Pigment epithelium–derived factor regulates the vasculature and mass of the prostate and pancreas


Angiogenesis sustains tumor growth and metastasis, and recent studies indicate that the vascular endothelium regulates tissue mass. In the prostate, androgens drive angiogenic inducers to stimulate growth, whereas androgen withdrawal leads to decreased vascular endothelial growth factor, vascular regression and epithelial cell apoptosis. Here, we identify the angiogenesis inhibitor pigment epithelium–derived factor (PEDF) as a key inhibitor of stromal vasculature and epithelial tissue growth in mouse prostate and pancreas. In PEDF-deficient mice, stromal vessels were increased and associated with epithelial cell hyperplasia. Androgens inhibited prostatic PEDF expression in cultured cells. In vivo, androgen ablation increased PEDF in normal rat prostates and in human cancer biopsies. Exogenous PEDF induced tumor epithelial apoptosis in vitro and limited in vivo tumor xenograft growth, triggering endothelial apoptosis. Thus, PEDF regulates normal pancreas and prostate mass. Its androgen sensitivity makes PEDF a likely contributor to the anticancer effects of androgen ablation.


The adult vasculature is typically maintained in a quiescent state by sufficient amounts of angiogenic inhibitors within the tissue matrix1,2. Tumors require neovascularization for sustained growth and metastasis and use a variety of mechanisms to shift the angiogenic balance toward an angioinductive environment1,2,3. As an integral stromal component, the vascular endothelium is uniquely positioned to receive signals from paracrine factors and circulating hormones. The intimate link between hormones, blood vessels and epithelial cells is best shown in studies of the prostate. In the adult rat, castration-induced prostate regression occurs in an orderly way whereby decreased vascular endothelial growth factor (VEGF), accelerated endothelial cell apoptosis and decreased blood flow precede epithelial apoptosis4,5,6,7,8. Conversely, testosterone-stimulated prostate regrowth is preceded by endothelial cell proliferation and increased blood flow4, indicating that normal prostate growth is angiogenesis dependent9. VEGF returns to precastration levels 3 d after castration8, possibly influenced by local hypoxia10,11. If VEGF was the sole regulator of the prostate vasculature, this increase in VEGF would stimulate endothelial cell proliferation and epithelial regrowth. Instead, gland regression continues, strongly indicating that additional factors are participating. Here, we provide evidence that PEDF is a crucial regulator of prostate growth.

PEDF is a 50-kDa secreted glycoprotein expressed in many tissues12,13,14,15,16. Its cell-type-specific functions include promoting neuronal cell survival17,18,19 and acting as a neurotrophin for retinoblastoma cells20,21. It is also a potent natural inhibitor of angiogenesis in the eye15 that, when added exogenously, is capable of suppressing retinopathy by inducing endothelial cell apoptosis22. Here we show that PEDF has a key role as a natural angiogenesis inhibitor in two hormone-sensitive organs, the prostate and the pancreas. In the absence of PEDF, both organs developed substantial stromal vascularity and epithelial cell hyperplasia. In the prostate, PEDF expression was sensitive to both hypoxia and androgens, and its expression was decreased in human prostate cancers. Treatment with recombinant PEDF targeted both endothelial cells and prostate cancer epithelial cells, triggering their apoptosis. These data show that PEDF is a key regulator of prostatic and pancreatic growth and indicate that treatments replacing PEDF may stabilize or suppress tumor growth.


Increased microvascular density in PEDF-deficient mice

To evaluate the function of PEDF in normal growth and development, we generated PEDF-deficient mice. The null allele construct disrupted Serpinf1, the gene encoding PEDF, with a cassette containing an internal ribosomal entry site and encoding the β-galactosidase and neomycin resistance genes (IRES-LacZ-Neo; Fig. 1a). Of 168 neomycin-resistant 129/SvJ embryonic stem cell clones screened, only one had the expected 9.7-kb EcoRI null-allele fragment (Fig. 1b). These cells were injected into C57B16 blastocysts, and the two male chimeras produced were crossed to C57Bl6/J females. Serpinf1+/− males were crossed to C57Bl6/J females. Offspring were genotyped by PCR (Fig. 1c), and F1 heterozygotes were crossed to generate F2 mice for analysis. Serpinf1−/− mice were viable and fertile with litters of normal size (6.0 ± 0.88).

Figure 1: Generation of PEDF-deficient mice.

