Reciprocal deregulation of NKX3.1 and AURKA axis in castration-resistant prostate cancer and NEPC models

NKX3.1, a prostate-specific tumor suppressor, is either genomically lost or its protein levels are severely downregulated, which are invariably associated with poor prognosis in prostate cancer (PCa). Nevertheless, a clear disconnect exists between its mRNA and protein levels, indicating that its post-translational regulation may be critical in maintaining its protein levels. Similarly, AURKA is vastly overexpressed in all stages of prostate cancer (PCa), including castration-resistant PCa (CRPC) and neuroendocrine PCa (NEPC), although its transcripts are only increased in ~ 15% of cases, hinting at additional mechanisms of deregulation. Thus, identifying the upstream regulators that control AURKA and NKX3.1’s levels and/or their downstream effectors offer an alternative route to inhibit AURKA and upregulate NKX3.1 in highly fatal CRPC and NEPC. AURKA and NKX3.1 have not linked to each other in any study to date. A chemical genetic screen revealed NKX3.1 as a direct target of AURKA. AURKA-NKX3.1 cross-talk was analyzed using several biochemical techniques in CRPC and NEPC cells. We uncovered a reciprocal loop between AURKA and NKX3.1 in CRPC and NEPC cells. We observed that AURKA-mediated NKX3.1 downregulation is a major mechanism that drives CRPC pathogenesis and NEPC differentiation. AURKA phosphorylates NKX3.1 at three sites, which degrades it, but AURKA does not regulate NKX3.1 mRNA levels. NKX3.1 degradation drives highly aggressive oncogenic phenotypes in cells. NKX3.1 also degrades AURKA in a feedback loop. NKX3.1-AURKA loop thus upregulates AKT, ARv7 and Androgen Receptor (AR)-signaling in tandem promoting highly malignant phenotypes. Just as importantly, we observed that NKX3.1 overexpression fully abolished synaptophysin and enolase expression in NEPC cells, uncovering a strong negative relationship between NKX3.1 and neuroendocrine phenotypes, which was further confirmed be measuring neurite outgrowth. While WT-NKX3.1 inhibited neuronal differentiation, 3A-NKX3.1 expression obliterated it. NKX3.1 loss could be a major mechanism causing AURKA upregulation in CRPC and NEPC and vice versa. NKX3.1 genomic loss requires gene therapy, nonetheless, targeting AURKA provides a powerful tool to maintain NKX3.1 levels. Conversely, when NKX3.1 upregulation strategy using small molecules comes to fruition, AURKA inhibition should work synergistically due to the reciprocal loop in these highly aggressive incurable diseases.

