**2. Results**

### *2.1. LSD1 Is Highly Expressed in CRPC*

To determine the status of LSD1 in human prostate cancer, the expression of LSD1 and Nkx3.1, a sensitive specific marker of differentiated adenocarcinoma originating from the prostate [18], was examined in human prostate biopsy specimens by immunohistochemistry and the staining intensity scored. We found that LSD1 expression levels in CRPC were very high, as previously found in prostate biopsy specimens obtained from patients (Figure 1A–F,K). Interestingly, neuroendocrine-differentiated tumors after androgen deprivation therapy, which had no expression of Nkx3.1, had high levels of LSD1 expression (Figure 1G–J).

**Figure 1.** (**A**–**F**) Hematoxylin and eosin (HE) staining (**A**), and immunohistochemistry for Nkx3.1, a sensitive specific marker of differentiated adenocarcinoma originating from the prostate (**B**), and histone lysine demethylase 1 (LSD1) (**C**) of castration-naïve prostate cancer (castration-naïve PC) specimens obtained by prostate biopsy for an initial diagnosis in patients. HE staining (**D**), and immunohistochemistry for Nkx3.1 (**E**), and LSD1 (**F**) of castration-resistant prostate cancer (CRPC) specimens obtained by prostate biopsy after the acquisition of castration resistance and treatment with androgen deprivation therapy. Nuclei were counterstained with hematoxylin. Scale bar is 50 μm. (**G**–**J**) HE staining (**G**), and immunohistochemistry for Nkx3.1 (**H**), LSD1 (**I**), and synaptophysin (**J**) in prostate biopsy specimens obtained after the acquisition of castration resistance and neuroendocrine differentiation after treatment with androgen deprivation therapy. Scale bar is 50 μm. (**K**) A graphical comparison of intensity levels of LSD1 expression between castration-naïve PC and CRPC biopsy samples. n.s.: not significant.

### *2.2. LSD1 Expression in Prostate Cancer Cell Lines and Suppression of Prostate Cancer Cell Proliferation by NCL1*

First, to determine whether LSD1 inhibition influences specific gene methylation status, 22Rv1 cultured prostate cancer cells treated with NCL1 were subjected to chromatin immunoprecipitation (ChIP) assay. As a result, and consistent with our previous report, NCL1 specifically impaired the demethylation of histone H3 lysine 4 (H3K4me2) at the containing promoter lesion of P21 genes (Figure 2A), reflecting the increased level of fold enrichment compared with the IgG control in ChIP assay, and increased levels of P21 protein expression in Western blot analysis, compared to the control. Next, we examined the status of LSD1 in CRPC cell lines, and found by Western blot analysis that LSD1 protein was highly expressed (Figure 2B). The proliferation of prostate cancer cells was significantly decreased by NCL1 treatment in a dose-dependent manner in the cancer cell lines tested, as determined by cell proliferation assay (Figure 2C). These findings sugges<sup>t</sup> that NCL1 attenuated CRPC cell proliferation by demethylating H3K4me2 via LSD1 inhibition.

**Figure 2.** (**A**) Chromatin immunoprecipitation (ChIP) analysis in 22Rv1 cells using a histone H3 lysine 4 dimethylation (H3K4me2) antibody showed that NCL1 induced the attenuation of demethylation of H3K4me2 in the promoter regions of P21. Western blot analysis of P21 in 22Rv1 cells is shown. The protein expression of P21 was increased, reflecting the results of the ChIP analysis. β-actin was used as an internal loading control. (**B**) Western blot analysis of PCai1CS, 22Rv1, and PC3 cells for LSD1. All castration-resistant prostate cancer (CRPC) cell lines expressed LSD1. β-actin was used as an internal loading control. (**C**) PCai1CS and 22Rv1 cells were treated with vehicle (control) or NCL1, and subjected to WST-8 assay to measure cell proliferation. NCL1 treatment reduced the cell viability of the two CRPC cell lines in a dose-dependent manner. (**D**) Western blot analyses 48 h after NCL1 treatment of 22Rv1, PCai1CS, and PC3 cells. The cell cycle-related protein expression of cyclin B1, cyclin D1, CDK2, CDK4, and p27KIP were unchanged. Treatment with NCL1 resulted in a marked elevation in cleaved caspase 3 without any change in caspase 3. In addition, protein expression of microtubule-associated protein light chain 3 (LC3)-II was elevated in NCL1-treated CRPC cells. β-actin was used as an internal loading control. (**E**) Guava® apoptosis analysis of PC3 and 22Rv1 cells. NCL1, the autophagy inhibitor chloroquine (CQ), and a combination of these drugs induced apoptosis in CRPC cells. Mean ± standard deviation (SD); \* *p* < 0.05, \*\* *p* < 0.001, \*\*\* *p* < 0.0001.