(a) The eight exons of mouse Serpinf1 (top, black rectangles) on the 18.8-kb EcoRI (R) fragment. In the null allele construct (bottom), exons 3–6 were replaced with an IRES-LacZ-Neo cassette. LoxP sites (gray triangle) bordered the Neo cassette and added an EcoRI site. An IRES site (stippled box) allowed translation of the gene encoding β-galactosidase driven by the Serpinf1 promoter, and Neo expression was driven by the phosphoglycerate kinase promoter. Arrows, translational start sites. (b) Southern blot analysis of Eco RI-digested DNA isolated from embryonic stem cell clones hybridized to the probe shown in a. The wild-type allele generates an 18.8-kb fragment (lane 1), and the 9.7 kb fragment identifies the Serpinf1 allele after homologous recombination (lane 2). (c) Genotyping of mouse DNA by PCR using primers specific for the wild-type allele (WT; 449-bp product) and the null allele (KO; 737-bp product). Lanes: M, 1-kb ladder (sizes, left margin); 1, wild-type; 2, heterozygote; 3, homozygous null; 4, positive control for null allele; 5, negative control for null allele; 6, positive control for wild-type allele; 7, negative control for wild-type allele. (d) Pancreas tissue sections from 3-month-old Serpinf1+/+ (+/+) and Serpinf1−/− (−/−) mice, immunostained for smooth muscle actin or PCNA. The inset in the PCNA-stained tissue sections shows the paucity of lumen (L) in Serpinf1−/− mice. (e) Quantification of PCNA-positive cells in Serpinf1+/+ and Serpinf1−/− pancreas tissue. HPF, high-power field. *, P ≤ 0.001.

PEDF is a neuroprotective agent17,18,19 and angiogenesis inhibitor in the eye15. Loss of PEDF is linked to a spectrum of adult-onset eye diseases23, and its absence substantially alters the developing retina. In mice 3 months of age, Serpinf1−/− retinas had malpositioned vessels, irregular pigmentation, a reduced number of ganglion cells (S.E.C. and V.M.S., unpublished observations) and increased microvessel density (MVD; Table 1). Although the eye and nervous system were the likely candidates for abnormalities, several other organs of the Serpinf1−/− mice had excessive angiogenesis (Table 1). In two hormone-sensitive organs, the pancreas and prostate, distinct stromal-epithelial phenotypes were evident. On gross examination, the Serpinf1−/− pancreas (3-month-old mouse) appeared enlarged, expanding well beyond the midline. Microscopically, the acinar epithelial cells seemed less well differentiated, with increased nuclear-to-cytoplasmic ratios and cytologic atypia (Fig. 1d). The pancreatic blood vessels were excessive in number and dilated and had thickened media, as shown by staining with antibody specific to smooth muscle–specific (Table 1 and Fig. 1d). The exocrine glands were more abundant, with many cells seeming to make only abortive attempts at true glandular structures with lumens (Fig. 1d). This growth disturbance was confirmed by a 3.8-fold increase in epithelial cells positive for proliferating cell nuclear antigen (PCNA) in Serpinf1−/− pancreas compared with that of wild-type controls (Fig. 1e; 8.33 ± 0.54 versus 2.21 ± 0.28, P < 0.001).

Table 1 Angiogenic phenotype in Serpinf1−/− mouse tissues

Serpinf1−/− mice developed prostatic hyperplasia

In contrast to Serpinf1+/+ prostates, in which a simple columnar epithelial layer lined most glands, there was considerable epithelial cell hyperplasia with stratification of the nuclei in prostates of 3-month-old Serpinf1−/− mice (Fig. 2a). Hyperplasia was also evident in Serpinf1+/− mice but was less pronounced (data not shown). By 6 months of age, Serpinf1+/− mice also had moderate nuclear pleomorphism and hyperchromatic nuclei (data not shown), indicating a progressive phenotype. Confirming a defect in growth control, the number of PCNA-positive epithelial cells was significantly higher in Serpinf1−/− and Serpinf1+/− prostates than in Serpinf1+/+ prostates (Fig. 2a,b; P < 0.014). Even when adjusted for total cell numbers, the increase remained significant (data not shown). Because androgens regulate prostate development and growth, we measured serum testosterone; we found no significant differences between Serpinf1+/+ (11.35 ± 1.3 ng/ml; n = 5) and Serpinf1+/− (15.5 ± 5.1; n = 3, P = 0.36) or Serpinf1−/− (17.4 ± 5.7; n = 6, P = 0.43) mice or between Serpinf1+/− and Serpinf1−/− mice (P = 0.86), indicating that the hyperplasia was not due to altered amounts of testosterone.

Figure 2: Prostatic hyperplasia in PEDF-deficient mice.

(a) Prostate tissue sections from 3-month-old Serpinf1+/+ and Serpinf1−/− mice stained with H&E or immunostained for PCNA or smooth muscle actin. Black arrowheads, vessels; gray arrowheads, smooth muscle layer surrounding the glands. Original maginifications, ×40 and ×100. (b,c) PCNA-positive cells (b) and factor VIII–related antigen–positive vessels (c) in prostate tissues of Serpinf1+/+ (+/+) Serpinf1+/− (+/−) and Serpinf1−/− (−/−) mice. HPF, high-power field. *, P ≤ 0.013 as compared with Serpinf1+/+.

The pathologic changes within Serpinf1−/− prostates were not confined to epithelial cells. The scant numbers of stromal cells typical of wild-type controls were replaced by stroma with higher cellularity and increased MVD. Immunostaining for smooth muscle–specific actin showed a broader band of smooth muscle encircling the Serpinf1−/− glands than in controls (Fig. 2a). Similar to the pancreas phenotype, the vessels in the Serpinf1−/− prostates were more muscular than in the wild-type controls, with occasional intraglandular vessels, a feature never seen in controls (Fig. 2a). The stromal MVD was 3.2- and 1.9-fold higher in Serpinf1−/− and Serpinf1+/− prostates, respectively, than in Serpinf1+/+ prostates (Table 1 and Fig. 2c; P < 0.005). These data indicated that dysregulated angiogenesis may have been one mechanism underlying the hyperplastic epithelial growth in PEDF-deficient mice.

