Pigment epithelial-derived factor (PEDF), an angiogenesis inhibitor with neurotrophic properties, balances angiogenesis in the eye and blocks tumor progression. Its neurotrophic function and the ability to block vascular leakage is replicated by the PEDF 44-mer peptide (residues 58-101). We analyzed PEDFs' three-dimensional structure and identified a potential receptor-binding surface. Seeking PEDF-based antiangiogenic agents we generated and tested peptides representing the middle and lower regions of this surface. We identified previously unknown antiangiogenic epitopes consisting of the 34-mer (residues 24-57) and a shorter proximal peptide (TGA, residues 16-26) with the critical stretch L19VEEED24 and a fragment within the 44-mer (ERT, residues 78-94), which retained neurotrophic activity. The 34-mer and TGA, but not the 44-mer reproduced PEDF angioinhibitory signals hinged on c-jun-NH2-kinase–dependent nuclear factor of activated T cell deactivation and caused apoptosis. Conversely, the ERT, but not the 34-mer/TGA induced neuronal differentiation. For the 44-mer/ERT, we showed a novel ability to cause neuroendocrine differentiation in prostate cancer cells. PEDF and the peptides bound endothelial and PC-3 prostate cancer cells. Bound peptides were displaced by PEDF, but not by each other, suggesting multiple receptors. PEDF and its active fragments blocked tumor formation when conditionally expressed by PC-3 cells. The 34- and 44-mer used distinct mechanisms: the 34-mer acted on endothelial cells, blocked angiogenesis, and induced apoptosis whereas 44-mer prompted neuroendocrine differentiation in cancer cells. Our results map active regions for the two PEDF functions, signaling via distinct receptors, identify candidate peptides, and provide their mechanism of action for future development of PEDF-based tumor therapies.
- antitumor activity
- neuronal differentiation
- receptor-interacting surfaces
- functional epitopes
Pigment epithelial-derived factor (PEDF) increases neuronal survival and differentiation, and blocks angiogenesis by inducing endothelial cell death ( 1, 2). In the activated endothelial cell, PEDF up-regulates FasL death ligand, which engages inducer-generated Fas receptor causing apoptosis ( 3). PEDF also blocks basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF)-induced transcription by deactivating the nuclear factor of activated T cells (NFAT). PEDF activates c-jun-NH2 kinases (JNK), which restore NFAT inactive state, thereby blocking expression of endogenous caspase inhibitor, c-FLIP (FLICE inhibitory protein; ref. 4). Conversely, PEDF neuroprotective function entails nuclear factor κB–dependent events ( 5– 7).
Studies of retinopathy and macular degeneration show PEDF importance in the eye ( 8– 13), where it suppresses angiogenesis and vascular leakage ( 14– 16). PEDF controls tumor growth; it is a candidate tumor suppressor in neuroectodermal tumors, mouse melanoma, and in ovarian cancer ( 17, 18). Forced PEDF expression delays the growth and invasion of lung carcinoma, hepatocellular carcinoma, melanoma, and glioblastoma, where it blocks neovascularization ( 19– 24). Moreover, decreased PEDF levels in the metastatic prostate adenocarcinoma in rat and humans, compared with the nonmetastatic disease imply that the loss of PEDF contributes to the progression to a metastatic phenotype ( 25). PEDF antitumor effects are not limited to antiangiogenesis. In neuroblastoma, PEDF from Schwannian stroma simultaneously blocks angiogenesis and redifferentiates the tumor ( 26). In PEDF−/− mice, hypervascularization of the prostate is combined with hyperplasia ( 27), whereas in cultured prostate adenocarcinoma, exogenous PEDF induces apoptosis ( 27). Finally, PEDF is repressed by testosterone in the cultured prostate epithelium and increased in the prostate in vivo upon castration, suggesting that PEDF is as a key hormone-regulated angiogenesis inhibitor in this organ ( 27).
