The clock gene BHLHE40 and atypical CCNG2 control androgen-induced cellular senescence as a novel tumor suppressive pathway in prostate cancer

The LNCaP cell line, representing a castration-sensitive prostate cancer model, was acquired from Protopopov et al. in 2004 [23]. The C4-2 cell line, exhibiting castration-resistant characteristics in prostate cancer, was obtained from Thalmann et al. in 2007 [24]. As previously described, LNCaP and C4-2 cells were cultured in RPMI and DMEM media, respectively, at 37 °C in a humidified incubator with 5% CO2 [5]. 1nM R1881 (Merck; Germany; R0908), was defined previously serving as SAL [5], or 0.1% DMSO (ROTH; Germany; 4720.1) as a solvent control was used for cell treatments.

To modulate BHLHE40 expression levels in order to investigate the role of BHLHE40 in cellular senescence induction by SAL, C4-2 and LNCaP cell lines were subjected to knockdown by shRNA, siRNA, or overexpression. For shBHLHE40 KD plasmid construction, the backbone EZ-Tet-pLKO-Puro plasmid was obtained from Addgene (#85,966). shRNA targeting BHLHE40 with the following sequence (5'GCCCTGCAGAGTGGTTTACAA3') was inserted into the backbone. Scrambled plasmid was obtained from Addgene (#47,541) as a non-targeting control. For overexpression of BHLHE40, the cDNA sequence of BHLHE40 was purchased from Eurofins Genomics. pCDH-CMV-puro, was used to insert the BHLHE40 cDNA. The JetPrime reagent from PolyPlus Co (PolyPlus; France; 101,000,015), was utilized as per the manufacturer's instructions. Transfection experiments were conducted 48 h (h) before treating cells with SAL or DMSO. si-mediated knockdown was performed in both cell lines by using ON-TARGETplus Human BHLHE40 siRNA with the 'AAAGAGACGUGACCGGAUU' sequence (Dharmacon; USA; L-010318–00-0010) with a final concentration of 25 nM. As a negative control, ON-TARGETplus nontargeting control siRNA with the 'UGGUUUACAUGUUGUGUGA' sequence (Dharmacon; USA; D-001810–04-20) was used. Moreover, for CCNG2 KD ON-TARGETplus Human CCNG2 siRNA with the 'CAUGAUGUGAUCCGGAUUA' sequence was used (Dharmacon; USA; L-003217–00-0010). siRNAs were transfected by the DharmaFECT reagent (Dharmacon; USA; T-2003–02) according to the manufacturer's protocol. Transfection was performed 24 h prior SAL or DMSO treatments.

For SA β-Gal activity, 50,000 cells for both C4-2 and LNCaP cell lines were seeded per well in a 6-well plate. The staining was performed as described previously [5]. Briefly, cells were washed with 1 × PBS prior fixation by 1% glutardialdehyde for 5 min 72 h post treatments. For the next step, cells were washed with 1 × PBS and incubated at 37 °C in SA β-gal staining solution. Staining solution contains 40 mM citric acid/sodium phosphate buffer (pH 6.0), 1 mg/ml X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside), 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 150 mM sodium chloride, 2 mM magnesium chloride.

Crystal violet was used to stain cells according to previously described protocol [25]. In short, 3 days after treatments, C4-2 and LNCaP cells were fixed with 1% glutaraldehyde and stained with 0.1% crystal violet to measure the cell population. Stained cells were washed with Sörenson's solution which contains 0.9%, w/v tri-sodium citrate, 2% HCl, 40% ethanol and water. Absorbance was measured at 590 nm.

RNA was isolated from the cells using RNA-solv reagent (omega; USA; R6830) according to the manufacture's protocol. In brief, cell suspension was collected in a 1.5 ml Eppendorf tube. After adding RNA-solv and chloroform and through mixing, centrifugation was performed for phase separation. RNA was precipitated by isopropanol and centrifugation. RNA concentration was measured by Nanodrop ND-1000 Spectrophotometer. 2 µg RNA was converted to cDNA using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; Lithuania; 4,368,814), the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad; USA; 1,725,271), gene specific primers, and Bio-Rad CFX Duet Real Time PCR machine. All primers used are listed in Supplemental Table S1. Plots represent the Mean of the expression levels and error bars represent SEM.