Loss of PEDF in human prostate cancers

The hyperplasia seen in Serpinf1−/− mice prompted us to examine PEDF expression in human prostate tissues. In normal donors (15–20 years old), we found strong PEDF immunostaining in prostate epithelial cells and stromal cells, especially in smooth muscle cells (Fig. 3a). In benign prostatic hyperplasia (BPH) tissues, the epithelial cells and stroma were also intensely stained for PEDF (Fig. 3a; mean patient age 67.8 ± 3.2 years), whereas in low-grade (Gleason score 2–6) and high-grade (Gleason score 7–10) prostate cancer tissues, PEDF staining was minimal to absent (Fig. 3a; mean patient age 68.9 ± 2.4 years). The decreased PEDF expression was probably not an age-related loss, as there was no statistically significant age difference between the BPH and cancer groups.

Figure 3: PEDF expression and function in prostate cells.

(a) Human normal prostate, BPH, and low-grade and high-grade prostate cancers immunostained for PEDF. (b) PEDF detected by immunoblot in the conditioned media of prostate cells. Lanes: +, recombinant PEDF protein (positive control); 1, PrEC; 2, DU145; 3, TSU-Pr1; 4, LNCaP; 5, PC-3. Right margin arrow, molecular weight standard. Below, identically loaded Coomassie-stained gel (loading controls). (c) Angiogenic activity in media conditioned by PrEC or a short-term normal stromal cell culture using the endothelial cell (EC) migration assay with (+) or without (−) neutralizing antibody to PEDF (αPEDF). Data were pooled from at least two independent assays. 100%, migration elicited by the media tested alone. (d) Angiogenic activity in media conditioned by PC-3 or TSU-Pr1 as described in c. *, P ≤ 0.02.

PEDF was a functional angiogenesis inhibitor in prostate cells

We assessed PEDF in media conditioned by normal prostate epithelial cell (PrEC) and stromal cell (PrSC) strains and cancer cell lines. There was a strong PEDF signal in both the epithelial and stromal cell media (Fig. 3b, lane 1 and data not shown). In contrast, the DU145, TSU-Pr1 (studied as a prostate cancer cell line for years24, although a recent study indicates that it is of bladder origin rather than prostate25), LNCaP and PC-3 cancer cell lines secreted much less to no detectable PEDF, with 82%, 93%, 100% and 35% reductions, respectively, compared with that of epithelial cells, by densitometry (Fig. 3b).

We used the microvascular endothelial cell migration assay, an in vitro assay that correlates well with in vivo angiogenic activity15,26, to test the antiangiogenic function of PEDF. Consistent with previous findings27, media conditioned with prostate epithelial cells and normal stromal cells induced endothelial cell migration; however, the addition of neutralizing antibodies to PEDF to media from PrEC or a short-term normal stromal cell culture increased the activity compared with that of media alone (Fig. 3c; P < 0.013). We obtained similar results using PrSC (data not shown). Previously, PC-3 cells were found to have the lowest angioinductive activity of these cancer lines27. The fact that PC-3 cells secreted the most PEDF (Fig. 3b) may explain this finding, as the angioinductive activity was significantly increased when PEDF was blocked (Fig. 3d; P = 0.0003). As expected, the addition of antibody to PEDF to media from TSU-Pr1 cells, which secreted little PEDF, did not alter activity (Fig. 3d; P = 0.469). These data indicated that PEDF secreted by prostate cells had antiangiogenic activity.

PEDF expression was sensitive to androgen and hypoxia

Androgens can stimulate pro-angiogenic amounts of VEGF and increase prostate vascularity28,29,30. Androgens had the opposite effect on antiangiogenic PEDF. Treatment with increasing concentrations (10−12, 10−10 and 10−7 M) of dihydrotestosterone (DHT) decreased PEDF secretion by 10%, 35% and 47%, respectively, in PrSC compared with cells grown in androgen-free media (Fig. 4a). A short-term stromal cell culture showed similar decreases, ranging from 25% to 55% (data not shown).

Figure 4: Androgen and hypoxia regulation of PEDF expression in prostate cells.

(a) PEDF in conditioned media collected from normal stromal cells treated with increasing concentrations of DHT, analyzed by immunoblot. Below each panel, identically loaded Coomassie-stained gels (loading controls). Left margin arrow, PEDF; right margin arrow, molecular weight standard (b,c). Immunohistochemistry for PEDF in ventral prostates from normal intact rats ('Intact') and from rats on day 3 and day 5 after castration (b) and in human prostate biopsies from two cancer patients (Pt 1 and Pt 2) before and 3–7 d after androgen ablation therapy (c). (d) PEDF in media conditioned by prostate epithelial cells (PrEC), prostate stromal cells (PrSC) and PC-3 cancer cells subjected to normoxic (N), CoCl2 (C) or hypoxic (H) conditions, analyzed by immunoblot as described in a. Left margin arrow, molecular weight standard.