Dissimilar molecular and cellular events triggered by PEDF point to distinct PEDF receptors, which elicit divergent signals. PEDF binds glycosaminoglycans, collagens ( 28– 30), and an 80 kDa surface protein in Y-79 retinoblastoma and cerebral granule cells ( 31, 32). Recently a 60 kDa candidate receptor was identified on endothelial cells ( 33). PEDF negatively charged, acidic region binds collagen, lacks neurotrophic activity and may confer antiangiogenic properties ( 30). Three aspartates (236, 238, and 280) and the basic cluster (Arg126, Lys127, Arg129) in mouse PEDF are critical to collagen and heparin binding ( 34). The NH2 terminus contains putative receptor-binding site where a 44-mer peptide (residues 58-101) with critical Glu81, Ile83, Leu92, and Ser95 ( 29, 31, 35) replicates neurotrophic function and antivasopermeability, whereas the more proximal 34-mer (residues 24-57) is non-neurotrophic ( 31).
We analyzed the peptides in PEDF putative receptor-interacting site, modeled from three-dimensional structure ( 36) and identified the following functional epitopes, a 44-mer, an antiangiogenic 34-mer, and an additional antiangiogenic fragment (TGA) upstream of the 34-mer. The 44-mer lacked inhibitory activity, but its internal fragment (ERT), which contains the residues controlling permeability, blocked angiogenesis.
The 34-mer, TGA, and ERT, but not the 44-mer elicited endothelial cell apoptosis, blocked migration and angiogenesis. The 34-mer, but not the 44-mer reproduced JNK-dependent NFAT deactivation and decreased c-FLIP expression. In contrast, the 44-mer and ERT, but not the 34-mer and TGA induced neurite outgrowth in the retinoblastoma cells.
To determine the contribution of PEDF, two functions in its antitumor effect, we expressed PEDF fragments in the PC-3 prostate adenocarcinoma cells. PEDF N-ter fragment containing both the 34- and 44-mer delayed tumor progression and blocked angiogenesis. Both antiangiogenic 34-mer and neurotrophic 44-mer reduced tumor growth; however, only the 34-mer reduced tumor angiogenesis. The 44-mer did not inhibit angiogenesis, but consistent with its neurotrophic function, caused neuroendocrine phenotype in the PC-3 cells and tumors.
PEDF bound endothelial cells with high affinity indicating the existence of specific surface receptor(s). Consistent with their location in the putative receptor-binding site, the 34- and 44-mer specifically bound both endothelial cell and PC-3 cells. The 34-mer binding was displaced by PEDF but not by the 44-mer, again suggesting two distinct receptors.
Our results suggest that distinct PEDF surfaces interact with nonidentical receptors on the prostate and endothelial cell. The 34-mer causes JNK-dependent endothelial cell apoptosis due to NFATc2 deactivation and c-FLIP blockade, whereas the 44-mer triggers neuroendocrine differentiation of the prostate epithelium. Mapping of the antiangiogenic and prodifferentiation PEDF functions to separate regions will facilitate the development of PEDF-based tumor therapies.
Materials and Methods
Cells and reagents. Human umbilical vascular endothelial cell (HUVEC, VEC Technologies, Inc., Rensselaer, NY) were grown in MCDB131 (Sigma, St. Louis, MO) and supplements (BioWhittaker, Walkersville, MD), Y79 in RPMI 1640 (Invitrogen, San Diego, CA), 10% fetal bovine serum, l-Gln, penicillin and streptomycin, and PC-3 expressing Tet repressor (PC3-TR4) cultured in RPMI 1640, 10% serum (Hyclone, Logan, UT), 1 μg/mL blasticidin (Invitrogen), 2% penicillin and streptomycin. Synthetic peptides (95% pure) have been dialyzed to remove guanidinium chloride.