Cells were lysed in 80 μl lysis buffer (20 mM Tris–HCl pH 8.0, 100 mM NaCl, 1 mM EDTA, 1% NP-40 and 1% Tergitol, 50 mM NaF, 100 μM Na₃VO₄, 10 mM β-Glycerophosphate) and centrifuged at 12 000 × g/4 °C for 10 min to obtain the cell extracts. Quantification of the protein concentration was performed with Nanodrop ND-1000 Spectrophotometer. Subsequently, 30 µg protein extract was loaded for Western blot and anti-beta Actin antibody served as loading control. Cell lysates were separated by 12% SDS-PAGE then proteins were transferred to PVDF membranes. Skim milk was used to block membranes then membranes were incubated with primary antibodies. The detection was performed by ImageQuantTM LAS 4000 (GE Healthcare Bio-Sciences AB). Quantification of bands was performed with the LabImage D1 software. All used antibodies are listed in Supplemental Table S2.

3D spheroids of C4-2 cells with and without BHLHE40 knockdown were generated according to previous protocol [10, 26]. In short, 1000 cells per well were seeded in 96-well ultralow attachment plates (PerkinElmer). The cells were centrifuged 3 times at 300 rpm for 3 min at room temperature (RT), plate was incubated at 37 °C, 5% CO₂ to form a spheroid. After 24 h of seeding, spheroids were treated for 6 days with SAL or DMSO. Cell culture medium was refreshed every 3 days. On day 6, spheroids were fixed with 4% PFA for 20 min at RT following fixation, 3 times washing with 1 × PBS (pH 6) was performed. spheroids were incubated with SA-β-Gal staining solution overnight at 37 °C. Next day spheroids were washed with 1 × PBS, paraffin-embedded and dehydrated in Xylol for 10 min, 100% EtOH, 96% EtOH, 70% EtOH each for 5 min. All steps were repeated twice. Afterwards heat-induced antigen retrieval in a steamer was performed for 20 min in a 1 × citrate buffer. The slides were then washed with 1 × PBS and permeabilization was performed with 0.2% Triton X100 for 10 min. Slices were used for the imaging by brightfield microscope CellObserverZ1 (Carl Zeiss). For Ki-67 staining, spheroid slices were permeabilized and incubated for 1 h at RT in 5% normal goat serum (NGS) (Biozol; Germany; ENG9010-10) for blocking and incubated with primary anti-Ki67 antibody overnight. The next day, spheroids were washed 3 times with 1 × PBS each for 10 min then, secondary anti-rabbit IgG Alexa 546 antibody was used for 1 h at RT in the dark. Washing steps were performed, and DAPI (Life Technologies; USA; H3569) solution (1 μg/ml in 1 × PBS) was used for nuclei staining for 10 min. After one time washing with 1 × PBS for 10 min, the spheroid slices were covered with Fluoromount-G® (SouthernBiotech; USA; 0100–01) and coverslips. Images were captured with the confocal laser scanning microscope (Carl Zeiss LSM 880). For immunofluorescence of BHLHE40, 15,000 LNCaP and C4-2 cells were seeded in Chambered Borosilicate Cover glass from Lab-Tek. Next day cells were treated with SAL or DMSO for 3 days. After seeding for 3 days. After 3 times washing cells with 1 × PBS and permeabilization with 0.2% Triton X100 for 10 min, cells were washed again with 1 × PBS and same as spheroid slices, cells were incubated 1 h at RT with NGS. All following steps were according to steps mentioned above. Antibodies are listed in Supplemental Table S2.

The ex vivo treatment of PCa samples from patients with radical prostatectomies was described previously with ethical approval (Reg.-Nr.: 2019–1502) [11, 27]. All the patients gave informed consent, and all were informed about the purpose of the study and conform to the Declaration of Helsinki. One prostatectomy sample was used from each patient. samples were treated for 48 h by DMSO or R1881. RNA was extracted same as the protocol mentioned in the RNA isolation section. For RT-qPCR, two technical replicates for each sample were performed. Expression levels were normalized to housekeeping genes, alpha-Tubulin and TBP. Sequence of the primers are mentioned in Supplemental Table S1. Gleason scores of prostatectomy samples are listed in Table S3.