prostate tumors (> 94%) [1]. In contrast, AURKA transcripts were upregulated in only 15.4% of prostate cancer (PCa) and 76.3% of BPH specimens [2], indicating that post-translational stabilization of AURKA is a critical factor in promoting its deregulation. AURKA levels were also significantly higher in local and metastatic CRPC tumor specimens as compared to hormone-naïve PCa samples [3]. AURKA overexpression is also a hallmark of de novo and treatment-induced neuroendocrine prostate cancer (NEPC) [4]. Many AURKA-selective inhibitors are in clinical trials. Alisertib (aka MLN8237), one of the selective-AURKA inhibitors, is currently being used in many Phase I and II trials against various cancers, including in CRPC [5]. Nevertheless, no AURKA-targeted drug has been approved yet, partly because alisertib has shown efficacy in only ~ 20-25% patients at the best, particularly in solid cancers. Furthermore, as AURKA is an essential kinase, its inhibition may cause substantial collateral toxicity in normal tissues. In contrast, AURKA inhibition in combination with chemotherapy, radiation, HDAC or MYCN inhibitors improves the efficacy in up to 40-50% patients [6][7][8]. Thus, an alternate approach is to identify the upstream regulators and downstream targets of AURKA, which could potentially be used as therapeutic intervention points to target AURKA-induced malignancy either alone or in combination with AURKA inhibitors. Previously, we have identified a few such oncogenic downstream targets and upstream regulators, which both regulate and are regulated by AURKA in a feedback loop [9][10][11][12][13][14]. Thus, specific inhibition of these substrates provides an effective alternate approach to indirectly modulate AURKA with potentially much less toxicity.
The present study focuses on one such feedback loop, which was discovered between AURKA and a tumorsuppressor NKX3.1 in CRPC and NEPC cells. We have identified NKX3.1 as a direct substrate of AURKA by employing a pioneering global screen [9,[15][16][17][18]. Unlike AURKA, which is ubiquitously expressed, NKX3.1 is a prostate-specific transcription factor, which is essential for the development and maintenance of prostate and testes [19]. NKX3.1 is also a tumor suppressor gene, situated on chromosome 8p21.2, which shows loss of heterozygosity (LOH) in up to 89% of high-grade prostatic intraepithelial neoplasia (HG-PIN) and up to 86% of prostatic tumors [20]. NKX3.1 is fully lost in up to 78% of metastatic lesions and 34% of CRPC [21]. Targeted disruption of NKX3.1 in mice causes prostatic epithelial hyperplasia and PIN [22]. Moreover, when combined with Pten disruption, loss of one or both NKX3.1 alleles causes more aggressive and rapid PCa [23]. Importantly, loss-of-function of Nkx3.1 and Pten facilitates androgen independence following castration [24]. Subsequently, the authors reported that Nkx3.1-Pten mice acquire androgen-independence even before the manifestation of PIN or PCa [25]. These findings indicate that loss of NKX3.1 is intimately linked with CRPC progression.
While the tumor-suppressive functions and loss of NKX3.1 are well established, a contradiction exists between its mRNA and protein levels in PCa. Most of the studies reported that mRNA levels of NKX3.1 are either increased or unchanged in PCa tissues compared to normal tissues [26]. In contrast, NKX3.1 protein was uniformly downregulated in IHC studies [22,27]. These findings suggest that NKX3.1 downregulation at the posttranslational stage may contribute significantly to PCa pathogenesis, which prompted us to examine whether the reciprocal levels of AURKA and NKX3.1 in PCa are related to each other.

Cell lines and antibodies
C4-2, 22Rv1, HEK-293T and Phoenix cells were purchased from American Type Culture Collection (ATCC; Manassas, VA, USA) and maintained according to the manufacturer's instructions. 49F cells were obtained from Dr. Amina Zoubeidi and maintained in RPMI medium, 10% FBS along with 10 μM enzalutamide (MedChemExpress, NJ, USA). The details of the antibodies used in this study are included in Additional file 1: Table S1.

In vitro phosphorylation assays
6x-His-tagged AURKA was purified using Ni-NTA beads. 6x-His-TPX2 was expressed and isolated from Escherichia coli. To remove the background signal, the AURKA-TPX2 complex was treated with kinase buffer (50 mM Tris, 10 mM MgCl 2 ) containing ATP (100 μM) for 1 h at 30 °C. Subsequently, the beads were washed three times with kinase buffer, followed by the addition of ~ 2 μg of 6x-His-tagged WT type or mutant NKX3.1 and 2 μCi of [γ-32 P] ATP for 25 min. The reaction mixture was boiled in SDS-PAGE dye for 5 min, proteins were separated using SDS-PAGE and exposed to X-Ray film.

Immunofluorescence
Immunofluorescence was performed as before [31]. PCa cells were plated on poly-lysine-coated coverslips for 16 h. The cells were treated with respective lentivirus (30 h) or Alisertib (1 μM for 12 h). The cells were fixed, permeabilized and blocked with PBS/0.1% triton X-100/2% BSA solution. The coverslips were incubated with substrate-specific antibodies overnight at 4 °C, followed by dye-conjugated secondary antibody for 2 h in dark. Cells were counterstained with DAPI (Sigma, MO, USA) for 5-10 min (dilution of 1:50,000). Images were taken using a Nikon Eclipse E600 microscope (Nikon Instruments, Melville, NY).

Real-time qPCR
Real-time qPCR experiments were conducted as reported before [32]. The primers are listed in Additional file 1: Table S2. Each experiment was carried out in triplicate at least three independent times.

Cycloheximide assay
The cells were seeded in 6-well plates for 12 h prior to infection. Subsequently, corresponding retro-or lentiviruses were added for an additional 32 h. Cycloheximide (20 μg/ml) was then added for the times indicated in the figures prior to lysis. The cell lysates were subjected to Western blot analysis.