### *2.3. NCL1 Inhibits CRPC Cell Growth by Apoptotic Mechanisms*

To determine how NCL1 induced growth inhibition, proteins involved in the cell cycle and apoptosis were examined in NCL1-treated CRPC cell lines. Reflecting the inhibition of LSD1, the expression of P21 was enhanced (Figure 2A). Cleaved caspase 3 was markedly elevated after treatment with NCL1 but caspase 3 expression remained unchanged. However, examination of cell cycle-related proteins showed that cyclin B1, cyclin D1, cyclin-dependent kinase (CDK)2, CDK4, and p27KIP expression were not changed by NCL1 treatment (Figure 2D). Therefore, analyses by Guava® ViaCount assay were performed. As a result, we found that NCL1 treatment of PC3 and 22Rv1 cells led to a significant induction of apoptosis (Figure 2E). These results suggested that selective attenuation of LSD1 using NCL1 inhibits cell proliferation by caspase-dependent apoptosis.

### *2.4. NCL1 Potentially Regulates Autophagy to Induce Cell Death in 22Rv1 Cells*

The conversion of microtubule-associated protein light chain 3 (LC3)-I to LC3-II and the formation of LC3 puncta were used to determine whether NCL1 treatment induced autophagy in CRPC cells. We found that NCL1 induced an increase of LC3-II protein levels in 22Rv1, PCai1CS, and PC3 cells as determined by Western blotting (Figure 2D). Therefore, to confirm the contribution of NCL1 to autophagy, we then raised the pH of the lumen of lysosomes and/or autolysosomes to inhibit autophagic flux using chloroquine (CQ), thereby preventing autophagic degradation. Flow cytometry revealed that a combination of NCL1 and CQ increased apoptotic cell numbers (Figure 2E). In addition, it was revealed by LysoTracker analysis that NCL1 treatment led to a further accumulation of activated lysosomes (Figure 3B), and the addition of CQ caused an attenuation of the phenomenon (Figure 3D). By WST-8 cell counting assay, CQ alone was shown to have an effect on 22Rv1 cell viability, while CQ enhanced the inhibition of cell growth by NCL1 (Figure 3E). Furthermore, combination index analysis revealed that the force of combination was shown to be synergistic (Figure 3F). NCL1 treatment for 3 h led to the formation of autophagosomes, as shown in Figure 3G. The cytoplasm also showed an increase in structures (shown in the 72 h figure) from 24 h to 72 h; the results obtained with LysoTracker sugges<sup>t</sup> that these structures were lysosomes. CQ treatment led to the inhibition of the degradation of structures incorporated into phagosomes. Using a combined treatment, these findings revealed colocalization (Figure 3G). These results sugges<sup>t</sup> that NCL1 may induce CRPC cell death by regulating autophagy potential in addition to regulating an apoptotic anticancer pathway.

