Growth and Survival Mechanisms Associated with Perineural Invasion in Prostate Cancer

INTRODUCTION

Perineural invasion (PNI) is the process by which cancer cells wrap around nerves. It was first described by Ernst in 1905 (1) . It is seen most frequently in certain types of cancer such as prostate, bile duct, and pancreatic carcinomas as well as head and neck cancers. Because of their predilection for nerves, these cancers are known as “neurotropic cancers.” For approximately 60 years, researchers believed that PNI was an expression of tumors cells traveling through lymphatic vessels within the perineural space. Rodin et al. (2) demonstrated that the perineural space was devoid of lymphatic vessels and, in the absence of a biological explanation, concluded that PNI was a purely physical phenomenon. Cancer cells found a pre-established dissemination route of “least resistance” in the perineural space (2) .

Prostate cancer currently is the most common noncutaneous malignancy and the second leading cause of cancer death in American men (3) . Mortality in prostate cancer patients is generally attributable to extracapsular spread, which often results in treatment failure and is associated with poor prognosis (4) . Perineural invasion is reported in 85% of prostate cancer patients. More importantly, at least 50% of cases involving extracapsular extension occurred by spread of the cancer within the perineural spaces (5) . Accordingly, current theories suggest that PNI is a key process for extraprostatic spread in prostate cancer.

A previous study with a small number of cases has reported a decrease in apoptotic rate and increased proliferation in cancer cells in a perineural location as compared with cells away from the nerve (6) . However, little is understood about the specific mechanisms that regulate PNI in prostate cancer. We are aware of no new biological studies involving PNI for the last 35 years.

It is likely that a set of key signaling pathways mediate interactions between nerves and cancer cells, which result in PNI and spread of prostate cancer beyond the gland. To address specific mechanisms, we recently developed the PNI in vitro model of PNI (7) . This model showed that specific interactions existed between prostate cancer cells and nerves, which led to costimulation of growth and PNI in vitro. As we reported previously, DU-145 cells and ganglion neuron neurites migrate toward each other, and the cancer cells contact the nerves and migrate in the perineural sheath (7) . Accordingly, the PNI in vitro model is useful in permitting the analysis of specific signaling mechanisms that regulate changes in cell proliferation and survival associated with PNI in prostate cancer. To understand mechanisms involved in PNI, we used the PNI in vitro model and human prostate cancer tissue arrays to identify and assess specific factors that might regulate this biology. We report here that prostate cancer cells in a perineural position exhibit a significantly reduced rate of apoptosis and an increased rate of proliferation. Furthermore, we show that up-regulation of nuclear factor κB (NFκB) and its downstream targets PIM-2 and defender against cell death 1 (DAD-1), all components involved in antiapoptosis signaling cascades, is associated with prostate cancer cells in PNI and most likely regulates the inhibition of apoptosis.

MATERIALS AND METHODS

Perineural Invasion In vitro System.

The PNI in vitro coculture model consists of human prostate cancer cells (DU-145) and mouse ganglia/nerves embedded in Matrigel following the conditions we have published previously (7) . Briefly, dorsal root ganglia/nerves (DRG) from 4- to 7-week-old C57BL/6 mice were removed surgically and cocultured with DU-145 cells in EHS Matrigel (Becton Dickinson, Bedford, MA). Cultures were grown in RPMI 1640 medium (GIBCO, Grand Island, NY) containing 10% heat-inactivated fetal bovine serum (GIBCO) in a humidified atmosphere of 5% CO₂. DU-145 alone and DRG alone were cultured as controls. Between days 9 and 11, some cultures were fixed (3.7% buffered formalin) and processed for immunohistochemistry. Identical cultures were processed for RNA extraction. Residual conditioned media (48 hours) from all cultures were used fresh in the microchamber system.

Microchamber System.