Cloning. PEDF cDNA fragments N-ter (57-419 bp), 34-mer (186-287) and 44-mer (288-419) were PCR-amplified from DNA template (pCEP-PEDF; ref. 37) with the following primers: N-ter, 5′-CTA GCT TAA GAA GCT TAG AAT GCA GGC CCT GGT-3′(forward); 5′-ACG CAC CGG TGG TAC CAT GGA TGT CTG GG-3′ (reverse, NT-r); 34-mer, 5′-CTA GCT TAA GAA GCT TAG GAT GGA TCC TTT CTT CAA-3′ (forward); 5′-ACG CAC CGG TTG GGC CCC TTG GGT-3′ (reverse); 44-mer, 5′-ACG CAC CGG TTT AGT TGG TCG TGG GAC TC-3′ (forward) and NT-r. PCR products were cloned in pcDNA 4/TO/myc-His, (AgeI/HindIII) and sequences. PC3-TR transfectants were selected on 50 μg/mL Zeocin, induced with 1 μg/mL doxycycline, and expression was tested by RT-PCR and immunostaining.
Endothelial cell chemotaxis. Endothelial cell chemotaxis was measured as in ref. ( 38). HUVECs starved in MDCB131, 0.1% bovine serum albumin (Sigma) were plated at 1.5 × 106/mL, on the lower side of porous membranes (8 μm, Nucleopore) in Boyden chambers, the samples were added to the upper side. The migration was counted in 10,400× fields (controls: 0.1% BSA, 10 ng/mL bFGF).
Apoptosis. HUVECs (5 × 104/well) were grown in 0.2% serum on gelatinized coverslips in the 24-well plates, treated for 18 hours with PEDF or PEDF peptides, fixed and stained with ApopTag kit (Serologicals, Norcross, GA). Apoptosis was measured in 600 to 1,200 cells per treatment.
Western blotting. HUVECs were harvested, lysed in 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1% NP40, 0.5% deoxycholate, protease/phosphatase inhibitors, cleared (50 minutes 4°C followed by centrifugation), resolved by SDS-PAGE (40-50 μg per lane), transferred to membranes, probed for phospho-JNK (Promega, Madison, WI, 1/5,000 in TBS-T, 4°C) and reprobed for JNK-1 (PharMingen, San Diego, CA, 1/500) to control loading. Serum samples from tumor-bearing mice (1 μg per lane) were resolved on 4% to 20% gradient PAAG, the membranes probed with mouse anti-ChrA (1/300, Lab Vision, Fremont, CA) or rabbit anti-NSE (1/1,000, Biomed, Aurora, OH).
Biotinylation. A mix of 100 μg peptide, 50 μL DMSO, 200 μL RIB buffer [200 mmol/L Tris base (pH 7.4), 1 mmol/L phenylmethylsulfonyl fluoride, 2 mmol/L benzamidine, 1 mmol/L aprotinin, 4 mmol/L chymostatin, 20 mmol/L leupeptin] ± 2 mg Biotin-X-X-NHS (Sigma) was incubated for 30 minutes (37°C) and dialyzed overnight (1,000 NMW, SpectraPor, 4°C) against HBBS (Sigma).
Binding. Cells were collected in 20 mmol/L EDTA, washed in cold PBS, resuspended in binding buffer (1.5 × 106/mL, 1% BSA, 0.02% NaN3 in PBS), incubated with biotinylated peptides or with radiolabeled PEDF (1 hour, 4°C), and washed thrice. Cy5-streptavidin was added (2 μg/mL for 30 minutes) followed by three more washes and fluorescence-activated cell sorting analysis.
Y-79 differentiation. Cells were cultured 5 days in defined medium [106/10 mL MEM, N2 supplements, 1× MEM nonessential amino acids, 10 mmol/L HEPES (pH 7.3), 200 mmol/L Gln, penicillin and streptomycin], harvested, and seeded in RPMI, 10% fetal bovine serum, 200 mmol/L Gln, 1% penicillin and streptomycin on poly-L-ornithine coated coverslips (Sigma; 105/well, six-well plates). After attachment, the cells were washed and grown 1 to 2 weeks in defined medium with test substances. The peptides were tested twice, a minimum of 200 aggregates per treatment was examined.
Corneal angiogenesis. Corneal angiogenesis was assessed as in ref. ( 39). Sucralfate pellets were implanted in the cornea of C57/B16 mice (National Cancer Institute). Capillary growth was scored as a positive response. Three to six animals (6-11 corneal implants) were examined per sample. Animals were treated following the NIH guidelines, and the protocols approved by the Northwestern University Animal Care and Use Committee.
