Induction of autophagy
Autophagy is a highly conserved catabolic process that regulates cytoplasmic biomass, organellar abundance, and organellar distribution, removing harmful protein aggregates and intracellular toxins. Moreover, autophagy provides regenerated metabolites to promote tumour cell survival. Autophagy can be regulated in multiple ways. In nutrient-rich conditions, the PI3K/AKT pathway inhibits autophagy through the activation of the mammalian target of rapamycin (mTOR). In the absence of nutrients or growth factors or under hypoxic conditions, autophagy is activated by the AMPK pathway, resulting in the upregulation of the transcription of autophagy genes and the inhibition of mTOR/S6K/4EBP activity by the phosphorylation of TSC2. Autophagy contributes to Enz resistance through the activation of AMPK and inhibition of the mTOR pathway (Farrow, Yang et al., 2014 , Smith and Macleod, 2019 , Nguyen, Yang et al., 2014).
Inhibition ofapoptosis
Apoptosis is a form of programmed cell death that plays a critical role in organism development and tissue homeostasis, resulting in the removal of damaged cells. However, cancer cells have lost their ability to undergo apoptosis, which lead to their uncontrolled proliferation. These cancer cells usually overexpress proteins related to the resistance of apoptosis cascade activation (Pistritto, Trisciuoglio et al., 2016). BCL2 proteins are frequently overexpressed in cancers and are associated with disease progression and treatment resistance. The BCL2 family contains anti-apoptotic proteins, including BCL-2, BCL-XL, MCL-1, and BCL-w, and pro-apoptotic proteins, including BAX, BAK, and BIM (Siddiqui, Ahad et al., 2015).
It has been reported that the anti-apoptotic BCL2 protein family members BCL-XL and MCL-1 are critical factors mediating Enz resistance. BCL-XL and MCL-1 bind and separate pro-apoptotic proteins BIM and BAX, resulting in the inhibition of apoptosis. Moreover, PI3K/AKT signalling is activated by Enz and can suppress BAD, a BH3-only protein that activates pro-apoptotic signalling through the inhibition of BCL-XL, mediating apoptosis evasion. However, different prostate cell lines showed differences in the expression of pro-apoptotic and anti-apoptotic BCL2 family proteins. Another study found that BCL-2 was exclusively upregulated in CRPC cells, but not BCL-XL and MCL-1, and was directly induced by Enz, which enhanced cancer stem cell (CSC)-enriched holoclones (Pilling and Hwang, 2019). The BCL-2-selective inhibitor ABT-199 combined with Enz in LNCaP cells drastically inhibited Enz-resistant CRPC. Thus, targeting BCL2 protein signalling plus Enz is a promising therapeutic treatment for ENZ resistance (Li, Deng et al., 2018). The novel small molecule PAWI-2 reduced the protein levels of BCL-2, BCL-XL, and MCL-1 to normal levels in the presence of Enz. PAWI-2 can inhibit tumour growth either as a single agent or in combination with Enz to overcome Enz resistance in CRPC (Cheng, Moore et al., 2019).
The inhibitor of apoptosis protein (IAP) family also inhibits apoptosis by regulating the activity of caspases, and some IAPs are overexpressed in cancers, mediating tumour pathogenesis, progression, and resistance. The IAP family contains eight members, including NAIP (BIRC1), cIAP1 (BIRC2), cIAP2 (BIRC3), XIAP (BIRC4), survivin (BIRC5), Apollon/BRUCE (BIRC6), ML-IAP/LIVIN (BIRC7), and ILP-2 (BIRC8) (Krajewska, Krajewski et al., 2003). BIRC6 played a vital role in Enz resistance in a high-fidelity, Enz-resistant, patient-derived CRPC tissue xenograft model named LTL-313BR. BIRC6 is an exclusive IAP member among the eight proteins that were upregulated in both Enz-resistant systems. Furthermore, an BIRC6-targeting antisense oligonucleotide (ASO-6w2) inhibited the growth of LTL-313BR xenografts and increased the apoptosis rate (Luk, Shresth et al., 2016). Another IAP inhibitor, AEG40995, expanded the caspase-mediated apoptotic response to Enz by inducing TNF-α signalling. Combining Enz with IAP antagonists may overcome Enz resistance in CRPC (Pilling, Hwang et al., 2017).
Lineage plasticity and phenotype switching
Cancer cell plasticity can be defined as the ability of a cell to substantially alter its characteristics and exhibit a new phenotype that is closer to a unique developmental lineage. Cell plasticity allows tumour cells to reversibly convert to cellular properties independent of the drug-targeted pathways. The increased diversity of tumour cells is associated with therapeutic resistance and metastasis, suggesting that the retention of pluripotent progenitors and the persistence of progenitor cells can re-proliferate resistant and metastatic tumour cells with different phenotypes. PCa is driven by androgens and developsvia ligand-mediated AR signalling. Although early treatment is effective, PCa cells can adapt to androgen deprivation and restore AR signalling, eventually progressing to CRPC, which can be treated by the AR antagonist Enz. However, after long-term AR inhibition, the tumour archetype changes, which results in histological de-differentiation and cell lineage alterations in the form of the EMT and/or neuroendocrine differentiation (Davie, Beltran et al., 2018).