Animal experiments were approved by the Thüringer Landesamt für Lebensmittelsicherheit und Verbraucherschutz, Germany (Reg.-Nr.: UKJ-23–013). C4-2 cell suspension (10⁶ cells per 50 μl 1 × PBS) was mixed 1:1 with Matrigel (CORNING; USA; 356,231) and injected subcutaneously (s.c.) into both flanks of the intact (non-castrated) nude mice (8-week-old male athymic nude mice, Janvier Labs, France). After the tumors reached a size of approximately 80 mm³, vehicle (0.5% Tween 80) or dihydrotestosterone (DHT, 50 mg/kg) corresponding SAL (AbMol BioScience; USA; M6033) was s.c injected daily. Tumor size was measured every 48 h using a caliper (tumor volume = (length × width²) × 0.52). Two-way ANOVA was used for statistical analysis. Mice were weighted every other day and sacrificed when tumor size reached a size of approximately 800 mm³ or if weight loss exceeded 20% of initial weight or injection reached to 5 weeks. Tumors were frozen in liquid nitrogen and RNA was extracted from tumors same as previously mentioned protocol [11].

For transcriptome analyses total RNA was isolated from C4-2 cells with and without BHLHE40 KD cells treated with and without SAL treatment, for 72 h in three independent biological replicates, using the mentioned protocol in section mRNA isolation and qRT-PCR. Samples were sent for paired-end sequencing to macrogen Europe. Samples quality check was performed using FastQC version 0.11.5. Adapters were trimmed from both ends by using Skewer version 0.2.1 [28] in Linux (Ubuntu). hg38 (BSgenome.Hsapiens.UCSC.hg38) as a reference genome was used for alignment. Alignment was performed by using the QuasR package and using Rhisat2 aligner algorithm. Read counting was carried out by using GenomicAlignments package. Normalization was performed by using DESeq2 package. Two replicates with confirmed regulation of known BHLHE40 target genes by knockdown were used for further analysis. GSEA (Gene set enrichment analysis) was performed using GSEA (RRID:SCR_003199) [29]. Pathway analysis was performed by using PathFindR and ClusterProfiler packages [30 –32]. The motif analysis for ChIP-seq data was performed via HOMER (RRID:SCR_010881).

AR ChIP-Seq was analyzed for finding AR binding sites under DHT treatment corresponding to SAL and overlapping genes between AR and RNA-Seq data from BHLHE40 Knockdown cells [33]. BHLHE40 ChIP-Seq was used for analysis to identify genome-wide BHLHE40 chromatin binding sites [34]. Two RNA-Seq datasets were used to analyze the expression level of BHLHE40 in prostatectomy samples compared to adjacent tumors [35, 36].

The promoter sequence of target gene was downloaded from UCSC Genome Browser (RRID:SCR_005780) [37] and the JASPAR (RRID:SCR_003030) [38] was used to predict binding sites of BHLHE40 in the promoter of target gene. The cytoHubba (RRID:SCR_017677) from Cytoscape (RRID:SCR_003032), and the STRING (RRID:SCR_005223) [39] were utilized to explore the established network connections between BHLHE40 and CCNG2.

Co-immunoprecipitation (Co-IP) experiments were performed as described previously [40]. Briefly, a specific antibody against BHLHE40 or normal rabbit IgG as a negative control was incubated with protein A magnetic beads. After extensive washing of beads, cell extracts from C4-2 and LNCaP cells were incubated with the antibody-loaded beads for 2 h at 4 °C. After washing steps, beads were resuspended in SDS buffer and boiled at 99 °C. SDS buffer containing precipitated protein complex was loaded to 12% SDS-PAGE for detection by Western blotting. Antibodies used with concentrations are listed in Supplemental Table S2.

Protein structure prediction for BHLHE40 and AR ligand binding domain with hinge region were performed by I-TASSER (RRID:SCR_014627) [41]. The C-score, a measure of confidence in the model, ranged from [-5,2]. A higher score indicates a higher confidence in the model. Also, the TM score > 0.5 indicates correct topology, while TM < 0.17 represents random similarity in the predicted model [42 –44]. Also, protein structure of AR-LBD domain with SAL was downloaded from RCSB Protein Data Bank (PDB ID 1E3G) as a positive control. Docking was performed via PatchDock (RRID:SCR_017589) [45] and ClusPro (RRID:SCR_018248) [46]. Visualization was performed in PyMOL version 1.8.0.0 (RRID:SCR_000305).

