagination were accompanied by a corresponding inward growth of some blood vessels that remained confined within the stromal compartment (Fig. 2D–F, arrows). The distribution of more peripheral stromal capillaries was unchanged (Fig. 2D–F, arrowhead). Similar to the normal prostate ductules, proliferating blood vessels (Fig. 2E, arrows) within a PIN stage dorsal lobe ductule did not have α-SMA immunoreactive pericytes, whereas the surrounding fibromuscular tunica was α-SMA immunoreactive (Fig. 2E). In the absence of immunostaining the intervening epithelial cells, vessels in PIN stage ductules had a normal organization and regular dimensions. Proliferating blood vessels were confined within the stromal space between the epithelial cells and surrounding α-SMA immunoreactive spindle cells, and neither invaded the ductule lumen or loose connective tissue (Fig. 2F). Larger blood vessels in the loose connective tissue (Fig. 2D, F, asterisks) had a similar morphology and distribution as in the normal prostate (compare with Fig. 1A,B, black arrowheads).

Blood vessels in WD stage tissues were more abundant than those found in PIN lesions and around normal ductules (Fig. 2G). Blood vessels around ductules in WD stage tissues were enlarged, tortuous, and some had endothelial cell sprouts (Fig. 2G, arrows). Although the number of blood vessels within E-cadherin–positive WD stage ductules was increased, blood vessels in the loose connective tissue around WD stage ductules had a noticeably greater density. The peripheral distribution of capillaries observed in the normal and PIN stage stroma was not a consistent feature at this stage. Instead, the ductule capillaries were disorganized and scattered, and extended with the deeply invaginated layers of the epithelium (Fig. 2G, arrowheads).

In the PD stage, blood vessels were heterogeneous in size and shape, and the occurrence of collapsed ductules that lacked blood vessels was sporadic (Fig. 2H). PD blood vessel diameters were either similar to that of PIN vessels or up to three times as wide (data not shown). Some regions of PD tumors were sparsely distributed with tortuous, large-diameter blood vessels, whereas other regions contained a densely packed vascular network comprised of smaller diameter, abnormally shaped blood vessels. The distribution of vessels with large and small diameter was variable in different tumors. Many blood vessels in PD tumors had endothelial cell sprouts (Fig. 2G, arrows).

Increased Blood Vessel Area Density in WD and PD Stages

Blood vessel area density measurements revealed a small increase in vascularity between normal ductules and PIN lesions (Fig. 3). Mean vessel area density of WD stage tissue specimens was five times greater than in the normal prostate. Vascular area density was lower at the PD stage than at the WD stage, indicative of the disorganized heterogeneity of blood vessels in the PD stage.

Figure 3.

Quantification of blood vessel area densities in the normal prostate and in three histopathologic stages in the TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) model. Mean blood vessel area density was maximal in well differentiated stage tissues. *P < 0.0001 versus normal blood vessels by analysis of variance. Bars: ± the standard error.

Vascular Proliferation in Human Prostate Carcinoma

Vasculature changes in the murine prostate carcinoma model were compared with the vasculature of the normal human and neoplastic prostate (Fig. 4). In the normal human prostate specimen, columnar epithelial cells of ducts and acini were surrounded by a layer of basal cells (Fig. 4A, arrowhead). Nuclei of the epithelial cells were basally located (Fig. 4A, arrow). Similar to the mouse prostate, the vasculature was confined to the periphery of the acini (Fig. 4B, arrow). Blood vessels residing in the thick fibromuscular stroma between acini had uniform dimensions and an organized distribution (Fig. 4B, arrowheads).

Figure 4.