DU-145 human prostate carcinoma cells were obtained from American Type Culture Collection (Manassas, VA) and cultured in RPMI 1640 supplemented with fetal bovine serum (10%), penicillin (100 units/ml), and streptomycin (100 μg/ml; Sigma, St. Louis, MO). They were maintained at 37°C in a humidified atmosphere of 5% CO_2. DU-145 cells were cultured in microchamber culture slides (VWR) as monolayer adherent cells for 11 days to approximately 70–80% confluence. The cultures were then switched to 50% fresh medium and 50% conditioned medium transferred from the PNI coculture model and controls (DU-145 cell growing alone; day 7). Fresh RPMI 1640-based culture medium was also used as a control. Genistein (Sigma) was dissolved in 0.1 mol/L Na₂CO₃ to make a 10 mmol/L stock solution and added directly to the culture medium at a final concentration of 50 μmol/L. Samples were treated with genistein or vehicle control for 96 hours. All samples were cultured for a total of 10 days and then fixed in 3.7% buffered formaldehyde and processed for in situ apoptotic body labeling.

BAY 11-7085 (Calbiochem, La Jolla, CA) was dissolved in dimethyl sulfoxide to make a 10 mmol/L solution and added directly to the culture medium at a final concentration of 10 μmol/L. Samples were treated with BAY 11-7085 or vehicle control also for 96 hours.

Labeling and Hybridization of cDNA for Microarray Analysis.

Microarray analysis was performed using 30 μg of total RNA. cDNA reverse transcription and fluorescent labeling reactions were carried out using Cy3-labeled nucleotides for control and Cy5-labeled nucleotides for experimental samples. Briefly, cDNA synthesis was initiated by incubating RNA with nonlabeled oligo(dT) primer (1.5 μg; Life Technologies, Inc.) in a 39-μl total volume at 70°C for 10 minutes and chilled on ice. Then, Cy3- and Cy5-labeled dCTPs (Amersham Life Science, Arlington Heights, IL) were added to the appropriate reactions. To each reaction, we added 5× first-strand buffer [12 μl; 250 mmol/L Tris-HCl (pH 8.3), 375 mmol/L KCl, and 15 mmol/L MgCl₂), unlabeled nucleotide mix (3 μl; 1 mmol/L dATP, dGTP, and dTTP and 0.1 mmol/L dCTP), 0.1 mol/L dithiothreitol (6 μl), 3 μl of Superscript II reverse transcriptase (200 units/μl), and RNase inhibitor (6 μl; 20 units/μl). After incubation at 42°C for 2 h, cDNA was denatured and neutralized by adding 3 μl of 5 mol/L NaOH and incubation at 37°C for 10 minutes, followed by the addition of 15 μl of Tris-HCl (pH 7.5) and 3 μl of 5 mol/L HCl. The mixture of Cy3 and Cy5 reactions was used as probe and purified using Qiagen PCR purification kit (Qiagen, Valencia, CA). The probe was then mixed with an equal amount of ultrahyb hybridization buffer (Ambion, Austin, TX) and denatured by a 2-minute incubation in boiling water to hybridize in Genomic Systems Hybridization station for 4 hours at 42°C against a microarray chip carrying 6,000 cDNAs obtained from Baylor Microarray Core Facility. 4 After hybridization and post-hybridization washes, the slide was scanned immediately in an Axon 4000A dual channel scanner (Axon Instruments, Foster City, CA), and the data were analyzed using the Gene Pix version 3.0 software package (Axon). Genes were considered up- or down-regulated if expression was changed at least 2-fold as compared with the control. Data with low signal intensity, high background, and high variability were eliminated.

First-Strand cDNA Synthesis.

First-strand cDNA were made from total RNA by using RETROscript First Strand Synthesis Kit (Ambion). One microgram of each RNA sample and control template RNA (Ambion) was heated to 75°C for 3 minutes in nuclease-free water containing 2 μl of random decamers. Two microliters of 10× reverse transcription buffer, 4 μl of 10 mmol/L deoxynucleoside triphosphate mix (10 mmol/L each dATP, dCTP, dGTP, and dTTP), 1 μl of placental RNase inhibitor, and 1 μl of Moloney murine leukemia virus reverse transcriptase (100 units/μl) were added to the sample. Each reaction volume was adjusted to 20 μl by adding nuclease-free water. The reaction mixture was incubated at 44°C for 1 hour, followed by incubation at 92°C for 10 minutes to inactivate the reverse transcriptase.