Ubiquitylation assay
Ubiquitylation assay was performed as described before [14]. Briefly, C4-2 and 22Rv1 cells were infected with 6x-His-Ubiquitin retrovirus along with either AURKA or NKX3.1 (WT or mutant) retrovirus for 30 h, followed by MG132 (10 μM) addition for 12 h. The corresponding lysates were incubated with either Ni-NTA beads or specific antibodies for 4 h. The proteins were separated by SDS-PAGE, transferred on PVDF membrane and the ubiquitylated proteins were detected using either 6x-His or substrate-specific antibody.

Isolation of cytosolic and nuclear fractions
C4-2 and 22Rv1 cells were washed twice with chilled PBS, resuspended in buffer A (10 mM Tris pH 7.9, 10 mM KCl, 0.5 mM DTT, 0.05% NP40, 1.5 mM MgCl 2 , and 1 mM PMSF) and placed on ice for 10 min, followed by centrifugation at 3000 rpm at 4 °C (10 min). To separate the nuclear fraction, the pellet was resuspended in buffer B (300 mM NaCl, 5 mM Tris pH 7.9, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 26% glycerol (v/v) and 1 mM PMSF). The suspension was homogenized using a 27½ gauge needle (ten times). The lysates were placed on ice for 30 min and the nuclear fraction was separated by centrifugation at 24,000×g at 4 °C for 20 min. The cytosolic and nuclear extracts were further analyzed by Western blotting [33].

Chemotaxis assay
Migration assay was performed using Boyden chambers as reported previously [18].

MTT assay
The MTT assay was conducted as before [34].

Neurite outgrowth assay
49F NEPC cells were seeded in a 6-well plate at a density of 5 × 10 4 cells/well. After 12 h, the cells were infected with the respective retroviruses to initiate ectopic overexpression of the wild-type and phospho-resistant NKX3.1. 36 h post infection, the cells were washed with PBS and imaged under AmScope light microscope. The definition of a neurite points to "an extension from the cell body equivalent or greater than 1× the cell body width" [35]. Bright field images were imported in ImageJ software and neurite length was calculated as fraction of cell body width. This length was normalized against the vector-treated cells. Ten different fields of cells were used for quantification from five different replicates.

Statistical analysis
All data are displayed as mean ± SEM of three or more experiments. Statistical analysis was performed using GraphPad Prism (version 6.07). Statistical significance of difference was determined by the one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc test. P < 0.05 was considered statistically significant.

AURKA directly phosphorylates NKX3.1 in vitro
As NKX3.1 was identified as an AURKA target in a global screen, we inspected whether AURKA directly phosphorylates NKX3.1 in vitro. AURKA in complex with its activator TPX2 was incubated with recombinant 6x-His-NKX3.1, which resulted in the phosphorylation of the latter, indicating that NKX3.1 is a substrate of AURKA (Fig. 1A, lane 3).

NKX3.1's nuclear residence is independent of AURKA
We examined whether AURKA regulates the subcellular location of NKX3.1. AURKA was knocked-down in C4-2 cells, which did not impact the nuclear residence of NKX3.1. Similarly, AURKA inhibition using Alisertib had no effect on NKX3.1 localization (Fig. 1D, E). Equivalent results were observed in 22Rv1 cells, where AURKA silencing or inhibition showed no change in NKX3.1 localization (Fig. 1F, G). To confirm these results, we performed subcellular fractionation in scrambled shRNA and AURKA-shRNA treated C4-2 and 22Rv1 cells. Neither of the cell-type showed any change in NKX3.1 localization upon AURKA silencing (Fig. 1H, I), thereby confirming that NKX3.1 nuclear residence is not controlled by AURKA in cells.