Vascular abnormalities in human prostate carcinoma. (A) A normal acinus exhibits the characteristic single layer of columnar epithelial cells (arrow) bounded by basal cells (arrowhead) in a hematoxylin and eosin (H & E) stained section of the normal human prostate. (B) A corresponding serial section shows CD34 immunoreactive capillaries (green) lie at the periphery of the acinus (arrow) shown in A. Blood vessels with regular dimensions and organized patterns also are present in the surrounding fibromuscular stroma (arrowheads). (C) In human prostate carcinoma, H & E staining reveals disorganized clusters of variably sized glands (arrows). Epithelial cell nuclei are more centrally located (arrowheads). (D) In a corresponding serial section, CD34 immunoreactive blood vessels are densely packed, variable in diameter, and disorganized in a hyperplastic acinus (arrow) and in the fibromuscular stroma (arrowhead). Scale bar = 65 μm (A, C); 70 μm (B, D).[The scale bar in Panel D applies to all figures.]

In well-differentiated, Gleason score 6 prostate carcinomas, acini were variable in size and shape, and lacked the basal cell layer (Fig. 4C, arrows). Epithelial cells had more centrally localized nuclei (Fig. 4C, arrowheads). Similar to that observed in the TRAMP WD stage tissue specimens, the vasculature was disorganized and more abundant in each acinus (Fig. 4D, arrow). The disorganized and profuse acinar blood vessels (Fig. 4D, arrow) could not be easily distinguished from proliferating capillaries in the fibromuscular stroma (Fig. 4D, arrowhead). Blood vessels in both compartments exhibited variable diameters and a tortuosity that were unlike those observed in the normal prostate specimen.

Pericyte Abnormalities in the TRAMP Tumor Vasculature

Although pericytes are closely associated with endothelial cells of normal blood vessels, in tumor blood vessels, pericytes are loosely attached and project multiple cytoplasmic processes. The vascular abnormalities we observed in TRAMP tumors led us to evaluate the morphology and immunoreactivity of pericytes associated with prostate blood vessels from nontransgenic littermates and TRAMP mice by immunohistochemistry (Fig. 5).

Figure 5.

Pericyte abnormalities in TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) tumors. (A) Platelet-derived growth factor receptor-β (PDGFR-β) positive pericytes (red, arrow) are closely associated with ductule capillaries (CD31, green) in normal prostates from nontransgenic littermates. (B) Numerous PDGFR-β immunoreactive pericytes (arrows) are associated with blood vessels in the well differentiated stage. (C) PDGFR-β positive pericytes in poorly differentiated (PD) tumors have large cell bodies that are loosely attached (arrow). (D) PD stage pericytes also are immunoreactive for alpha-smooth muscle actin (red, arrows) and have numerous cytoplasmic processes that extend away from the blood vessel wall (arrowheads). Scale bar = 25 μm (A, B); 15 μm (C, D). [The scale bar in Panel D applies to all figures.]

In the normal prostate and PIN stage, pericytes were tightly associated with the abluminal surface of endothelial cells. Pericytes associated with blood vessels in normal ductules were PDGFR-β immunoreactive (Fig. 5A, arrow), whereas pericytes in the normal and PIN stage ductules were not α-SMA immunoreactive (Fig. 2B, C, E, F). Pericytes in WD stage tissues had prominent cell bodies, were PDGFR-β positive (Fig. 5B, arrows), and α-SMA negative (data not shown). WD stage pericytes were more abundant in blood vessels residing in both the luminal epithelium of hyperplastic ductules and in the loose connective tissue around ductules (Fig. 5B, arrows). In WD stage tissues, pericytes maintained a close association with blood vessels but unlike the TRAMP PIN stage, cells other than pericytes in the loose connective tissue were weakly PDGFR-β immunoreactive (data not shown).

In the PD tumors, pericytes were both PDGFR-β (Fig. 5C, arrow) and α-SMA immunoreactive (Fig. 5D, arrows). PD stage pericytes were abnormally shaped, variable in size, loosely associated with blood vessels (Fig. 5C, D, arrows), and their coverage of the abluminal surface of PD stage tumor blood vessels was not uniform (Fig. 5D, arrows). Cytoplasmic processes of some pericytes extended beyond the tips of the endothelial cell sprouts and towards neighboring blood vessels (Fig. 5D, arrowheads).