The two-tailed student t-test was used for the comparison of the mean values between two groups and two-way ANOVA was used for multiple comparisons in the GraphPad Prism version 8.0 (RRID:SCR_000306).

All statistics represent the Mean of values and error bars represent SEM. For multiple comparisons, the post-hoc (Tukey) test was used.

The AKT pathway has been identified as a critical pathway in controlling pro-survival signaling SAL-induced cellular senescence in PCa and it was observed that SAL enhances phosphorylation of AKT at serine-473 [11]. Interestingly however, the KD of BHLHE40 rather further enhances p-AKT levels and p-p70S6K as a downstream factor of AKT although cellular senescence levels are decreased by the KD (Fig. 1O, P) suggesting that BHLHE40 rather inhibits the AKT signaling, which may be one tumor suppressive pathway mediated by BHLHE40. Thus, these findings suggest that BHLHE40 inhibits AKT phosphorylation and its downstream signaling in PCa cells.

The KD of BHLHE40 does not seem to modulate AR protein level and the expression of AR target genes (Fig. 1O, P, Fig. S1D-F). This indicates that BHLHE40 may only weakly regulate AR transcriptional signaling. Our previous RNA-seq data, using the AKT inhibitor AKTi, indicate a slight induction of BHLHE40 in C4-2 cell (Fig. S2C, S2D) [11] indicating that AKT signaling inhibits BHLHE40 expression likely in an indirect manner as a potential feed-back loop.

Collectively, the obtained data suggest in addition to the well-established p-AKT signaling, the existence of another signaling pathway that mediates SAL-induced cellular senescence by BHLHE40 indicating an alternative AKT pathway for SAL-induced cellular senescence.

Given that SAL induces the expression of BHLHE40 and that the BHLHE40 KD reduces the senescence level in PCa cell lines, transcriptome data were analyzed and classified into subsets of genes dependent on their up- or downregulation to identify specific AR-BHLHE40 transcriptome landscapes and specific regulated pathways by these two transcription factors. To achieve this, upregulated DEGs from the BHLHE40 KD SAL-treated samples were separated from the downregulated and each set was separately analyzed for their overlap with the DEGs of SAL-treatment. The Venn diagram depicts approximately 6300 differentially upregulated genes in BHLHE40 KD at SAL that were common with DEGs in SAL-induced samples. Only about 800 genes were upregulated, being unique to BHLHE40 KD SAL-treated samples (Fig. 3I). This suggests a large overlap of genes that are regulated by BHLHE40 and are co-regulated by SAL. The same analysis was performed for the downregulated gene set. Similarly, a large number of DEGs were common but only nearly 140 genes were downregulated and specific to BHLHE40 (Fig. 3J). Taking together, we identified four sets of genes being co-regulated by BHLHE40 and AR.

The FOXO-mediated transcription pathway is noteworthy as it appeared in both sets of common factors, albeit with different gene sets. This pathway is linked to cell cycle arrest, quiescence, growth, and stress resistance of cells [52 –54]. In addition, specific sets of genes for BHLHE40 KD from Fig. 3I and J were analyzed for pathways. Most of the known BHLHE40-involved pathways were identified, such as the circadian clock and epithelial-mesenchymal transition (Fig. S4, Table S4).

Furthermore, pathway analysis of common genes between available BHLHE40 ChIP-seq data and our BHLHE40 KD SAL-treated transcriptome (Fig. 3L) yielded cell cycle regulation, DNA repair, and interestingly the FOXO-mediated transcription pathway (Fig. 3M and Table 5).

Moreover, the cytoHubba from Cytoscape, and the STRING were utilized to explore the established network connections between BHLHE40 and CCNG2, along with several cellular senescence markers (Fig. S7) [57 –59]. The data suggest that BHLHE40 and CCNG2 are connected to each other in multiple ways and also to factors that regulate chromatin organization and cell cycle. Notably, the analysis indicates close interaction with cellular senescence markers and the potent cell cycle inhibitors p15^INK4b, p16^INK4a, and p21^WAF1/Cip1.