To evaluate PEDF expression after androgen ablation in vivo, we used a rat castration model. Strong epithelial cell staining was evident at 3 d after castration, whereas only weak epithelial PEDF staining was present in intact rat prostates (Fig. 4b). Stronger staining for PEDF was observed at all post-castration time points examined (3–21 d after castration; Fig. 4b and data not shown). Because of the paucity of cells in the rodent prostate stroma, we could not assess whether PEDF staining was altered in that compartment. To assess whether PEDF expression was sensitive to androgens in humans, we immunostained prostate biopsy specimens obtained before and after androgen-ablation therapy. PEDF expression was minimally evident in only one of eight specimens obtained before therapy (Fig. 4c). In contrast, PEDF strongly immunolocalized to tumor epithelial cells in seven of eight specimens obtained after therapy, with focal stromal positivity also noted in three of eight (Fig. 4c), indicating that androgen ablation stimulated PEDF production in human prostate tumors.

Vascular regression after androgen ablation leads to a hypoxic environment in the prostate10,11, and PEDF is known to be hypoxia sensitive in retinoblastoma cells15. To assess whether prostatic PEDF expression was hypoxia sensitive, we incubated prostate cells in conditions of normoxia or hypoxia, or with CoCl2, which simulates hypoxia31. In PrSC, CoCl2 and hypoxia treatment decreased secreted PEDF amounts by 78% and 97%, respectively, compared with that in normoxia-treated cells (Fig. 4d). Hypoxia also decreased PEDF secretion by PC-3 cancer cells by 50%, whereas CoCl2 treatment produced a more modest reduction (Fig. 4d). In contrast to the results in stromal cells, CoCl2 and hypoxia treatment of PrEC resulted in slightly increased amounts of PEDF (Fig. 4d) of 22% and 15%, respectively.

PEDF treatment increased necrosis in a xenograft tumor model

To assess whether exogenous PEDF treatment could suppress tumor growth, we treated TSU-Pr1 xenograft tumors with intra-tumoral injections of vehicle or 1.5 μg recombinant PEDF daily for 14 d. Although this short-term treatment did not produce a statistically significant difference in tumor size, microscopic examination showed large necrotic areas admixed with nuclear debris in all recombinant PEDF–treated tumors (n = 5), in contrast to the minimal to no necrosis in vehicle-treated tumors (n = 5; Fig. 5a). In addition, recombinant PEDF-treated tumor cells showed less cytologic atypia and more abundant cytoplasm than did vehicle-treated tumors that had confluent regions of viable tumor cells and considerable pleomorphism (Fig. 5a). We obtained similar results with recombinant PEDF–treated PC-3 xenografts (data not shown). In regions of viable tumor, vehicle-treated tumors had nearly twice the number of mitotic figures of recombinant PEDF-treated tumors (Fig. 5b; P = 0.0004). In addition, factor VIII–related antigen immunostaining showed a significantly lower MVD in recombinant PEDF–treated tumors than in controls (Fig. 5c; P = 0.0002). Treated tumors also showed an increase in endothelial cell apoptosis as shown by dual staining for factor VIII–related antigen and TUNEL (Fig. 5d,e; P < 0.00001). To examine whether PEDF might also target tumor epithelial cells, we treated TSU-Pr1 and DU145 cancer cells with PEDF in vitro. Apoptosis of tumor epithelial cells was dose dependent (Fig. 5f,g and data not shown; P < 0.0.03). When PEDF was administered in hypoxic conditions, apoptosis increased 3.3-fold compared with results obtained with PEDF treatment or hypoxia treatment alone (Fig. 5f,g; P < 0.0001).

Figure 5: Treatment of xenograft subcutaneous human tumors with recombinant PEDF.

(a) Tissue sections from subcutaneous TSU-Pr1 tumors in nude mice treated with PBS (vehicle) or recombinant PEDF and stained with H&E. N, areas of necrosis. (b,c) Cells in mitosis (b) and cells positive for factor VIII–related antigen (c), counted in five high-power fields (HPF) in tumor sections treated with vehicle and recombinant PEDF. (d) Tumor sections simultaneously stained for factor VIII–related antigen and by TUNEL using immunofluorescent labels; corresponding images were merged (far right). (e) Percent apoptotic cells calculated from the total number of endothelial cells (ECs) and apoptotic endothelial cells. (f,g) TSU-Pr1 cells treated with recombinant PEDF with (+) or without (−) CoCl2 in vitro and stained for TUNEL with propidium iodide (PI) counterstaining, then photographed (f) and counted to determine the percent apoptotic cells (g). b, c and e: *, P < 0.009 for treated samples as compared with controls. g: *a, P < 0.001 as compared with untreated; *b, P < 0.03 as compared with untreated; and *c P < 0.0001 as compared with untreated or CoCl2 or recombinant PEDF–treated samples.