Tumorigenicity assay. Athymic mice (nu/nu females, National Cancer Institute, 4-6 weeks old) received s.c. injections of PC-3 cells in the hindquarters (1 × 106 cells per site, 10 mice per clone). For inducible expression, doxycycline in drinking water (1 mg/mL) was given to half of the animals; tumors were monitored daily, frozen at harvest, the experiment was repeated twice.
Immunostaining. Cryosections were fixed for 10 minutes in acetone, 1:1 acetone/chloroform and acetone, washed in PBS and blocked with Biotin-Avidin kit (Vector, Marion, IA) and 2% donkey serum (Jackson Immunoresearch). The sections were incubated with CD31 antibodies (1/125, BD PharMingen) in 2% donkey serum, washed and followed by donkey anti-rat-rhodamin X (1/200, Jackson Immunoresearch, West Grove, PA). The slides were developed with 20 μg/mL streptavidin-Cy5 in PBS. Apoptosis was visualized by terminal nucleotidyl transferase–mediated nick end labeling (ApopTag).
Chromatin immunoprecipitation. Chromatin immunoprecipitation was done with a kit from Upstate Biotechnology (Lake Placid, NY) as in ref. ( 4). Formaldehyde was added to the medium, lysates sonicated to generate ∼1 kb DNA fragments and incubated at 4°C with NFATc2 pAb672 (J.M. Redondo, CNIC, Spain), or with control rabbit IgG (Santa Cruz). The complexes were isolated on salmon sperm DNA-protein A agarose, DNA extracted and amplified with c-FLIP promoter primers (5′-TCACGTTTGCTATGACTCCCAGAC-3′, 5′-TCCACGCGTTAGGAGTAAACACTG-3′; 382 bp product).
Statistical evaluation of the data. Statistical evaluation of the data was done by paired Student's t test or by Fisher's exact test. SE and P values are shown where appropriate.
Design of PEDF peptides representing putative receptor-interacting surface. A 44-mer NH2-terminal fragment retains PEDF neurotrophic activity ( 31), suggesting that the lower acidic surface including parts of the NH2-terminal helices, hA-hD, represents the neurotrophic receptor-binding region. We generated peptides covering this region ( Fig. 1A, left ) to determine if it encompasses antiangiogenic receptor-binding surface. The NH2-terminal peptide TGA represented the middle of acidic surface with the E21EED string as a folded unit. The 34-mer modeled part of hA accessible on the lower surface, and ERT represented the lower hC/loop within the 44-mer ( Fig. 1A, right). Two TGA variants were made, TGA-Q with three contiguous glutamates replaced with glutamines (TGALVQQQDPF), and TGA-S with hydrophobic leucine and valine changed to serines (TGASSEEEDPF).
The 34-mer, TGA, and ERT, but not the 44-mer were antiangiogenic. Seeking epitopes critical for antiangiogenesis, we screened the peptides in endothelial cell apoptosis and chemotaxis assays. Apoptosis is an essential component of antiangiogenesis ( 40). Similar to native PEDF, the 34-mer, TGA, and ERT caused endothelial cell apoptosis at 10 nmol/L. In contrast, the 44-mer failed to induce apoptosis at 10 nmol/L and reduced background apoptosis at 100 nmol/L ( Fig. 1B, left). The ability to interfere with endothelial cell migration correlates with apoptosis induction in vitro and with antiangiogenesis in vivo ( 41– 43). Full-size PEDF inhibits endothelial cell chemotaxis ( 37). Similarly the 34-mer, TGA and ERT blocked bFGF-induced endothelial cell migration ( Fig. 1B, right) but not the random motility (data not shown). In contrast, the 44-mer was noninhibitory ( Fig. 1B, right), and induced endothelial cell migration (data not shown). Thus, the 34-mer/TGA epitope was antiangiogenic and distinct from the 44-mer. In agreement with the in vitro observations, the 34-mer, TGA, and ERT were inhibitory in the mouse corneal assay whereas the 44-mer remained stimulatory ( Fig. 1C).