AURKA subcellular residence is independent of NKX3.1
Unlike NKX3.1, AURKA was present both in the cytoplasm and nucleus in C4-2 and 22Rv1 cells, although it was predominantly cytoplasmic (Fig. 1J, K). NKX3.1 knockdown had no impact on AURKA localization, indicating that it does not regulate AURKA localization in cells. Subcellular fractionation further confirmed these results in both C4-2 and 22Rv1 cells (Fig. 1L, M).
We next examined whether AURKA-mediated regulation of NKX3.1 occurs at the mRNA stage. AURKA overexpression resulted in over threefold increase in its transcripts, however, NKX3.1 mRNA levels remained the same in C4-2 cells (Fig. 2K). AURKA silencing also failed to change NKX3.1 mRNA levels in C4-2 cells (Fig. 2L). 22Rv1 cells showed analogous results upon AURKA overexpression and knock-down, respectively ( Fig. 2M, N). and AURKA bind each other in C4-2 cells. AURKA was immunoprecipitated and its association with NKX3.1 analyzed. IgG was used as the negative control, and NKX3.1 IP was used as a positive control. C NKX3.1 and AURKA bind each other in C4-2 cells. NKX3.1 was immunoprecipitated and its binding with AURKA was analyzed. IgG was used as the negative control, and AURKA IP was used as a positive control. D AURKA knockdown does not impact the subcellular localization of NKX3.1 in C4-2 cells. Immunofluorescence micrographs of C4-2 cells infected with either scrambled or AURKA shRNA followed by probing with NKX3.1 antibody (red). Nuclear counterstain is represented by DAPI (blue). (Scale bar = 20 μm). AURKA knockdown was confirmed using Western blot analysis. Images for control cells (having much lower NKX3.1 expression levels) were shown in enhanced gain to assist in visualization of the red signal. E AURKA inhibition does not alter the subcellular localization of NKX3.1 in C4-2 cells. Immunofluorescence images representing the subcellular distribution of NKX3.1 (red) in response to Alisertib in C4-2 cells. The blue channel represents DAPI for the nuclear counterstain. (Scale bar = 20 μm). Images for DMSO treated cells (having much lower relative NKX3.1 expression levels) were shown in enhanced gain to assist in visualization of the red signal. F AURKA depletion does not affect the subcellular localization of NKX3.1 in 22Rv1 cells. Immunofluorescence analysis of 22Rv1 cells with and without AURKA knockdown. Texas Red was used for probing NKX3.1 and DAPI (blue) is used for nuclear counterstain. (Scale bar = 20 μm). Western blot for confirmation of AURKA knockdown in 22Rv1 cells. G Inhibition of AURKA activity has no effect on the subcellular localization of NKX3.1 in 22Rv1 cells. Images obtained from immunofluorescence microscopy with red-NKX3.
To further confirm that NKX3.1 overexpression causes AURKA degradation, we overexpressed S185A-NKX3.1 mutant in C4-2 cells, which is more stable than WT, and thus is expressed at relatively higher levels in cells [28]. While WT NKX3.1 expression decreased AURKA levels, S185A expression decreased it even more significantly, confirming that NKX3.1 downregulates AURKA levels in a dose-dependent manner (Fig. 3Q, R).

Phospho-resistant NKX3.1 shows significantly enhanced stability
We next tested the consequences of AURKA-triggered phosphorylation of NKX3.1. Both WT and 3A-NKX3.1 were ectopically expressed in C4-2 cells. As expected, phospho-resistant 3A-NKX3.1 showed higher levels as compared to WT (Fig. 5A, B). Conversely, AURKA levels showed the opposite pattern with the highest levels in control C4-2 cells, followed by WT and least in 3A-NKX3.1 cells, confirming the negative regulation by NKX3.1. We observed a similar pattern in 22Rv1 cells (Fig. 5C, D).
The stability of WT and 3A mutant was compared using cycloheximide. As shown, 3A-NKX3.1 showed a substantially longer half-life as compared to the WT allele (Fig. 5E). Figure 5F shows WT and 3A-NKX3.1 degradation patterns from three independent experiments. Comparable regulation was observed in 22Rv1 cells (Fig. 5G, H).
AURKA was further overexpressed in control C4-2, WT and 3A-NKX3.1-C4-2 cells to investigate whether it reverses the negative impact of NKX3.1 in cell motility. AURKA overexpression increased chemotaxis in C4-2 cells as predicted. AURKA could rescue cell motility in NKX3.1-C4-2 cells as well, but not in 3A-NKX3.1 cells, indicating that phospho-resistant NKX3.1 is fully capable of counteracting the oncogenicity of AURKA (Fig. 6J, K). Together, these results implicate that the balance between AURKA and NKX3.1 levels is crucial in dictating the aggressiveness of PCa tumors.