The EMT is a process in which the epithelial phenotype is lost and mesenchymal characteristics are acquired. The expression of EMT drivers (ZEB1, ZEB2, Snail, Twist, and FOXC2) and mesenchymal markers (N-cadherin, fibronectin, and vimentin) in PCa cells was significantly enhanced by Enz treatment, promoting the migration of PCa cells and inducing the transformation of prostate cancer into spindle and fibroblast-like forms (Miao, Yang et al., 2017). Transforming growth factor β (TGF-β) is one of several molecular drivers contributing to the EMT process (Quintanal-Villalonga, Chan et al., 2020). RNA-seq data from CTCs have shown that TGF-β and cyclin D1 (CCND1) signalling pathways are significantly upregulated in drug-resistant CTCs. Moreover, the key regulators of the TGF-β pathway, namely, SMAD family member 3 (SMAD3) and CCND1, participate in the resistance to Enz (Pal, Patel et al., 2018). A new study found that the loss of the transcription factor FOXA1 led to the distinct upregulation of TGF-β3, inducing TGF-β signalling and the EMT. Compared with a single Enz treatment, adding the TGF-β receptor I kinase inhibitor galunisertib (LY2157299) enhanced the efficacy of Enz in inhibiting CRPC xenograft tumour growth and metastasis (Song, Park et al., 2019 , Paller, Pu et al., 2019). The transcription factor STAT3 is also associated with the EMT and ENZ resistance. Metformin, a TGF-β1/STAT3 axis inhibitor, alleviated resistance to Enz by suppressing the EMT (Liu, Tong et al., 2017). Recently, the quinazoline-derived agent DZ-50 was found to reverse EMT to MET by targeting insulin-growth factor binding protein-3 (IGFBP-3), resulting in the re-differentiation of prostate cancer. Phenotype reversal in prostate cancer cells contributes to re-sensitizing PCa cells to Enz to overcome resistance to anti-androgen therapy (Hensley, Cao et al., 2019).
Previous studies have demonstrated that the loss of the retinoblastoma tumour suppressor gene RB1 promotes lineage plasticity and the metastasis of PCa with PTEN mutations. Simultaneously losing RB1 and the tumour suppressor gene TP53 facilitated resistance to anti-androgen therapy. RB1 and TP53 suppress epigenetic reprogramming factors such as EZH2 and SOX2, which play critical roles in generating induced pluripotent stem cells (Ku, Rosario et al., 2017 , Ge, Wang et al., 2020). The enhancer of EZH2 is usually overexpressed in PCa and associated with poor prognosis. EZH2 directly binds to the promoter of prostate-specific antigen (PSA) and suppresses its expression in Enz-resistant PCa cells. EZH2 inhibition/depletion boosts the efficacy of Enz in resistant CRPC (Xiao, Tien et al., 2018). Studies have proven that AR inhibition drives the transformation of prostate cancer into a neuroendocrine phenotype (Carceles-Cordon, Kelly et al., 2020). The understanding of neuroendocrine prostate cancer (NEPC) biology remains poorly understood. Zhang et al. (Zhang, Zheng et al., 2018) demonstrated that neuroendocrine differentiation (NED) and angiogenesis were regulated by ADT-activated CREB, which in turn enhanced EZH2 activity. Notably, anti-angiogenic factor thrombospondin-1(TSP1) was a direct target of EZH2 epigenetic repression. Castration activated the CREB/EZH2 axis, thereby upregulating NE markers and downregulating TSP1. The CREB/EZH2/TSP1 axis was found to be a new pathway for CRPC/NEPC treatment. The long non-coding RNA-p21 (lncRNA-p21), as a TP53 co-repressor, was upregulated upon Enz treatment to induce NED. Enz promoted the transcription of lncRNA-p21 by changing the AR binding affinity for different AREs, converting the function of EZH2 from being a histone-methyltransferase to a non-histone methyltransferase, which led to STAT3 methylation resulting in NED. The Enz/AR/lncRNA-p21/EZH2/STAT3 signalling pathway provides novel targets for developing new therapeutic methods for Enz-resistant CRPC (Luo, Wang et al., 2019). SOX2 is a reprogramming transcription factors and is upregulated after the function of TP53 and RB1 is lost, leading to drug resistance. However, previous studies did not indicate whether elevated SOX2 was sufficient to induce NE marker expression or Enz resistance (Mu, Zhang et al., 2017). By utilizing LNCaP cells engineered for the inducible elevation of SOX2, Metz et al. (MetzWilder et al., 2020) showed that increasing SOX2 did not diminish the growth inhibition of Enz or its ability to reduce the expression of PSA. Thus, the identification of the exact mechanisms of the synergistic effects of SOX2 elevation in NE plasticity and the discovery of therapeutic