Residues LVEEED in TGA were critical for the antiangiogenic activity. Structural analysis suggested the importance of the highly charged VEEED stretch within the TGA peptide. Two mutations in this area (TGA-Q and TGA-S) destroyed TGA's ability to block endothelial cell chemotaxis and angiogenesis in vivo ( Fig. 1D).
The 34-mer, TGA, and ERT reproduced PEDF signaling events. PEDF triggers apoptosis and stops angiogenesis by inactivation of NFATc2 transcription factor via JNK-1 and -2 ( 4). Like PEDF, the inhibitory 34-mer, TGA and ERT, increased active JNK in bFGF-stimulated endothelial cells. Conversely, the 44-mer had only modest effect ( Fig. 2A, left ). The substitutions in LVEEED motif abolished JNK phosphorylation ( Fig. 2A, right). Similarly, the 34-, but not the 44-mer decreased NFATc2 activation as was determined by Western blot for phospho-NFATc2 ( Fig. 2B). Moreover, the 34-mer but not the 44-mer increased JNK-2 association to NFATc2 ( Fig. 2C). The mRNA and endogenous promoter activity for NFAT target gene, c-FLIP ( 3, 4) were reduced by PEDF and the antiangiogenic 34-mer, but not by the neurotrophic 44-mer ( Fig. 2D).
PEDF fragments blocked tumor growth via distinct mechanisms. We generated PC-3 clones conditionally expressing PEDF fragments: the 34-mer, the 44-mer and the N-ter, which retains both regions. The expression of PEDF fragments, inducible by doxycycline ( Fig. 3A ), had no effect on PC-3 growth in vitro (data not shown). N-ter PEDF delayed tumor progression (P < 0.0001). The antiangiogenic 34-mer and the neurotrophic 44-mer also reduced tumor growth ( Fig. 3B, left; P < 0.0005 and P < 0.0003, respectively). The lack of an in vitro effect suggested angiogenesis-dependent action. Indeed, N-ter and 34-mer substantially decreased tumor microvessel density. However, in the tumors expressing the 44-mer, tumor microvessel density remained high ( Fig. 3B and C). In addition, the N-ter or 34-mer, but not the 44-mer caused >2-fold increase in tumor cells apoptosis ( Fig. 3B and C).
PC-3 tumors expressed significant levels of FasL. However, the number of capillaries positive for FasL was higher in the tumors expressing the N-ter and 34-mer, but not the 44-mer (data not shown).
PEDF, 44-mer, and ERT induced neuronal differentiation in Y-79 and PC-3 cells. To define PEDF surfaces responsible for the neurotrophic activity, we tested synthetic peptides in Y-79 differentiation assay. As was shown earlier, PEDF and the 44-mer induced neurite outgrowth. The 34-mer and TGA had no effect and ERT induced neurites, but less potently than PEDF or 44-mer ( Fig. 4A and B ).
Prostate epithelium may assume neuroendocrine phenotype manifested by the dendrite-like processes, by increased neuron-specific markers like neuron-specific enolase (NSE) or chromogranin A (ChrA; see Discussion), and by secretion of neuropeptides, including serotonin and gastrin-releasing peptide (bombesin; ref. 44). PC-3 cells expressing N-ter or the 44-mer formed processes typical of neuroendocrine differentiation (26% and 33% of the cells, respectively), compared with the vector or 34-mer (2-4%, P < 0,0003; Fig. 4C). Cytokeratin K8 was present in control tumors and decreased by the 44-mer and PEDF (data not shown) indicating less pronounced epithelial phenotype. The mRNA for gastrin-releasing peptide/bombesin was increased in the LNCaP prostate adenocarcinoma cells by exogenous PEDF or by 44-mer, but not 34-mer (data not shown). RT-PCR showed increased mRNA for neuroendocrine differentiation marker ChrA in PC-3 cells expressing PEDF or the 44-mer (data not shown). This remained true in vivo because ChrA and NSE were elevated in the serum of mice bearing tumors, expressing PEDF and the 44-mer, but not 34-mer ( Fig. 4D).