Signals from vascular endothelial cells directly participate in the development of major organ systems, including the pancreas32,33, and it has been suggested that the mass of individual organs can be controlled by the extent of their vasculature9. The function of angiogenesis inhibitors in these processes is unclear. Here, we have presented data showing that a developmental deficiency in the angiogenesis inhibitor PEDF caused profound changes in the size and cellularity of the prostate and pancreas. Loss of PEDF resulted in epithelial cell hyperplasia in both the prostate and pancreas, where it was associated with an increased number of PCNA-positive cells. Angiogenesis was also excessive within the stroma of both organs, as shown by significant increases in MVD in Serpinf1-null mice. Thus, the action of PEDF may be indirect, with dysregulated stromal angiogenesis triggering epithelial hyperplasia, although a direct effect on the normal epithelial cells has not been ruled out. The observation that in Serpinf1-null mice both the pancreas and prostate had similar increases in vessel density (about threefold) and in hyperplasia (three- to fourfold) indicated that the vessels had a critical role in maintaining tissue homeostasis.

Endocrine pancreatic development and differentiation relies on signals from the vasculature or endothelial-derived factors32. As pancreatic growth progresses, the endoderm makes frequent contacts with the endothelium and this coincides with the expression of PDX134,35 and VEGFR236. Mice transgenic for expression of PDX-VEGF have hypervascular pancreatic tissue and islet cell area expansion at the expense of the acinar area32. In mice deficient in angiogenic inhibitor thrombospondin-1, islets became hyperplastic and angiogenic, whereas the acinar area remained relatively stable37. Here we have shown that a developmental loss of PEDF also resulted in an increase in pancreatic stromal vascularity; however, instead of targeting the islet cells, the exocrine epithelial cells showed evidence of atypical hyperplasia. The islet morphology and hormone-secreting subpopulation of cells were not substantially altered (S.E.C. and V.M.S., unpublished observations). These data indicated that the gain or loss of a single angiogenic mediator can contribute to atypical growth of epithelial cells of the endocrine or exocrine pancreas.

The vasculature of most adult organs is quiescent; however, both adipose tissue38 and prostate tissue are prone to increases in mass through adulthood, indicating that these organs retain the ability to stimulate neovascularization after maturity. The signals responsible for modulating endothelial cells in normal mature tissues are poorly understood. In the prostate, it is well recognized that the modulation of the vasculature is essential in androgen ablation–induced tissue regression4,5,7. Interactions between the stromal and epithelial cells are essential to normal and neoplastic growth, with the stromal cells mediating the growth-regulatory effects of androgens39,40,41. The observation that Serpinf1−/− mice developed prostatic epithelial hyperplasia in the context of increased stromal vascularity indicated that PEDF may be critical to growth regulation in this organ. In the setting of experimental castration or tumor-related androgen ablation therapy, increased PEDF expression would seem likely to dampen the stromal vasculature. Amounts of PEDF peaked at 3 d after castration in normal rat prostate tissue, coincident with the onset of tissue hypoxia and the VEGF 'rebound'10,11, indicating that it may function to counterbalance VEGF and promote continued glandular involution. Because androgens and hypoxia regulate PEDF expression in vitro, it is likely that in vivo both tissue oxygenation and androgens are involved in regulating PEDF expression in the prostate. Moreover, exogenous PEDF treatment suppressed xenograft tumor growth by inducing endothelial cell apoptosis, but the endothelial cells were not the only target for PEDF. PEDF was capable of triggering apoptosis of cultured prostate tumor epithelial cells, an effect that was significantly augmented by hypoxia. These data indicate that multifunctional PEDF may be an effective antitumor agent.


Transgenic mice.

PEDF-deficient mice were generated by homologous recombination using a construct that replaced Serpinf1 exons 3–6 with a cassette containing an internal ribosomal entry site and the β-galactosidase and neomycin-resistance genes (Genome Systems). Embryonic stem cell clone DNA digested with EcoRI was screened by Southern blotting. Mouse genomic DNAs were genotyped using PCR primers amplifying the wild-type (737bp) and null (449bp) alleles and compared with a positive control (1μg C57B16/J DNA with 5.2 pg Serpinf1-null vector DNA, representing endogenous amounts of Serpinf1 per 1 μg haploid genome) and a negative control (all reagents except DNA). Genotypes were verified on a second DNA sample obtained at the time the mice were killed. Serum was prepared from blood collected by inferior vena cava puncture. Testosterone (2- to 9-month-old mice) was measured using the Access Analyzer (Beckman Coulter).

Animal studies.

Xenograft tumors in nude mice were generated by injecting 2.5 × 106 TSU-Pr1 cells subcutaneously in the right flank. When tumors reached 5 mm in diameter, recombinant PEDF15 (1.5 μg/50 μl) or PBS vehicle (50 μl) was injected intratumorally. Ventral prostate tissues were obtained from 2-month-old rats before and 3–21 d after surgical castration. Animals were housed in accordance with guidelines from the American Association for Laboratory Animal Care, and research protocols were approved by Northwestern's Animal Care and Use Committee.

Human tissues.