AURKA upregulates AKT, AR and ARv7 signaling via NKX3.1 phosphorylation
Activation of the PI3K-AKT pathway plays a critical role in the initiation and progression of CRPC. As NKX3.1 inhibits the AKT pathway, we investigated whether AURKA activates the AKT pathway via NKX3.1 in CRPC. WT and 3A-NKX3.1 were overexpressed in C4-2 cells, which fully inhibited AKT activation, although there was no change in AKT levels (Fig. 6L, M). We observed similar AURKA-mediated regulation of AKT signaling in 22Rv1 cells (Fig. 6N, O). As both WT and 3A-NKX3.1 fully inhibited phospho-AKT signaling in C4-2 and 22Rv1 cells, we tested the potential impact of these two alleles in AURKA-overexpressing-C42 and 22Rv1 cells. AURKA overexpression strongly increased phospho-AKT signal at S473 and T308 sites in both C4-2 and 22Rv1 cells (Fig. 5P-S), which was robustly decreased by ectopic expression of WT-NKX3.1. 3A-NKX3.1 expression was relatively more effective than WT in diminishing phospho-AKT levels in these cells (Fig. 5P-S).
We next examined whether AURKA also upregulates the AR pathway by degrading NKX3.1. Ectopic expression of WT and mutant NKX3.1 completely abolished AR levels in C4-2 cells (Fig. 6T, U). In 22Rv1 cells, both AR and Arv7 levels were severely reduced upon WT and phospho-resistant NKX3.1 overexpression (Fig. 6V, W). Together, these results revealed a direct link of AURKA in activating AKT and AR pathways in CRPC via degradation of NKX3.1.

AURKA-NKX3.1 cross-talk in NEPC cells
AURKA amplification is one of the salient features of NEPC, and is causally linked to neuroendocrine differentiation in both de novo and ADT-resistant CRPC tumors [4]. In contrast, NKX3.1 mRNA and protein levels have only been analyzed in a few NEPC tumors, most of which show downregulation of NKX3.1 [37,38]. These findings prompted us to investigate whether AURKA upregulation could be linked to NKX3.1 levels in NEPC. We also wondered whether NKX3.1 could regulate neuroendocrine phenotypes, which has not been shown in any study to date.