PEDF, 34-mer, and 44-mer bind endothelial and PC-3 cells. PEDF bound HUVECs with KD = 5 nmol/L ( Fig. 5A and B ) and radioligand displaced with cold PEDF (EC50 = 5 nmol/L, Fig. 5C), suggesting endothelial cell surface receptors. Biotin-labeled 34- and 44-mers (34-merB and 44-merB) bound HUVECs in a dose-dependent manner ( Fig. 6A ) and were displaced by 10× excess of unlabeled PEDF ( Fig. 6C). Both peptides also bound PC-3 cells ( Fig. 6B). The 34-mer binding to PC-3 cells was displaced by unlabeled PEDF, but not by the 44-mer ( Fig. 6D) also suggesting two PEDF receptors.
PEDF neurotrophic activity maps to the residues 58 to 101, and the 44-mer ( 36), whereas antiangiogenic epitope is elusive. The 44-mer also blocks leakage of the retinal vasculature ( 35) and was earlier identified as an interactive epitope for a putative neurotrophic receptor ( 29, 31, 32) in the acidic surface including NH2-terminal helices, hA and hC ( 36). Assuming that antiangiogenic and neurotrophic receptor-binding surfaces map to the same area, we generated peptides mimicking the middle and lower regions of PEDF and ascribed the two functions to adjacent but distinct epitopes. Two NH2-terminal peptides, TGA and the 34-mer, induced apoptosis, blocked endothelial cell migration and corneal angiogenesis, but failed to induce Y-79 differentiation. The more distal 44-mer was noninhibitory, induced endothelial cell chemotaxis, and sustained angiogenesis in vivo. The 44-mer induced neuronal differentiation of retinoblastoma and prostate carcinoma. ERT, its short internal peptide, was both neurotrophic and antiangiogenic. It induced Y-79 differentiation, caused endothelial cell apoptosis, blocked migration and angiogenesis. ERT spans the hC region of the 44-mer and contains three of the four residues deemed critical for blocking vascular leakage ( 35).
A putative 80 kDa neurotrophic receptor (PEDF-RN), interacting with the 44-mer was identified on Y-79 cells, cerebellar and motor neurons, and in neural retina ( 29, 31, 32). Our data suggest that PEDF blocks angiogenesis via a distinct receptor (PEDF-RA), which interacts with the 34-mer/TGA epitope. ERT may interact with PEDF-RN on Y-79 cells causing differentiation. Our hypothesis is supported by the discovery of a 60 kDa putative endothelial cell receptor ( 33). The 44-mer proangiogenic activities imply that there may be endothelial PEDF-RN. Interestingly, antipermeability is typical for some angiogenic stimuli ( 45). Our binding studies also indicate two PEDF receptors on the endothelial and prostate adenocarcinoma cells.
Consistent with our hypothesis, 34- and 44-mer triggered distinct events in remodeling endothelial cells. The 34- but not the 44-mer activated JNK, increased inactive NFATc2, and reduced NFAT binding to the endogenous c-FLIP promoter, thus blocking proangiogenic survival signal. Thus, two PEDF functional epitopes cause divergent signals. JNK activation was reproduced by the short antiangiogenic peptides and was lost upon mutation of the charged LVEEED stretch.
We found that forced expression of PEDF fragments inhibited in vivo growth of PC-3 prostate adenocarcinomas. This result is consistent with PEDF antiangiogenic function ( 2, 27, 46) and with its role as a unique regulator of prostate growth ( 25, 27). Interestingly, the PEDF peptides used distinct anticancer pathways. The 34-mer decreased tumor microvessel density and increased apoptosis, whereas the 44-mer lacked antiangiogenic effects but prompted neuroendocrine differentiation of PC-3 cells.
