**3. Discussion**

In the present study, we examined whether CRPC proliferation is repressed by LSD1 and targeted this molecule using a specific inhibitor, NCL1. The inhibition of LSD1 by NCL1 led to an increase in H3K4me2 modifications in the promoter region (Figure 2A) and induced increases in P21 protein expression (Figure 2A). NCL1 significantly inhibited growth in vitro (Figure 2C) as well as tumor growth ex vivo at low concentration levels (Figure 4A). In addition, adverse events were not noted in the general condition of the mice and in their blood analyses (Table 1-2). These results demonstrated the safety and efficacy of NCL1 in CRPC, highlighting its potential as a new treatment for this disease. We are the first to demonstrate the therapeutic potential of NCL1 ex vivo using a CRPC animal model.

Currently, CRPC patients are treated by ADT, including AR- and non-AR-targeting drugs [19].However, the selection of more appropriate treatment and the sequencing of these drugs is increasingly being investigated. Above all, NED phenotypes have been identified in about 50% of cases of CRPC, which express NED markers such as synaptophysin and chromogranin [20]. The presence of NED markers has been shown to indicate a poorer prognosis when treated with AR-targeting drugs, including enzalutamide [21,22]. In previous reports, including our previous article, the overexpression of LSD1 in prostate cancer was shown to be a predictive marker for aggressive tumor biology and tumor recurrence during therapy [16,23,24]. However, reports describing changes in LSD1 expression using consecutive pre- and post-ADT specimens do not exist. Our immunohistochemical analyses demonstrated that the overexpression of LSD1 in aggressive cancer was maintained in castration-resistant cancer cells (Figure 1A–F). In addition, although only in one case, the overexpression of LSD1 was maintained after changes to an NED tumor (Figure 1G–J).Furthermore, using a PCai1 ex vivo model, a high level of LSD1 expression was maintained from 1 week after castration (Figure 4D,E), and cell growth of castration-resistant PCai1 cells was effectively suppressed by NCL1 both in vitro and ex vivo. Therefore, NCL1 may have therapeutic potential for CRPC, including NED phenotypes, from an early phase after the acquisition of castration resistance.Further studies are needed to clearly test its in vivo potential in combination with ADT, as well as with AR-targeting drugs.

We observed high protein expression of LSD1 in CRPC cells (Figure 2B), and the inhibition of LSD1 activity using NCL1 to reduce cell proliferation in vitro (Figure 2C). In addition, the common mechanism of cell death induced both in vitro and ex vivo by NCL1 was revealed to be caspase-dependent apoptosis (Figure 2D,E and Figure 4A,B,H,I). Apoptosis is an active cell suicide process that maintains cellular homeostasis; however, cancer cells can override apoptotic cell death by upregulating anti-apoptotic machinery and/or downregulating pro-apoptotic programs [25,26]. It is generally accepted that autophagy can function as an adaptive response to maintain cell survival and

growth [27,28]. A recent report reveals that the mammalian target of rapamycin (mTOR) inhibition protects cancer cells from apoptosis during nutrient limitation [29]. Several studies have established that autophagy is associated with drug resistance in prostate cancer cells to ADT and inhibitors of PI3K/Akt/mTOR signaling [30–32]. In a previous study, we reported that LSD1 inhibition stimulated autophagy in castration-naïve prostate cancer [16]. Therefore, to confirm this phenomenon in CRPC, we examined drug-induced autophagy in 22Rv1 cells by detecting LC3-II expression, as well as LysoTracker analysis, a non-specific autophagic marker, using TEM and Western blotting. We showed that NCL1 induced autophagy in 22Rv1, PC3, and PCai1CS cells in a concentration-dependent manner (Figure 2D). WST-8 assay revealed that the anti-tumor effect of NCL1 was reinforced when autophagy was inhibited by CQ in 22Rv1 cells. In addition, combination index analysis revealed that a combination of these drugs showed a synergistic effect (Figure 3F). These results sugges<sup>t</sup> that the stimulation of autophagy in CRPC protects cells against anti-tumor agents through LSD1. Also, when treated in combination with drugs that regulate autophagy, NCL1 may be more effective in the suppression of CRPC growth.

LSD1 plays a key role in many physiological functions. Previous studies have described how LSD1 inhibition reduces cell growth by affecting the expression of several genes involved in proliferation and the cell cycle [33–37]. Previous reports described that estrogen-induced demethylation of H3K9me2 by LSD1 caused reactive oxygen species-induced DNA damage and subsequently caused apoptosis via the regulation of phosphorylation with DNA damage repair enzymes in hormone-responsive cells [38,39]. In addition, a recent report described how a combination of LSD1 knockdown and cisplatin effectively suppressed the proliferation of PC3 cells, and that vascular endothelial growth factor, one of the most important promoters of angiogenesis, was downregulated by LSD1 siRNA treatment [40]. Our results sugges<sup>t</sup> a similar mechanism. The inhibitory mechanisms of cancer growth by NCL1 appeared to be related not only to direct effects on cell proliferation but also to effects on angiogenesis, as shown by a reduction in CD 31-positive vessels ex vivo (Figure 4C,J,K). Recently, an abnormal mRNA splice variant of the androgen receptor, called AR-V7 [41,42], was shown to convey resistance to ADT [43,44]. In addition, in patients with metastatic CRPC, the presence of detectable AR-V7 transcripts in circulating tumor cells has been associated with a high positive predictive value for a non-response to AR-targeting agents, including enzalutamide, in several studies [45,46]. Prospective trials are ongoing to develop the best biomarker strategy for identifying treatment-resistant patients. Interestingly, in 2018, Regufe da Mota et al. [47] reported that LSD1 inhibition caused attenuation of the expression of not only wild-type AR, but also AR-V7. Further prospective trials using biomarkers to help select patients are warranted to evaluate the benefits of a strategic sequence of several drugs, including NCL1, for patients with CRPC.