Based on the bioinformatic predictions and the regulation of CCNG2 by BHLHE40 and AR, the hypothesis was that CCNG2 is part of the SAL-induced cellular senescence. Knockdown of CCNG2 by siRNA indicates that SAL enhances CCNG2 mRNA expression (Fig. 4C) and reduces the SAL-induced cellular senescence (Fig. 4D, S8A). This suggests that CCNG2 mediates SAL-induced cellular senescence being in line with enhanced growth for CCNG2 KD-treated SAL samples (Fig. 4E, S8B). Also, in the C4-2 xenograft tumors and ex vivo treated prostatectomy samples the expression level of CCNG2 was induced by SAL (Fig. 4F, G). Further, the Kaplan–Meier survival plot for CCNG2 in prostate cancer suggests that low CCNG2 expression is linked with diminished survival rates in PCa over time (Fig. 4H) suggesting that CCNG2 is part of a tumor suppressive program induced by SAL.

Since the KD of BHLHE40 significantly reduces the CCNG2 mRNA (Fig. 4I) it suggests that BHLHE40 up-regulates CCNG2 expression and the atypical CCNG2 is part of the BHLHE40 tumor suppressive pathway. Vice versa, the CCNG2 KD leads to a significant reduction in the expression of BHLHE40 mRNA and protein (Fig. 4J, K). These data suggest a reciprocal control loop between BHLHE40 and CCNG2. Accordingly, the decrease in cellular senescence by CCNG2 KD is associated with decline in CDKN2B (Fig. 4L). The induction of p21^WAF1/Cip1, by SAL was reversed after CCNG2 KD (Fig. 4K) supporting that CCNG2 regulates cellular senescence. In accordance with the results of BHLHE40 KD in the C4-2 cell line, the phosphorylation of AKT at S473 was strongly induced after CCNG2 KD but no changes were detected in the phospho-p70S6K (Fig. 4M). This indicates that CCNG2 represses p-AKT level and further confirms that the regulation of cellular senescence is not only mediated by p-AKT levels but rather through an alternative pathway. Collectively, these data indicate that CCNG2 and BHLHE40 act in tandem to promote cellular senescence in the AR signaling.

Surprisingly, for both KD and overexpression scenarios, a reduction in growth was observed (Fig. 5B, D, and S9B). These data suggest that not only does BHLHE40 act in the regulation of cellular senescence, but it might also play a role in other mechanisms in PCa cells including dormancy. Protein levels of p15^INK4b and p21^WAF1/Cip1 were increased in BHLHE40-overexpressed cells (Fig. 5G, I). These data confirm the SA b-Gal activity assay. Concerning BHLHE40-overexpression analysis of phosphorylation of AKT and p70S6K suggests that induction of p-levels of both kinases by SAL are reduced (Fig. 5J, S10). Thus, these data confirm the induction of AKT and p70S6K phosphorylation by BHLHE40 KD suggesting that the BHLHE40 mediated cellular senescence especially under SAL treatment is associated with a pathway that leads to suppression of AKT and p70S6K phosphorylation and thus provides a novel AKT-alternative pathway to control cellular senescence in PCa.

Based on this, we hypothesized that the reads obtained by AR-ChIP-seq also contain bHLH motifs. Interestingly, motif analysis for AR ChIP-seq data of C4-2 cells treated with SAL revealed many AR binding sites that also contain a bHLH motif in reads. Focusing on genes, it revealed around 2400 genes with both AR and bHLH binding sites, which may explain the observed large overlap of DEGs between these two transcription factors supporting our hypothesis. With pathway analysis of this gene set, the pathways of Circadian rhythm, Heme signaling, BMAL1-Clock activates circadian gene expression, and cellular senescence emerged (Table S5). This suggests a link between androgen signaling and circadian rhythm and cellular senescence.

To assess potential interactions between AR and BHLHE40, protein structure predictions and protein–protein docking modeling were conducted for the AR-DBD-hinge-LBD region with BHLHE40 protein. The 3D AR-DBD-hinge-LBD region prediction is depicted in (Fig. 6D, S12A), while the BHLHE40 structure and the region interacting with DNA, along with the involved amino acids, are illustrated in (Fig. 6E). Both AR and BHLHE40 structures exhibit high confidence scores and topology with a strong score for AR DBD-hinge-LBD and BHLHE40. The docking predictions show a high score for the interaction between AR and BHLHE40-bound DNA (Fig. 6F, S12B). The modeling suggests AR interacts with BHLHE40 through its ligand binding domain and that DNA binding domains of both proteins are positioned near each other.

Supplementary Material 1. Supplementary Material 2.