Real-Time Quantitative Polymerase Chain Reaction.

The ABI Prism 7000 sequence detection system (Applied Biosystems, Foster City, CA) was used for quantitative PCR analysis using hypoxanthine phosphoribosyltransferase 1 (HPRT1) as endogenous control. All probes (6-FAM dye labeled at the 5′ end and a nonfluorescent quencher at the 3′ end of the probe) were combined with primers (Assay-on-Design; Applied Biosystems). Validation experiments were performed using 1:2 diluted templates. The log input amount of RNA versus ΔCt was generated to demonstrate that the efficiencies of the targets and references were approximately equal. DAD-1, NFκB, PIM-2 (sequences available on request), and HPRT1 genes were amplified in separate wells in duplicate. Reaction conditions included 10 μl of 2× TaqMan Universal Master Mix (with UNG), 1 μl of HPRT1 or target primers and probes mixture, 100 ng of template cDNA (DU-145 alone, PNI coculture, and PNI coculture with genistein), and nuclease-free water to a 96-well reaction plate. The total reaction volume is 20 μl. The TaqMan cycling conditions were as follows: 2 minutes at 50°C; 10 minutes at 95°C, and 40 cycles of 15 seconds at 95°C followed by 1 minute at 60°C.

Calculation was performed using values of Δ cycle threshold (ΔCt) as the calibrator from DU-145 cells cultured alone or from DU-145 cells cocultured with DRG. For each experimental sample, the Ct of target, normalized to an endogenous reference and relative to a calibrator, is given by: 2^−ΔΔCt.

Cohort Enrollment and Tissue Microarrays.

More than 4,000 patients underwent radical prostatectomies at one of the Baylor College of Medicine-affiliated institutions and willingly provided tissues (IRB H-1158). A single pathologist (T. M. W.) performed the pathological analysis of the radical prostatectomies. Of these, 226 were used to build tissue microarrays with the manual tissue arrayer (Beecher Instruments Microarray Technology, Silver Spring, MD). Entry criteria included: (a) no preoperative treatment; (b) surgery between 1983 and 1998; (c) prostate cancer present in the surgical specimen and large enough to be cored (2-mm cores) for microarrays; and (d) PNI present and large enough to be cored. A smaller array using tissues from 50 patients was also built.

Whole mount slides were examined under light microscope, and areas of nerves with PNI were marked. Areas of cancer far away from the nerve were also selected [non-PNI (NPNI)]. One tissue cylinder (2 mm) was punched from the PNI area and the NPNI area of each specimen and transferred to an empty recipient paraffin block. The Baylor Institutional Review Board (IRB H-13024) approved this study.

In situ Labeling of Apoptotic Bodies.

The detection of DNA fragmentation was determined in situ by the terminal deoxynucleotidyl transferase-mediated nick end labeling (TUNEL) technique as described previously (8) . We used the TACS·XL-diaminobenzidine in situ apoptosis kit (Trevigen Inc., Gaithersburg, MD) following the manufacturer’s indicator with minor modifications and counterstained with methyl green. The TUNEL assay and immunohistochemistry were performed on the tissue microarray slides and paraffin-embedded sections of PNI coculture and controls. A positive control slide prepared by TACS-nuclease and a specimen known to be positive for apoptotic cells were used as positive controls. Substitution of terminal deoxynucleotidyl transferase with distilled water was used as a negative control.

Apoptotic bodies were counted under a light microscope (×400) equipped with an ocular grid (10,000 μm²). The area with highest positive stain was selected for counting. The number of apoptotic bodies was determined in a total of approximately 2,000 prostate cancer cells, normalized to 100 cells, and defined as apoptotic index (AI). At least 10 representative areas without necrosis were selected. Positively stained cells or bodies located in the stroma and lumen were excluded because these apoptotic cells or bodies might have originated from other cell types.

Immunohistochemistry.