DISCUSSION

In the current study, we demonstrate a stromal vascular network surrounds the ductules in the normal dorsolateral prostate that exhibits distinct changes in morphology and architecture in prostates from TRAMP mice. Structural changes and increased density of the vasculature coincided with proliferation of the epithelium in three histopathologic stages of the TRAMP model. Proliferation and expansion of the microvasculature throughout the TRAMP tumors were similar to those found in human prostate carcinoma. Pericytes associated with tumor blood vessels became loosely attached to endothelial cells and acquired α-SMA immunoreactivity.

The current findings of the dorsolateral lobe prostate vasculature from nontransgenic littermates and TRAMP mice are summarized (Fig. 6). Capillaries in the normal prostate were limited to the surrounding stroma of each ductule and neither penetrated the basal surface of the luminal epithelium nor the surrounding thin fibromuscular tunica. Arterioles and venules of regular diameter and distribution were present around the ductules in the loose connective tissue. In the PIN stage, proliferation and invagination of the epithelium were reciprocated by the vasculature within the ductule stroma. In the WD stage, ductule epithelial cell proliferation and invagination were increased, and blood vessels in both the ductules and surrounding loose connective tissue were disorganized, of variable diameter, and had numerous closely associated pericytes. In PD adenocarcinoma, pericytes acquired α-SMA immunoreactivity and were loosely associated with heterogeneously sized blood vessels. Small clusters of E-cadherin immunoreactive ductules were randomly distributed among the dedifferentiated tumor cells in PD tumors.

Figure 6.

Summary of epithelial and vascular changes in the normal and TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) dorsolateral prostate. The upper ductule in each TRAMP stage illustrates changes in the epithelium and vasculature relative to the lower normal ductule. Epithelial cell proliferation and invagination in the prostatic intraepithelial neoplasia (PIN) and well differentiated (WD) stages are accompanied by increased vascular proliferation within the stroma of each ductule. In WD stage tissues, the density of blood vessels and associated pericytes increases. Unlike the normal prostate, and TRAMP PIN and WD stages, poorly differentiated stage pericytes are loosely connected with vessels and express α-smooth muscle actin (blue).

Down-regulation of E-cadherin expression in TRAMP PD tumors is consistent with other studies of TRAMP mice and human prostate tumors and reflects dedifferentiation of the glandular epithelium.19, 31, 32 Reduced E-cadherin expression and dedifferentiation of epithelial cells to a fibroblastic or mesenchymal phenotype are associated with poor prognosis in gastric carcinoma,33 breast carcinoma,34 nonsmall cell lung carcinoma,35 and pancreatic adenocarcinoma.36 In the WD stage tissues, some PDGFR-β and α-SMA immunoreactive cells were found interspersed in the loose connective tissue (data not shown) that may be undifferentiated perivascular mesenchymal cells or vascular smooth muscle cells.37

The onset of α-SMA expression by pericytes in PD tumors is consistent with a similar phenotypic switch observed in pericytes in insulinomas in RIP-Tag2 transgenic mice and cultured human brain pericytes.5, 38 Although pericytes express many proteins such as α-SMA, desmin,37 aminopeptidase A,39, 40 aminopeptidase N,41, 42 human melanoma proteoglycan/NG2,43 and PDGFR-β,44 a unique pericyte-specific protein has yet been identified. Expression of α-SMA by the pericytes in the transgenic TRAMP and RIP-Tag2 tumors suggests transcriptional activation may occur by a common molecular mechanism that is influenced by the tumor microenvironment.45

The vasculature of the mouse prostate was previously observed in H & E stained thin sections.19, 20, 30 Vascular studies of immunostained thin paraffin sections using PECAM/CD31 concluded that blood vessels were only present in the loose connective tissue outside the ductules.26 In the published studies and in our H & E stained sections (Fig. 1A, B), blood vessels within the ductules were not readily detectable except in the loose connective tissue. To visualize the arrangement of blood vessels within each normal prostate ductule, we evaluated prostate sections using immunofluorescence microscopy by systemically perfusing animals with fixative to preserve the microvasculature in an open state, and immunostained 80-μm thick sections so that the 3-dimensional organization of the vasculature within the ductules could be visualized by confocal microscopy. Our immunofluorescence studies show a uniform capillary network resides within the normal ductule stroma. Larger diameter blood vessels resident in the loose connective tissue may account for their easier detection in stained thin sections.