Normal prostate specimens were obtained by autopsy (n = 3), and BPH (n = 5) and cancer (n = 5) tissues were obtained from radical prostatectomy patients (Northwestern Memorial Hospital, Chicago, Illinois). Prostate biopsy tissues (n = 8) obtained before and 3–7 d after androgen ablation therapy were from Umeå University patients (Umeå, Sweden). Human tissue collections were approved by Internal Review Board committees at both institutions and all guidelines for patient consent were followed.

Cell culture.

Cell strains (PrEC, PrSC and endothelial cells; Clonetics) and cancer cell lines DU145, LNCaP, PC-3 and TSU-Pr1 (American Type Culture Collection) were maintained, short-term normal prostate stromal cell cultures were established, and serum-free conditioned media was collected as described27. For in vitro studies, cells at 70–80% confluence were washed twice with PBS, incubated 3–4 h in serum-free media and washed, and treatment media were added. Cells were incubated for 48 h in conditions of normoxia (20% O2, 5% CO2 and 75% N2) or hypoxia (0.5% O2, 5% CO2 and 94.5% N2) or with 50 μM CoCl2 in normoxia or for 24 h with 0 to 10−7 M DHT (Sigma). VEGF was measured as a control; it was increased in treated samples (data not shown), consistent with published studies42,43,44. For recombinant PEDF treatments, cells were incubated for 24–36 h with 0–100 nM recombinant PEDF with or without 100 μM CoCl2 in 10-cm dishes or on chamber slides for TUNEL staining. Cycloheximide (100 μM) served as a positive control for apoptosis studies (data not shown). Assays were done at least twice.

Tissue analysis.

Tissues were fixed with formalin, imbedded in paraffin, sectioned and stained with H&E. Antibodies to PEDF15, factor VIII–related antigen, PCNA and smooth muscle actin (DAKO) were used for PEDF, MVD, cell proliferation and smooth muscle immunostaining, respectively. Cells positive for factor VIII–related antigen and PCNA were counted in five high-power fields. For xenograft tumors, only non-necrotic fields were assessed.

Apoptosis was detected using the ApopTag Fluorescein TUNEL kit (Serologicals). Chamber slides were counterstained with propidium iodide (0.5 μg/ml in FluorSave; Calbiochem), and 600 cells were assessed per group. Tumor tissues were stained for TUNEL as above and factor VIII–related antigen using an R-phycoerythrin-conjugated secondary antibody (Sigma). Endothelial cells and TUNEL-positive endothelial cells were counted in ten medium-power fields (minimum 500 cells), and the percentage of apoptotic endothelial cells was calculated.


PEDF was detected using antibody to PEDF15 on blots (Hybond-C; Amersham) containing 20 μg sample separated by 10% SDS-PAGE. Identically loaded Coomassie-stained gels confirmed equal protein loading. Molecular size standards (BioRad) and recombinant PEDF (20 ng/lane) were included as controls. Band intensities were quantified by densitometry with equalization to Coomassie-stained gels.

In vitro angiogenesis assay.

The endothelial cell migration assay was done as described before27. Samples were assayed in quadruplicate. Serum-free medium plus 0.1% bovine serum albumin and VEGF (250–500 pg/ml; R&D Systems) served as background migration and positive controls, respectively. Purified proteins alone and with neutralizing antibodies were included as controls (data not shown).

Statistical analysis.

Data were analyzed using Student's t-test; P ≤ 0.05 was considered significant.


  1. 1

    Bouck, N., Stellmach, V. & Hsu, S.C. How tumors become angiogenic. Adv. Cancer Res. 69, 135–174 (1996).

    CAS  Article  Google Scholar 

  2. 2

    Hanahan, D. & Folkman, J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 86, 353–364 (1996).

    CAS  Article  Google Scholar 

  3. 3

    Folkman, J., Shochat, S.J., Strand, R.D. & Fellows, K.E. Tumor angiogenesis: therapeutic implications. N. Engl. J. Med. 285, 1182–1186 (1971).

    CAS  Article  Google Scholar 

  4. 4

    Franck-Lissbrant, I., Haggstrom, S., Damber, J.E. & Bergh, A. Testosterone stimulates angiogenesis and vascular regrowth in the ventral prostate in castrated adult rats. Endocrinology 139, 451–456 (1998).

    CAS  Article  Google Scholar 

  5. 5

    Shabsigh, A. et al. Rapid reduction in blood flow to the rat ventral prostate gland after castration: preliminary evidence that androgens influence prostate size by regulating blood flow to the prostate gland and prostatic endothelial cell survival. Prostate 36, 201–206 (1998).

    CAS  Article  Google Scholar 

  6. 6

    Shabsigh, A. et al. Early effects of castration on the vascular system of the rat ventral prostate gland. Endocrinology 140, 1920–1926 (1999).

    Article  Google Scholar 

  7. 7

    Lekas, E., Johansson, M., Widmark, A., Bergh, A. & Damber, J.E. Decrement of blood flow precedes the involution of the ventral prostate in the rat after castration. Urol. Res. 25, 309–314 (1997).

    CAS  Article  Google Scholar 

  8. 8

    Burchardt, M. et al. Vascular endothelial growth factor-A expression in the rat ventral prostate gland and the early effects of castration. Prostate 43, 184–194 (2000).