Discussion
NKX3.1 is predominantly expressed in prostate luminal epithelial cells, and promotes cellular differentiation and lineage plasticity [22]. The loss of NKX3.1 is a crucial event in PCa initiation. NKX3.1 is a haploinsufficient gene, and loss of heterozygosity (LOH) at the Nkx3.1 locus results in hyperplasia and eventually PIN formation. At the molecular level, loss of a single Nkx3.1 allele prolongs the proliferative stage of dividing luminal cells, causing hyperplasia [40]. Most importantly, the dosage of Nkx3.1 controls discrete subsets of genes, thus, loss of one Nkx3.1 allele results in complete loss of some target genes, while other genes require loss of both copies [40].
While the loss of NKX3.1 protein is a hallmark of PCa in clinical specimens and mouse models, the accompanying NKX3.1 mRNA levels show little correlation with its protein levels [21]. Additionally, loss of NKX3.1 protein shows little correlation with loss of its locus or with the failure to identify inactivating mutations [41]. All these findings indicate that post-translational regulation of NKX3.1 plays a critical role in diseased states. NKX3.1 protein stability is indeed shown to be differentially regulated by phosphorylation. DYRK1B directly phosphorylates NKX3.1 at S185, which causes its degradation [42]. In contrast, it has been postulated that PIM1 phosphorylates NKX3.1 at both S185 and S186, which increases its stability [43]. CK2 phosphorylates at T89 and T93, which stabilizes it [44]. Markowski et al. showed that during inflammation, tumor necrosis factor (TNF)-alpha and interleukin-1 beta (IL-β) causes NKX3.1 phosphorylation at S196, which promote NKX3.1 degradation, although the kinase was not identified [45]. During DNA damage, active ATM phosphorylates NKX3.1 at residues T134 and T166, accelerating NKX3.1 degradation [46]. PKC (Protein kinase C) was shown to phosphorylate NKX3.1 at S48, although whether it regulates its protein stability (See figure on next page.) Fig. 6 NKX3.1 and AURKA cross-talk regulates aggressive phenotypes including AR, ARv7 upregulation and AKT activation in CRPC cells. A Phospho-resistant NKX3.1 inhibits cell proliferation more effectively in C4-2 cells as compared to WT NKX3.1. Cell proliferation was measured at indicated times. *P < 0.05, **P < 0.01. B Phospho-resistant NKX3.1 inhibits cell proliferation more effectively in 22Rv1 cells as compared to WT NKX3.1. *P < 0.05, **P < 0.01. C Ectopic expression of AURKA increases cell proliferation in C4-2 and NKX3.1-C4-2 cells, but not in 3A-NKX3.1-C4-2 cells. AURKA retrovirus was transiently infected in C4-2, NKX3.1 and 3A-NKX3.1 cells and cell growth was measured after 36h using MTT assay. **P < 0.01, and ***P < 0.001. D Colony formation assay showed that 3A-NKX3.1 is more effective in inhibiting colony formation as compared to the WT allele. E Quantitative data analysis of the soft agar experiment from three independent experiments. *P < 0.05. F NKX3.1 and 3A-NKX3.1 fully suppress chemotaxis in C4-2 cells, whereas AURKA knockdown partially suppressed it. The cells were starved in serum-free media for 12 h. Chemotaxis was performed using Boyden chambers. G The plot shows mean ± SEM of cell motility in C4-2, AURKA-knocked down-C4-2, NKX3.1 and 3A-NKX3.1-C4-2 cells from three independent experiments. **P < 0.01. H NKX3.1 and 3A-NKX3.1 fully suppress chemotaxis in 22Rv1 cells, whereas AURKA knockdown partially suppressed it. I Bar graph indicating the extent of migration plotted as mean ± SD of three independent experiments such as the one indicated in H. *P < 0.05. J AURKA overexpression rescues chemotaxis more effectively in C4-2 and NKX3.1-C4-2 cells, as compared to 3A-NKX3.1-C4-2 cells. K Histogram representing the quantification of migration levels, plotted as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01. L Levels of phospho-AKT in NKX3.1 and 3A-NKX3.1 overexpressing C4-2 cells are significantly lower than control cells. Control, NKX3.1-C4-2 and 3A-NKX3.1-C4-2 cells were assayed for p-AKT levels along with AKT and actin. M Quantification of change in AKT phosphorylation levels in response to NKX3.1 and 3A-NKX3.1-expression. Data from three independent experiments was normalized against actin, and represented as mean ± SEM [**P < 0.01, ns not significant]. N Degree of AKT phosphorylation is lowered by ectopic overexpression of wild-type and 3A-NKX3.1 in 22Rv1 cells. O Quantification of AKT phosphorylation levels obtained from three independent experiments such as the one depicted in N. [**P < 0.01, ns not significant]. P WT and 3A-NKX3.1 retroviruses were infected in AURKA overexpressing C4-2 cells and p-AKT levels were analyzed along with AKT and tubulin. Q Data from three independent experiments as in 6P were used for quantification, *P < 0.05, **P < 0.01 relative to control. R AURKA overexpressing 22Rv1 cells were also assessed for p-AKT levels in response to WT and 