The current model for TRAMP tumor progression proposes an angiogenic initiation switch occurs at the PIN stage followed by an angiogenic progression switch in advanced prostate tumors.26 Induction of the initiation switch results from epithelial expression of hypoxia-inducible factor 1α, VEGF, and VEGF receptor 1 (VEGFR-1). Activation of the initiation switch leads to invasion of blood vessels from the loose connective tissue into the prostatic ductules.26 Although vasculature changes at the PIN stage suggest the capillaries within the stroma expanded to accommodate epithelial cell proliferation, capillaries in PIN stage ductules did not show phenotypic characteristics of angiogenesis such as endothelial sprouts.46 Furthermore, characteristics of the vasculature in the PIN stage dorsolateral prostate suggest a local, regulated increase of proangiogenic factors was limited to the ductule vasculature because the morphology, distribution, and density of blood vessels in the adjacent loose connective tissue did not change.

The angiogenic progression switch in advanced prostate tumors has been proposed to account for increased microvascular density and VEGFR-2 expression in endothelial cells.26 Indeed, angiogenic sprouting in the WD and PD stage blood vessels we observed is consistent with reports of VEGF mRNA expression.21, 26 Moreover, mouse retinal angiogenesis studies have shown endothelial tip cell migration is dependent on a VEGF gradient whereas proliferation is sensitive to VEGF concentration.46, 47

Although the tumor vasculature phenotype is unequivocal at the PD stage, whether PD adenocarcinomas progress from PIN is speculative. We found as many as 50% of [C57BL/6 × FVB/n] F1 TRAMP males age > 25 weeks in our colony never developed tumors. Histologic evaluation of the 4 prostate lobes by E-cadherin and CD31 immunoreactivities showed evidence of epithelial hyperplasia with corresponding vascular expansion similar to the PIN stage. However, the blood vessels did not acquire phenotypic characteristics of tumor blood vessels (data not shown). Given that TRAMP tumor growth is not age dependent, but is influenced by genetic background,30 suggests anti-angiogenic factor(s) may effectively arrest the hyperplastic stage, or other unknown genetic factors or subtle hormonal differences may influence prostate tumor growth in the TRAMP model. Nevertheless, rigorous analyses of tumor progression would require expression of biologic markers that can be followed by in vivo imaging, unlike the histologic snapshots that are presented here and by other groups of the TRAMP PIN, WD, and PD stages.

We have identified vascular changes in three histopathologic stages of the TRAMP prostate. The abnormal distribution, morphology, and composition of proliferating blood vessels may yield new markers for improved detection of prostate carcinoma. Recent studies show treatment of human prostate and glioblastoma xenografts in athymic mice using a combination of endostatin, a direct angiogenesis inhibitor, and SU5416, a small molecule inhibitor of VEGFR-2, results in a synergistic anti-angiogenic effect to decrease tumor cell proliferation and blood flow.48 Combination therapies that simultaneously target endothelial cells and pericytes have been shown to enhance tumor blood vessel regression in transgenic RIP-Tag2 pancreatic islet tumors49 and in rat glioma xenografts in nude mice.50 By identifying novel molecular changes in pericyte and endothelial cells, improved strategies for early diagnosis and treatment of prostate carcinoma may be possible.

Acknowledgements

The authors thank Yan Sun and Cindy Soto (The University of Texas M. D. Anderson Cancer Center, Houston, TX) for technical assistance with the human prostate tissue specimens, Norman Greenberg (Fred Hutchinson Cancer Research Center, Seattle, WA) for the C57BL/6 TRAMP mice, Gyulnar Baimekanova and Jie Wei (University of California, San Francisco) for genotyping the TRAMP mice, and Ricardo Giordano (M. D. Anderson Cancer Center, Houston, TX) for his critical reading of the article.

Ancillary