    CAS  Article  Google Scholar 

  9. 9

    Folkman, J. Is tissue mass regulated by vascular endothelial cells? Prostate as the first evidence. Endocrinology 139, 441–442 (1998).

    CAS  Article  Google Scholar 

  10. 10

    Buttyan, R., Ghafar, M.A. & Shabsigh, A. The effects of androgen deprivation on the prostate gland: cell death mediated by vascular regression. Curr. Opin. Urol. 10, 415–420 (2000).

    CAS  Article  Google Scholar 

  11. 11

    Shabsigh, A. et al. Biomarker analysis demonstrates a hypoxic environment in the castrated rat ventral prostate gland. J. Cell. Biochem. 81, 437–444 (2001).

    CAS  Article  Google Scholar 

  12. 12

    Pignolo, R.J., Rotenberg, M.O. & Cristofalo, V.J. Analysis of EPC-1 growth state-dependent expression, specificity, and conservation of related sequences. J. Cell. Physiol. 162, 110–118 (1995).

    CAS  Article  Google Scholar 

  13. 13

    Tombran-Tink, J. et al. Organization, evolutionary conservation, expression and unusual Alu density of the human gene for pigment epithelium-derived factor, a unique neurotrophic serpin. Mol. Vis. 2, 11–19 (1996).

    CAS  PubMed  Google Scholar 

  14. 14

    Becerra, S.P. Structure-function studies on PEDF. A noninhibitory serpin with neurotrophic activity. Adv. Exp. Med. Biol. 425, 223–237 (1997).

    CAS  Article  Google Scholar 

  15. 15

    Dawson, D.W. et al. Pigment epithelium-derived factor: a potent inhibitor of angiogenesis. Science 285, 245–248 (1999).

    CAS  Article  Google Scholar 

  16. 16

    Crawford, S.E. et al. Pigment epithelium-derived factor (PEDF) in neuroblastoma: a multifunctional mediator of Schwann cell anti-tumor activity. J. Cell. Sci. 114, 4421–4428 (2001).

    CAS  PubMed  Google Scholar 

  17. 17

    Taniwaki, T., Becerra, S.P., Chader, G.J. & Schwartz, J.P. Pigment epithelium-derived factor is a survival factor for cerebellar granule cells in culture. J. Neurochem. 64, 2509–2517 (1995).

    CAS  Article  Google Scholar 

  18. 18

    Bilak, M.M. et al. Pigment epithelium-derived factor (PEDF) protects motor neurons from chronic glutamate-mediated neurodegeneration. J. Neuropathol. Exp. Neurol. 58, 719–728 (1999).

    CAS  Article  Google Scholar 

  19. 19

    Cao, W. et al. Pigment epithelium-derived factor protects cultured retinal neurons against hydrogen peroxide-induced cell death. J. Neurosci. Res. 57, 789–800 (1999).

    CAS  Article  Google Scholar 

  20. 20

    Tombran-Tink, J., Chader, G.G. & Johnson, L.V. PEDF: a pigment epithelium-derived factor with potent neuronal differentiative activity. Exp. Eye Res. 53, 411–414 (1991).

    CAS  Article  Google Scholar 

  21. 21

    Tombran-Tink, J. & Johnson, L.V. Neuronal differentiation of retinoblastoma cells induced by medium conditioned by human RPE cells. Invest. Ophthalmol. Vis. Sci. 30, 1700–1707 (1989).

    CAS  PubMed  Google Scholar 

  22. 22

    Stellmach, V.V., Crawford, S.E., Zhou, W. & Bouck, N. Prevention of ischemia-induced retinopathy by the natural ocular antiangiogenic agent pigment epithelium-derived factor. Proc. Natl. Acad. Sci. USA 98, 2593–2597 (2001).

    CAS  Article  Google Scholar 

  23. 23

    Bouck, N. PEDF: anti-angiogenic guardian of ocular function. Trends Mol. Med. 8, 330–334 (2002).

    CAS  Article  Google Scholar 

  24. 24

    Iizumi, T., Yazaki, T., Kanoh, S., Kondo, I. & Koiso, K. Establishment of a new prostatic carcinoma cell line (TSU-Pr1). J. Urol. 137, 1304–1306 (1987).

    CAS  Article  Google Scholar 

  25. 25

    van Bokhoven, A., Varella-Garcia, M., Korch, C, & Miller, G.J. TSU-Pr1 and JCA-1 cells are derivatives of T24 bladder carcinoma cells and are not of prostatic origin. Cancer Res. 61, 6340–6344 (2001).

    CAS  PubMed  Google Scholar 

  26. 26

    Good, D.J. et al. A tumor suppressor-dependent inhibitor of angiogenesis is immunologically and functionally indistinguishable from a fragment of thrombospondin. Proc. Natl. Acad. Sci. USA 87, 6624–6628 (1990).

    CAS  Article  Google Scholar 

  27. 27

    Doll, J.A. et al. Thrombospondin-1, vascular endothelial growth factor and fibroblast growth factor-2 are key functional regulators of angiogenesis in the prostate. Prostate 49, 293–305 (2001).