The proliferation rate of prostate cancer cells in human tissues and in the PNI in vitro model was determined using the immunoperoxidase method with antibodies against Ki-67 (Santa Cruz Biotechnology, Santa Cruz, CA). Briefly, the slides were deparaffinized and rehydrated in graded ethanol, and antigens were retrieved using steam in 10 mmol/L citrate buffer (pH 6.0) for 30 minutes. The slides were subsequently blocked with blocking protein (DAKO, Carpinteria, CA) for 10 minutes and then incubated with the polyclonal antibody Ki-67 (C-20), for 1 hour at room temperature at a dilution of 1:50. The secondary biotinylated antibody was applied for 30 minutes followed by a 30-minute incubation with streptavidin peroxidase (DAKO LSAB+HRP kit). After rinsing, slides were visualized by diaminobenzidine chromogen solution (DAKO) and counterstained with routine hematoxylin. Positive staining of Ki-67 was confined to the nucleus. The proliferation index (PI) was defined as the ratio of KI-67-positive cancer cells to total cancer cells in the highest positive stain fields (at least 2,000 cells), using a microscopic grid at ×400 magnification.

Expression of NFκB, PIM-2, and DAD-1 protein in histological sections was analyzed in the smaller tissue array by immunohistochemistry, as described above, with the following antibodies: anti-NFκB (Santa Cruz Biotechnology; sc-8008; mouse monoclonal IgG; 1:10 dilution; 30′ incubation time); PIM-2 (Santa Cruz Biotechnology; sc-13514; mouse monoclonal IgG; 1:10 dilution; 30′ incubation time); and DAD-1 (Santa Cruz Biotechnology; sc-12173; goat polyclonal IgG; 1:10 dilution, 30′ incubation time). NFκB expression was also analyzed in the large PNI array (226 patients). Cancer cells in PNI or NPNI areas were interpreted for immunoreactivity using a 0 to 3+ quantitative scoring system for both stain intensity and percentage of positive cells (labeling frequency percentage). For intensity, the grading scale ranged from no detectable signal (0) to strong signal seen at low power (3). A designation of 2 was assigned to a moderate signal seen at low to intermediate power, and a designation of 1 was assigned to a weak signal seen only at intermediate to high power.

Nuclear expression of NFκB was also quantified in the large PNI tissue microarray using a 0 to 3 scale. Zero correspond to no nuclear staining, 1 is faint nuclear staining, 2 is strong staining observed at medium power in <20% of the cells in the PNI cancer, and 3 is strong staining observed at medium power in >20% of the cells.

Immunofluorescence.

Tissue culture samples were rinsed in Tris-buffered saline followed by a 60-minute incubation with primary antibodies [NFκB, 1:50 mouse monoclonal antibody; DAD-1, 1:50 goat polyclonal antibody, (Santa Cruz Biotechnology)]. After a second rinse, the secondary antibodies [fluorescein isothiocyanate-conjugated rabbit antimouse (1:50) and Texas red-conjugated donkey antigoat (1:50), Zymed Laboratories] were added for 30 minutes and rinsed. 4′,6-Diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA) was used as a counterstain. Observation was performed using a fluorescent inverted Nikon microscope and a Q-capture imaging system.

Statistical Analysis.

Statistical analysis was conducted using Student’s t test for analysis of significant difference between different values and was considered significant at a 95% confidence interval (P < 0.05). The differences in NFκB labeling frequencies were compared between PNI cancer and NPNI specimens by using the matched pair Wilcoxon signed-rank test. The correlation of NFκB expression with patients’ clinical and pathological variables was analyzed by the Spearman or Pearson correlation test. The predictive value of NFκB for recurrence-free survival was determined using the Kaplan-Meier actuarial analysis and the log-rank test. In addition, the Cox proportional hazard regression model was used to analyze the value of using NFκB and other pathological and clinical markers to predict the risk of recurrence. The risk ratio and its 95% confidence interval were recorded for each marker. P values of <0.05 were considered statistically significant in all of our analyses. All analyses were performed with statistical software (SPSS 11.0).

RESULTS

Increased Proliferation and Reduced Apoptosis in Prostate Cancer Cells in the Perineural Space.