targets to restore cell sensitivity to Enz are urgently needed. SOX9 is a recently discovered SOX family member that is upregulated in stem-like reprogrammed prostate cancer cells. Transient SOX9 expression facilitated the resistance to Enz via NF-κB dimer activation (Nouri, Massah et al., 2020). Another epigenetic remodelling regulator, the RE1-silencing transcription factor (REST), mainly binds to chromatin in the proximity of neuron-specific genes. REST silencing is a key driver of neuron-specific gene expression in prostate cancer (Flores-Morales, Bergmann et al., 2019). Serine/arginine repetitive matrix 4 (SRRM4) is an upstream regulator of REST gene functions in PCa cells, promoting the progression of NEPC. Both AR inhibition and RB1 and TP53 loss enhanced the SRRM4-induced acquisition of the neuroendocrine phenotype in prostate cancer (Li, Donmez et al., 2017). A new study (Tiwari, Manzar et al., 2020) showed that serine peptidase inhibitor Kazal type 1 (SPINK1) was transcriptionally repressed by AR and its corepressor REST. AR inhibition triggered SPINK1 upregulation, and increased SOX2 expression directly transactivated SPINK1, which facilitated the acquisition of the NE-like phenotype.
In addition, neuroendocrine lineage reprogramming is constitutively associated with MYCN. N-Myc is encoded by MYCN and overexpressed in NEPC. Epigenomic and transcriptomic reprogramming can be induced by N-Myc overexpression and its subsequent DNA binding, contributing to the lineage-plastic phenotype. Additionally, EZH2 inhibitors reversed the N-Myc-induced inhibition of epithelial lineage genes. N-Myc depended on the BET family of epigenetic readers, especially BRD4, which promote the expression of the target genes. N-Myc also overlapped with HOXB13, a lineage-specific homeodomain-containing transcription factor, located predominantly at genomic loci implicated in neural lineage specification (Dardenne, Beltran et al., 2016 , Berger, Brady et al., 2019). Nerlakanti et al. (Nerlakanti, Yao et al., 2018) found that BRD4 epigenetically promotes HOXB13 expression and binds the enhancer of HOXB13. This BRD4-HOXB13 axis activated AR-independent cell cycle programmes, promoting the proliferation of CRPC. The neural transcription factor BRN2 was also a major driver of NEPC and was required for the expression of terminal NE makers and the aggressive growth of Enz-resistant CRPC tumours (Bishop, Thaper et al., 2017). Mucin 1 (MUC1) (Yasumizu, Rajabi et al., 2020) is a heterodimeric protein that suppresses AR signalling and induces BRN2 in association with the induction of MYCN and EZH2, and upregulation of the NE markers related to NEPC progression. BRN4 was found to be a novel transcription factor that drove neuroendocrine differentiation by interacting with BRN2. Enz treatment contributed to the amplified release of BRN2 and BRN4 in PCa extracellular vesicles (EVs), facilitating neuroendocrine differentiation (Bhagirath, Yang et al., 2019). The tumour suppresser protein kinase C (PKC)λ/ι also played a critical role in NEPC progression. The downregulation of PKCλ/ι promoted cancer cell plasticity and NEPC differentiation via the mTORC1/activating transcription factor 4 (ATF4)/PHGDH axis. This metabolic reprogramming augmented intracellular S-adenosyl methionine (SAM) levels to meet the epigenetic changes that facilitated the development of NEPC properties. Notably, the loss of PKCλ/ι led to the activation of mTORC1, which drove an ATF-dependent gene transcription programme that increased flux through the one-carbon pathway (Reina-Campos, Linares et al., 2019).
Tumour heterogeneity has been widely considered to negatively affect targeted cancer therapy, especially when genomic changes occur at the same time, which reduces the reliance on proto-oncogene drivers. In an in vivo small hairpin RNA (shRNA) screening of 730 prostate cancer genes, CHD1 loss was found to be the cause of anti-androgen resistance. ATAC-seq and RNA-seq analyses demonstrated that CHD1 loss led to overall changes in open and closed chromatin and associated transcriptomic changes, conferring resistance to Enz. Notably, four transcription factors (NR3C1, POU3F2, NR2F1, and TBX2) were found by screening to participate in Enz resistance, and they can be future therapeutic targets. To better understand the genetic and epigenetic mechanisms of neuroendocrine transformation, the CTC-derived explant (CDX) model is a distinct tool that can also be used as an effective therapeutic screening approach (Zhang, Zhou et al., 2020 , Faugeroux, Pailler et al., 2020).