### *2.5. Ex Vivo Regulation of Tumorigenesis by NCL1*

We examined the expression level of LSD1 after castration in a PCai1 subcutaneous tumor model. We found a high level of LSD1 expression that remained unchanged 1 week after castration (Figure 4D,E), and continued at this level for 8 weeks. We next assessed the role of NCL1 in tumor progression ex vivo. After PCai1 cells were injected subcutaneously into castrated nude mice, animals were subsequently treated with vehicle control or 1.0 mg/kg of NCL1. The NCL1-treated group showed a significant inhibition of tumor size compared to vehicle controls (Figure 4A). The size of other organs and body weights were not affected by NCL1 treatment, and differences in the relative weights of organs and blood parameters between the two groups were not found (Tables 1 and 2). Vacuolization was found to be increased in the groups treated with NCL1 compared with the vehicle control group (Figure 4F,G). Mechanisms involved in the inhibition of tumor growth by NCL1 in an animal model were examined using terminal deoxy nucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assays. Increased numbers of TUNEL-positive cells, and therefore apoptosis, were noted after treatment with NCL1 compared with vehicle (Figure 4B,H,I). In addition, we undertook immunohistochemical staining of CD31 to examine tumor vascularity. NCL1-treated tumors were found to have significantly decreased numbers of CD-31 positive blood vessels (Figure 4C,J,K). These results sugges<sup>t</sup> that NCL1 regulates apoptosis to induce cell death and decrease tumor vascularity, both in vitro and ex vivo.

**Figure 3.** (**A**–**D**) Detection of the activation of lysosomes using LysoTracker analysis in 22Rv1 cells. Cells were treated with vehicle control (**A**), 50 μM NCL1 (**B**), 50 μM chloroquine (CQ) (**C**), or with 50 μM NCL1 and 50 μM CQ (**D**). Blue: nuclei, red: lysosomes. (**E**) 22Rv1 cells were treated with 50 μM NCL1 and/or 50 μM CQ. A WST-8 assay, in which the dye absorption rate positively correlated with cell viability, revealed that a combination of NCL1 and CQ decreased cell growth. (**F**) A combination index was calculated from the results of the WST-8 assay in Figure 3E. The combination of NCL1 and CQ showed a synergistic effect. (**G**) Cells were treated with 50 μM NCL1 for 3 h, 12 h, and 72 h. Three hours after NCL1 treatment, the formation of autophagosomes was noted by transmission electron microscopy (TEM). The cytoplasm also showed increased numbers of structures (visible in the 72 h figure) from 24 h to 72 h. Scale bar is 20 μm.

**Figure 4.** (**A**) Tumor growth was significantly inhibited in mice treated with 1.0 mg/kg NCL1 as compared to vehicle controls. (**B**) A terminal deoxy nucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay was performed in NCL1-treated and control mice, and quantified as the mean TUNEL labeling percentage based on at least five randomly selected high-power microscope fields per individual. (**C**) Immunohistochemistry for CD31. Positivity was quantified as the mean number of vessels/mm<sup>2</sup> based on at least five randomly selected high-power microscopic fields per individual. (**D**,**E**) Representative immunohistochemistry of LSD1 in a subcutaneous PCai1 tumor. Uncastrated tumor (**D**), and 1 week after castration (**E**). Scale bar is 50 μm. (**F**,**G**) Hematoxylin and eosin (HE) staining in subcutaneous tumors from vehicle control (**F**) and 1.0 mg/kg NCL1-treated (**G**) mice. Vacuolation (black arrowheads) was increased in the NCL1-treated group compared with controls. Scale bar is 50 μm. (**H**,**I**) TUNEL staining for apoptosis in subcutaneous tumors from vehicle control (**H**) and 1.0 mg/kg NCL1-treated (**I**) mice. White arrowheads indicate TUNEL-positive cells. Scale bar is 50 μm. (**J**,**K**) Representative immunohistochemical images of CD31 in subcutaneous tumors from control (**J**) and 1.0 mg/kg NCL1-treated (**K**) mice. Black arrowheads indicate CD31-positive cells. Scale bar is 50 μm. Mean ± standard deviation (SD); \* *p* < 0.05.

**Table 1.** Relative organ weights at the experiment's termination in a PCai1 mouse tumor model. BW: body weight; R: right; L: left.



**Table 2.** Blood results at the experiment's termination in a PCai1mouse tumor model. AST: aspartate aminotransferase; ALT: alanine aminotransferase; ALP: alkaline phosphatase; T-Bil: total bilirubin; T-Chol: total cholesterol; Crea: creatinine; BUN: blood urea nitrogen; Na: sodium; K: potassium; Cl: chloride; Ca: calcium. Mean ± standard deviation (SD).