Matched pair analysis of the human tissue microarrays showed that prostate cancer cells associated spatially with nerves (PNI) exhibited a significant increase in PI as compared with prostate cancer cells in NPNI sites, as shown in Fig. 1 ⇓ . The PI in human adenocarcinoma cells at PNI sites had a mean value of 10.25 ± 1.2, whereas the PI in NPNI sites had a mean value of 8.16 ± 0.98 (P = 0.034). The AI in cancer cells at sites of PNI was significantly lower as compared with AI in prostate cancer cells not associated with PNI. The mean AI at PNI sites was 0.68 ± 0.16 as compared with a mean AI of 1.8 ± 0.42 at sites of cancer with no PNI (P = 0.010; Fig. 2 ⇓ ).

Results with the PNI in vitro model showed similar differences when prostate cancer cells were cocultured with dorsal root ganglion nerve tissue. Analysis of PI and AI between DU-145/DRG cocultures and control DU-145 cells showed that the PI in cocultures exhibited a mean of 9.87 ± 2.1 (n = 12), whereas the mean PI in control was lower at 6.05 ± 2.0 (n = 12; P = 0.001). The AI in the DU-145/DRG coculture was 11.28 ± 2.22, whereas the AI in control DU-145 cells was significantly higher at 27.22 ± 4.33 (n = 12; P = 0.004; Fig. 2 ⇓ ).

Overexpression of Genes Related to Proliferation and Apoptosis in Prostate cancer in Perineural Location.

Profiling of genes differentially expressed in DU-145/DRG cocultures relative to DU-145 controls by gene microarray analysis showed overexpression (of at least a 3-fold differential) of 15 genes. Of these, three were known components of antiapoptosis signaling pathways. These included NFκB and its downstream targets PIM-2 and DAD-1. Overexpression of these factors in perineural cancer cells relative to cancer cells not associated with nerves was confirmed by quantitative PCR and showed that NFκB, DAD-1, and PIM-2 were each overexpressed 1.9-, 1.6-, and 2.02-fold, respectively, in DU-145/DRG cocultures as compared with DU-145 cells growing alone (Fig. 3) ⇓ .

Confirmation that protein was also overexpressed was shown using immunostaining. Immunofluorescence studies with the DU-145/DRG cocultures demonstrated that NFκB is predominantly cytoplasmic in the DU-145 cells cultured without the nerve (control), whereas an overall increase in staining intensity and increased nuclear localization (nuclear translocation) was observed in DU-145 cells cocultured with DRG nerves in the PNI model (Fig. 4) ⇓ . In addition, overexpression of DAD-1 protein was evident in DU-145/DRG cocultures as compared with DU-145 controls (Fig. 4) ⇓ , and DAD-1 exhibited colocalization with NFκB in the nucleus. Immunohistochemistry for NFκB, DAD-1, and PIM-2 on paraffin-embedded tissues of the PNI coculture samples confirmed the quantitative PCR and immunofluorescence results (data not shown).

Examination of the human tissue microarray slides showed that the staining intensity of PIM-2 was also elevated in approximately 30% of the prostate cancer PNI sites (data not shown).

Reversal of Perineural Invasion Antiapoptotic Effect by Genistein and BAY 11-7085.

DU-145 cells exposed to 50% (v/v) of conditioned medium derived from DU-145/DRG cocultures also showed a decrease in the AI relative to vehicle controls. There was a decrease in the mean AI in the cells cultured with PNI coculture-conditioned medium (AI = 11.02; n = 8) compared with cells cultured with control DU-145-conditioned medium (AI = 13.52; n = 8; P < 0.000; Fig. 2C ⇓ ). The AI index showed a significant increase with the addition of genistein (AI = 17.32; P < 0.000; n = 16). The addition of genistein also lowered the levels of its effector targets, NFκB (0.62×), PIM-2 (0.57×), and DAD-1 (0.77×; Fig. 3 ⇓ ). The decrease at the RNA level was translated into a decrease in protein expression as demonstrated by immunofluorescence (data not shown).

The addition of the inhibitor of IκB phosphorylation, BAY 11-7085, demonstrated similar results. In contrast to controls (DU-145 cocultured with nerves), nuclear translocation of NFκB was not identified by immunofluorescence of the PNI in vitro model treated with BAY 11-7085 (Fig. 5) ⇓ . Quantitative PCR showed a decrease in NFκB and its downstream effectors DAD-1 and PIM-2. NFκB RNA levels in BAY 11-7085-treated cells were 0.60 those of control. Similarly, DAD-1 RNA levels in BAY 11-7085-treated cells were 0.85 those of control. We also identified a decrease in PIM-2 in BAY 11-7085-treated cells (0.48) as compared with controls (Fig. 3) ⇓ . More significantly, the apoptotic ratio increases in BAY 11-7085-treated cells, indicating a functional response to the IκB phosphorylation inhibition. The mean AI increase in cells treated with BAY 11-7085 was 125.5 ± 5.44 (n = 4) as compared with control DU-145-conditioned medium (AI = 104.25 ± 10.14; n = 3; P = 0.0164; Fig. 5 ⇓ ).

Biological Significance of Nuclear Translocation of Nuclear Factor κB in Human Perineural Invasion.

Data were available in 156 patients, due to core loss and/or lack of PNI cell in the core. Overall NFκB levels were higher in the cells in PNI than in those located away from the nerve (index, 7.18 ± 2.39 versus 2.34 ± 2.5; P = 0.000; Fig. 6B ⇓ ). Cytoplasmic expression of NFκB was present in all cells in PNI. Nuclear expression was identified in 72.6% of the patients (98 of 135); however, strong nuclear expression (grades 2 and 3) was present in only 38.5% and 8.1% of the patients, respectively (Fig. 6A) ⇓ . Strong nuclear NFκB expression in PNI correlated with other clinicopathological parameters including extracapsular extension (r² = 0.213; P = 0.131) and Gleason score (r² = 0.235; P = 0.006).

Strong levels of nuclear NFκB expression, defined as grade 2 and 3, were associated with a decrease biochemical-free survival. There was a significant difference in recurrence-free survival between cases with no expression or low expression of nuclear NFκB in the PNI (mean survival, 78.62 months) and those with strong expression [median survival, 67.52 months; hazard ratio = 2.053 (1.123–3.755); P = 0.0195; Fig. 6C ⇓] . This difference was significant on univariate analysis but not on multivariate analysis.

DISCUSSION

PNI was described as a morphological entity in the early 1900s (1) . However, biological studies were never accomplished, probably due to the lack of a model. This study is the first to show biological mechanisms associated with PNI. We have demonstrated that PNI in prostate cancer resulted in inhibition of apoptosis and increased proliferation in the cancer cells associated with nerves, both in human prostate tissues and in the PNI in vitro model. In addition, expression of NFκB and its downstream effectors DAD-1 and PIM-2 was elevated in perineural cancer cells. The addition of genistein or BAY 11-7085 reversed alterations in apoptosis and lowered RNA levels of NFκB, PIM-2, and DAD-1, suggesting that the NFκB signaling pathway regulates the alterations observed in PNI. Furthermore, nuclear translocation of NFκB was associated with decreased biochemical recurrence-free survival, corroborating the significance of the phenomenon in human tissues.

Neurons have numerous interactions with the epithelial and stromal components of the prostate gland. They are involved in prostate development and maintenance of histological architecture and function in the adult gland (9, 10, 11, 12, 13, 14) . However, the most common interaction between nerves and prostate epithelium during tumorigenesis is PNI. The biological and clinical significance of PNI in prostate cancer has been the subject of intense debate (15, 16, 17, 18) . Numerous other authors have reported varying degrees of clinical prognostic significance for PNI (15 , 17 , 19, 20, 21, 22, 23) . A significant study demonstrated that the volume of tumor around the nerve, labeled as the PNI diameter, was an independent predictor of survival (24) . The greater the PNI diameter, the stronger the risk of recurrence for the patient becomes. These data suggest that that not all PNIs are equal and that proximity to the nerve might provide greater growth and survival advantages in various degrees.

These studies, taken together, suggest that cancer cells in the perineural environment are influenced by the nerve to evolve a growth and survival advantage, the manifestation of which is increased tumor volume around the nerve. It follows that cancers that exploit the advantages found in the perineural microenvironment would be predicted to have a more aggressive biology and therefore the worst survival. These findings show the critical importance of studying the cancer microenvironment, including the neuroenvironment, as a key regulatory component of cancer progression. It seems likely that cancer cells have different biological abilities according to the elements that surround them. It is possible that the perineural space creates an environment that not only provides an avenue for prostate cancer invasion and spread but also creates the conditions for heightened aggressiveness that has been shown to be associated with PNI in prostate cancer. Data presented here support this hypothesis and suggest that the NFκB pathway is key in mediating this biology.

NFκB is a known regulatory element in a major survival pathway. In the present study, we demonstrate that NFκB is up-regulated in the PNI in vitro model at the RNA and protein levels. The data also strongly suggest that NFκB is translocated to the nuclei in prostate cancer cells in a perineural location. This process occurs in the in vitro PNI model as well as in human tissues. The latter is corroborated by the fact that strong nuclear NFκB expression in human perineural cells is associated with decreased biochemical recurrence-free survival. The lack of multivariate significance could be due to a number of factors such as a small population and the relatively strong correlation between nuclear NFκB and extracapsular extension and Gleason score. More importantly, this finding is a clear demonstration that nuclear expression/translocation is clearly related to a more aggressive phenotype of prostate cancer and is a clear demonstration of the biological significance of nuclear translocation of NFκB in the perineural space.

PIM-2 and DAD-1 are downstream targets in the NFκB survival pathway (25) . Both have been associated with antiapoptotic functions. Data presented here show up-regulation of these genes. Functional studies with genistein, a known inhibitor of constitutive and inducible NFκB activation (26 , 27) , and BAY 11-7085, a specific inhibitor of IκB phosphorylation (28) , suggest that the NFκB and its downstream effectors are likely to be a key regulatory component in PNI. Both genistein and BAY 11-7085 seem to inhibit the up-regulation of NFκB as well as PIM-2 and DAD-1. Nuclear translocation seems to be inhibited and is best noted with BAY 11-7085, as would be expected with a molecule that inhibits phosphorylation of IκB. More importantly, the functional consequence of the activation of the NFκB pathway in the perineural space, inhibition of apoptosis, is partially reversed with both compounds. Accordingly, targeting the NFκB pathway may be a novel therapeutic approach to limit metastatic spread via PNI.

The upstream regulatory pathways that affect NFκB in PNI are currently unknown. Whereas neural cell adhesion molecule (NCAM) has been described in numerous cancers (29, 30, 31) , we have not been able to identify NCAM in prostate cancer cells. However, nerves with prostate cancer PNI have higher levels of NCAM as compared with nerves without PNI (32) . The NCAM promoter is known to be activated by neural injury, such as cancer invading the nerve. Krushel (33) demonstrated that NFκB-mediated transcription was increased after NCAM binding to astrocytes and neurons. NCAM homophilic binding at the cell membrane leads to increased NFκB binding to DNA and transcriptional activation in both neurons and astrocytes. Because NCAM and NFκB are present in the post-synaptic density, it is possible that higher levels of NCAM in the nerve might increase NFκB levels in the prostate cancer cells as they contact the nerve in PNI. This represents a potential route allowing direct communication between the cancer cells and the nerves. The relationship between NFκB and NCAM is bidirectional because nitric oxide also stimulates protein binding to a NFκB site in the promoter of the NCAM gene (34) . Other potential upstream regulators are neuropeptides, which are known to be present in prostate nerves. Both bombesin (35) and dopamine (36) have been shown to stimulate NFκB activation.

Our studies showed that PNI is an interactive process between the cancer cells and nerves (7) . In the present study, we demonstrate that this interaction likely results in growth and survival advantages for the prostate cancer cells in the perineural location and provides a putative mechanism for this phenomenon. We believe that this is the first description of a biological mechanism and functional significance associated with PNI. A more clear understanding of the specific biological mechanisms involved in PNI and the inhibition of cancer cell apoptosis may lead to improved diagnostics/prognostics and to the development of new therapeutic strategies.