3A-NKX3.1 overexpression. S Three independent experiments as in 6R were used for quantitative analysis, **P < 0.01. T Both wild-type NKX3.1 and 3A-NKX3.1 deplete AR protein levels in C4-2 cells. U Histogram showing change in AR and NKX3.1 protein levels. Normalized data from three independent experiments, with actin as loading control, was plotted, **P < 0.01 compared to control cells. V Ectopic expression of NKX3.1 and 3A-NKX3.1 depletes AR protein levels in 22Rv1 cells. W Histogram depicting changes in AR protein levels in 22Rv1, NKX3.1-22Rv1 and 3A-NKX3.1-22Rv1 cells. The data from three independent experiments was plotted as mean ± SEM, **P < 0.01 vs 22Rv1 control cells is unknown [47]. In contrast to NKX3.1 regulation by several kinases, there is only one report that showed that NKX3.1 also regulates its kinase in a feedback mechanism [28]. We showed that LIMK2 kinase directly phosphorylates NKX3.1 at S185 and degrades it. NKX3.1 in return degrades LIMK2 as well by increasing its ubiquitylation [28]. The present study exposed that AURKA phosphorylates NKX3.1 at S28, S101 and S209, which triggers its ubiquitylation. Thus, regulating NKX3.1's protein stability is critical both under normal and diseased conditions. Furthermore, as 3A-NKX3.1 is more resistant to AURKA-mediated ubiquitylation, we believe that unlike N-Myc and FOXM1, AURKA uses its kinase activity to degrade NKX3.1. AURKA is highly expressed in C4-2 and 22Rv1 cells regardless of cell cycle, indicating that AURKA-mediated degradation of NKX3.1 is not cell cycle-dependent, although it may increase during mitosis, when AURKA levels are relatively higher. Our results further show that neither the depletion nor inhibition of AURKA has any impact on nuclear localization of NKX3.1. Nevertheless, as AURKA directly regulates NKX3.1 levels via phosphorylation, AURKA is expected to have significant control the transcriptional output of NKX3.1. Future studies are required to fully address AURKA-mediated transcriptional regulation of NKX3.1. Like NKX3.1, AURKA mRNA levels also show little correlation with protein levels in PCa, underscoring that both their regulation at the protein level is critical. NKX3.1 binds AURKA and triggers its ubiquitylation. 3A-NKX3.1 is resistant to AURKA-mediated degradation, hence, it effectively degrades AURKA and reverses oncogenic phenotypes. Thus, genomic loss of NKX3.1 could be a dominant factor contributing to AURKA protein upregulation in a significant percentage of PCa. While the exact mechanism by which NKX3.1 promotes the ubiquitylation of AURKA remains unclear, it could be at least partly mediated by AKT/GSK3b/FBXW7 pathway [48]. FBXW7 is a F-box protein, which is a part of the substrate recognition component of the SCF E3 ubiquitin ligase. FBXW7 regulates proteasome-mediated degradation of many oncoproteins including AURKA.
GSK3β inhibition reduces the binding affinity between AURKA and FBXW7, leading to AURKA stabilization. As NKX3.1 inhibits AKT signaling and AKT inactivates GSK3b by phosphorylation at Ser9, we speculate that NKX3.1 activates GSK3b by inhibiting AKT signaling, leading to AURKA degradation via FBXW7. Future studies are required to validate the exact mechanisms of AURKA degradation by NKX3.1.
Recent studies have uncovered a few mechanisms by which AURKA upregulates AR and ARv7 signaling in CRPC. AURKA stabilizes YBX1 causing upregulation of AR protein and ARv7 mRNA levels [7]. AURKA also degrades SPOP stabilizing both AR and ARv7 proteins [14]. We show that AURKA-mediated NKX3.1 degradation is another mechanism by which AURKA increases both AR and ARv7 levels (Fig. 7K).
Several studies have shown that AURKA inhibition downregulates AKT activation, however, the molecular players mediating this response largely remain unknown. Previously, we identified that AURKA directly phosphorylates and degrades PHLDA1 in breast cancer cells [5]. PHLDA1 is a repressor of AKT signaling, which could lead to AKT activation by AURKA. However, future studies are needed to establish whether PHLDA1 is regulated by AURKA in PCa. This study uncovered that NKX3.1 degradation is a key mechanism by which AURKA augments AKT signaling in CRPC cells (Fig. 7K).
AURKA overexpression or amplification is a hallmark of NEPC. In contrast, little is known about the role of NKX3.1 in NEPC. Nkx3.1 was genomically lost in p53-and Rb-deficient mouse prostate tumors exhibiting neuroendocrine phenotypes [37]. Similarly, human prostate tumors immunoreactive for neuroendocrine markers lacked or minimally showed NKX3.1 immunoreactivity [38]. Although a recent study showed that majority of AR-positive neuroendocrine tumors also express NKX3.1, which is consistent with its origin as an AR-regulated gene [49]. The mechanism by which NKX3.1 could be downregulated at the protein level in NEPC has not been investigated. Our study showed that AURKA overexpression is a major mechanism by which