Scattered neuroendocrine cells in the prostate form dendrite-like processes, express neuronal markers (NSE and ChrA) and secrete neuropeptides (reviewed in ref. 47). Neuroendocrine differentiation can be induced by interleukin-6 (IL-6), serum- and androgen deprivation; ectopic IL-6 or its constitutively active receptor induce neuroendocrine differentiation via cyclic AMP-dependent pathway ( 48). Androgen withdrawal causes neuroendocrine differentiation in androgen-dependent prostate adenocarcinomas, whereas androgen represses neuroendocrine transdifferentiation ( 49). This repression may be explained, at least in part, by testosterone-driven PEDF down-regulation ( 27). In our study, PC-3 N-ter and the 44-mer decreased K8, a luminal epithelial cell marker. Consistent with less pronounced luminal epithelial phenotype, overexpression of N-ter PEDF fragment and of 44-mer by PC-3 cells induced dendrite-like processes suggestive of a neuroendocrine differentiation ( 47) and increased the message for ChrA and NSE, whereas the 34-mer had no effect. Moreover, expression of the N-ter or 44-mer in PC-3 tumors was associated with high serum ChrA and NSE levels in tumor-bearing mice, indicative of neuroendocrine differentiation. The exogenous PEDF and 44-mer, but not the 34-mer, also increased the message for gastrin-releasing peptide, another neuroendocrine differentiation marker. Thus, the effects of PEDF on prostate adenocarcinomas are twofold, an angiogenesis blockade via the 34-mer induction of endothelial cell death, and transdifferentiation by the 44-mer. Unlike Doll et al. ( 27), we observed no apoptosis by PEDF in prostate adenocarcinoma cells, possibly because different cell lines were used.
The role of neuroendocrine differentiation in prostate adenocarcinoma progression is not clear and the link between neuroendocrine differentiation and poor prognoses is often disputed ( 47, 50– 52). On one hand, neuroendocrine cells cease to divide ( 49) and several neuroendocrine differentiation–promoting agents delay prostate adenocarcinoma progression in mouse models. For instance, melatonin increases prostate adenocarcinoma transdifferentiation and decreases proliferation ( 53). Interestingly, it is lower in the aging population, where prostate cancer is prevalent. Constitutive activation of the IL-6 receptor also delays prostate adenocarcinoma growth ( 48). In contrast, neuroendocrine cells escape apoptosis by tumor necrosis factor-α, thapsigargin, or Ca2+ influx ( 54) due to increased antiapoptotic proteins survivin and clusterin, calreticulin, and SERCA 2b ( 54– 56). Moreover, LNCaP xenografts regress in castrated mice but resume growth in the presence of neuroendocrine tumors. Secretions of neuroendocrine cells augment AR induction by testosterone ( 57), suggesting that neuroendocrine differentiation contributes to AR hypersensitivity and promotes the hormone refractive state. In our model, prostate adenocarcinoma cells expressing PEDF or the 44-mer displayed both neuroendocrine differentiation features and slower growth, thus the overall effect of neuroendocrine differentiation was beneficial.
In conclusion, mapping of PEDF functional surface localized its antiangiogenic activity to the N-proximal regions, and identified a critical stretch within the NH2-terminal area. Our results suggest PEDF interaction with two putative receptors, PEDF-RA and PEDF-RN differentially presented by the endothelial and tumor cells. We sought the contribution of the PEDF functional surfaces to its antitumor activity and showed PEDF neurotrophic epitope to promote neuroendocrine differentiation in prostate adenocarcinomas. These findings add a new dimension to the known PEDF activities, expand the knowledge of its structure and function and define PEDF short functional fragments as candidate anticancer agents. Because the 44-mer induces neuroendocrine differentiation, whose role in prostate adenocarcinoma growth requires further studies, its utility for cancer treatment remains questionable. However, the antiangiogenic 34-mer, TGA, and their peptide and nonpeptide derivatives provide an attractive new option for the treatment of prostate cancer and angiogenesis-dependent disease.
Grant support: NIH grants HL 068033-04 (O.V. Volpert), and GM47522 (K. Volz). Y. Mirochnik was supported by an NIH training grant T32 DK 62716-02.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank A. Chlenski, S. Cohn, B. Jimenez, and Z. Wang for helpful discussion and Dr. A. Rademaker (Northwestern University Biostatistics Core) for help with the statistical analysis.
- Received November 3, 2004.
- Revision received March 28, 2005.
- Accepted April 4, 2005.
- ©2005 American Association for Cancer Research.