    CAS  Article  Google Scholar 

  28. 28

    Joseph, I.B., Nelson, J.B., Denmeade, S.R. & Isaacs, J.T. Androgens regulate vascular endothelial growth factor content in normal and malignant prostatic tissue. Clin. Cancer Res. 3, 2507–2511 (1997).

    CAS  PubMed  Google Scholar 

  29. 29

    Jain, R.K. et al. Endothelial cell death, angiogenesis, and microvascular function after castration in an androgen-dependent tumor: role of vascular endothelial growth factor. Proc. Natl. Acad. Sci. USA 95, 10820–10825 (1998).

    CAS  Article  Google Scholar 

  30. 30

    Sordello, S., Bertrand, N. & Plouet, J. Vascular endothelial growth factor is up-regulated in vitro and in vivo by androgens. Biochem. Biophys. Res. Commun. 251, 287–290 (1998).

    CAS  Article  Google Scholar 

  31. 31

    Goldberg, M.A., Dunning, S.P. & Bunn, H.F. Regulation of the erythropoietin gene: evidence that the oxygen sensor is a heme protein. Science 242, 1412–1415 (1988).

    CAS  Article  Google Scholar 

  32. 32

    Lammert, E., Cleaver, O. & Melton, D. Induction of pancreatic differentiation by signals from blood vessels. Science 294, 564–567 (2001).

    CAS  Article  Google Scholar 

  33. 33

    Matsumoto, K., Yoshitomi, H., Rossant, J. & Zaret, K.S. Liver organogenesis promoted by endothelial cells prior to vascular function. Science 294, 559–563 (2001).

    CAS  Article  Google Scholar 

  34. 34

    Jonsson, J., Carlsson, L., Edlund, T. & Edlund, H. Insulin-promoter-factor 1 is required for pancreas development in mice. Nature 371, 606–609 (1994).

    CAS  Article  Google Scholar 

  35. 35

    Offield, M.F. et al. PDX-1 is required for pancreatic outgrowth and differentiation of the rostral duodenum. Development 122, 983–995 (1996).

    CAS  PubMed  PubMed Central  Google Scholar 

  36. 36

    Christofori, G., Naik, P. & Hanahan, D. Vascular endothelial growth factor and its receptors, flt-1 and flk-1, are expressed in normal pancreatic islets and throughout islet cell tumorigenesis. Mol. Endocrinol. 9, 1760–1770 (1995).

    CAS  Google Scholar 

  37. 37

    Crawford, S.E. et al. Thrombospondin-1 is a major activator of TGF-b1 in vivo. Cell 93, 1159–1170 (1998).

    CAS  Article  Google Scholar 

  38. 38

    Rupnick, M.A. et al. Adipose tissue mass can be regulated through the vasculature. Proc. Natl. Acad. Sci. USA 99, 10730–10735 (2002).

    CAS  Article  Google Scholar 

  39. 39

    Chung, L.W. The role of stromal-epithelial interaction in normal and malignant growth. Cancer Surv. 23, 33–42 (1995).

    CAS  PubMed  Google Scholar 

  40. 40

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

    CAS  Article  Google Scholar 

  41. 41

    Wong, Y.C. & Wang, Y.Z. Growth factors and epithelial-stromal interactions in prostate cancer development. Int. Rev. Cytol. 199, 65–116 (2000).

    CAS  Article  Google Scholar 

  42. 42

    Cvetkovic, D. et al. Increased hypoxia correlates with increased expression of the angiogenesis marker vascular endothelial growth factor in human prostate cancer. Urology 57, 821–825 (2001).

    CAS  Article  Google Scholar 

  43. 43

    Liu, X.H. et al. Upregulation of vascular endothelial growth factor by cobalt chloride-simulated hypoxia is mediated by persistent induction of cyclooxygenase-2 in a metastatic human prostate cancer cell line. Clin. Exp. Metastasis. 17, 687–694 (1999).

    CAS  Article  Google Scholar 

  44. 44

    Semenza, G.L. Hypoxia, clonal selection, and the role of HIF-1 in tumor progression. Crit. Rev. Biochem. Mol. Biol. 35, 71–103 (2000).

    CAS  Article  Google Scholar 

  45. 45

    Abramson, L.P. et al. Wilms' tumor growth is suppressed by antiangiogenic pigment epithelium-derived factor in a xenograft model. J. Ped. Surg. 38, 336–342 (2003).

    Article  Google Scholar 

Download references


We thank the Pathology Core Facility of the Robert H. Lurie Comprehensive Cancer Center of Northwestern University for assistance in specimen procurement. This work was supported in part by National Institutes of Health grant CA64329 to S.E.C. and Department of Defense grant DAMD17-99-1-9521 to J.A.D.

Author information



Corresponding author

Correspondence to Susan E Crawford.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Doll, J., Stellmach, V., Bouck, N. et al. Pigment epithelium–derived factor regulates the vasculature and mass of the prostate and pancreas. Nat Med 9, 774–780 (2003). https://doi.org/10.1038/nm870

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing