Distinct outcomes of CRL–Nedd8 pathway inhibition reveal cancer cell plasticity

Inhibition of protein degradation by blocking Cullin-RING E3 ligases (CRLs) is a new approach in cancer therapy though of unknown risk because CRL inhibition may stabilize both oncoproteins and tumor suppressors. Probing CRLs in prostate cancer cells revealed a remarkable plasticity of cells with TMPRSS2-ERG translocation. CRL suppression by chemical inhibition or knockdown of RING component RBX1 led to reversible G0/G1 cell cycle arrest that prevented cell apoptosis. Conversely, complete blocking of CRLs at a higher inhibitor dose-induced cytotoxicity that was amplified by knockdown of CRL regulator Cand1. We analyzed cell signaling to understand how varying degrees of CRL inhibition translated to distinct cell fates. Both tumor suppressor and oncogenic cell signaling pathways and transcriptional activities were affected, with pro-metastatic Wnt/β-catenin as the most upregulated. Suppression of the NF-κB pathway contributed to anti-apoptotic effect, and androgen receptor (AR) and ERG played decisive, though opposite, roles: AR was involved in protective quiescence, whereas ERG promoted apoptosis. These data define AR–ERG interaction as a key plasticity and survival determinant in prostate cancer and suggest supplementary treatments that may overcome drug resistance mechanisms regulated by AR–ERG interaction.


Data quantification
The effect of each treatment was measured in 4 well replicates split into two 384 (or 96)-microwell plates (2 replicates per plate). The fluorescent signals (Hoechst and CellEvent) were measured in 9 fields per well-replicate. Based on these signals, the "Cell Health Profiling" program performs cell segmentation and obtains information about each individual cell, including the intensity of the signal from the channel (Hoechst and CellEvent), the area of the nuclei and the total number of cells per field. Based on the CellEvent signal intensities from negative siAllStars and positive siCellDeath controls , the threshold that distinguishes live cells from apoptotic was established. Based on this threshold, the median value of the percentage of dying cells was then calculated. Statistical analyses have been performed using the statistical software R (http://www.r-project.org/). Box and whiskers plots represent the distribution of the data, with the box delimiting the central half of the data (from first to third quartiles); the segment is the median of the data. The whiskers delimit the rest of the data if its length does not exceed 1.5 times the size of the box, other data points are indicated by circles. P-values are calculated using the two-sided Wilcoxon rank-based test. For analysis and comparison of the effects of siRNA in various cell lines we used robust Z-score (RZ), which allows the data to be less dependent on "outliers". The Z-score is the distance from the mean of the whole plate normalized by its standard deviation. Its robust version, RZ, calculates as follow (Birmingham et al, 2009): where RZ w is a robust Z-score for the well w, X w is the value for the well (either percentage of dying cells or cell number); Md is a median value of the X w for the whole plate, MAD is a median absolute deviation of X w for the whole plate and k is the constant scale factor, which depends on the distribution, and is equal to 1.4826 in case of the Gaussian distribution.

DNA synthesis assay by EdU incorporation
The quantity of proliferating cells in the population was determined by measurement of DNA synthesis (by EdU incorporation) using Click-iT® EdU Alexa Fluor® 647 Flow Cytometry Assay Kit from Invitrogen (C-10419). This involved the cells being seeded in black plates with transpare nt bottoms suitable for fluorescent measurements (Fisher Scientific, 781091). The treatments were performed on the day after cell seeding. At indicated times, VCaP cells were treated with EdU for 5 hours and then fixed and stained according to the manufacturer's protocol. Finally, Hoechst reagent was added to the cells followed by incubation for 30 minutes. The labeled cells were covered with Glycerol and PBS++ solution (in a ratio 1:1) and stored at 4°C. Image acquisitions were performed using CellInsight™ NXT High Content Screening Platform (Thermo Scientific). The images were analyzed and quantified by the "Cell Health Profiling" program, installed within CellInsight™. The identification of cells (cell segmentation) was based on the detection of nuclei by the Hoechst channel. The EdU signal was quantified for each nucleus, and the results were presented as a percentage of cells having nuclear EdU staining above a threshold.

Cell-cycle analysis
The cell cycle was analyzed by the measurement of total DNA content using flow cytometry. Cells were grown in culture medium with or without drug treatment. They were harvested with trypsin, neutralized by culture medium and washed once in PBS. Then the cells were fixed with 70% fridge-cold ethanol for 30 minutes, and BSA was added to a final concentration of 0.5%. The cells were spun at 3000 rpm for 7 minutes, the supernatant was discarded, and the cells were resuspended in 0.25 % BSA in PBS and spun again at 3000 rpm 7 minutes. The supernatant was discarded, and replaced by a 50 µg/ml 7-AAD (7aminoactinomycin D) solution in PBS. The 7-AAD labeled cells were analyzed by BD™ LSR II flow cytometer from BD Biosciences.

Senescence test
Analysis of senescence was performed by measuring of β-galactosidase activity according to the previously described protocol (Debacq-Chainiaux et al, 2009). The protocol was slightly modified when the test was performed with spheroids. This involved the suspension of VCaP cells being distributed in ultra-low attachment U-bottom plates (Falcon,353910) in concentrations of 500 cells/well/100 µl and being incubated during 10 days. Typically, 30 spheroids per condition were used. Next, the spheroids were harvested, pelleted at 600 rpm for 5 min, resuspended in 150 µl of 1.5% low-melting agarose (Sigma, A9414) and distributed into Lab Tek chambers (Dominique Dutscher, 055082). After polymerization for about 30 minutes, the gels were washed twice with PBS, fixed with 2% formaldehyde and 0.2% glutaraldehyde in PBS during 7 minutes. This was followed by a double wash with PBS, and the addition of a staining solution (containing citric acid/Na phosphate buffer, 5 mM K 4 [Fe(CN) 6 ]x3H 2 O, 5 mM K 3 [Fe(CN) 6 ], 150 mM sodium chloride, 2 mM magnesium chloride and 1 mg/ml X-gal in distilled water). The spheroids were then incubated at 37°C during 5 hours. Then the gels were washed multiple times with PBS to remove the background yellow staining of agarose. As a final step, the gels were washed with methanol for 1 min and viewed by bright field microscopy.

Immunofluorescence microscopy
Cells were grown on plasma-treated glass slides. The culture medium was removed and replaced by 4% paraformaldehyde (PFA) for 15 min at RT. Then PFA was replaced with 0.2% Tween for 5 min at RT. Treatment with Tween was followed by the addition of NH 4 Cl 0,1M for 10 min at RT. Then the slides were washed briefly in PBS +Ca 2+ +Mg 2+ (PBS++, Sigma, P4417) and blocked with 3 % BSA in PBS++ which had been filtered through a 0.2 µm filter for 30 min. Primary antibodies were added in 1.5 % BSA (filtered) and left for 2 hours at RT. This was followed by 3 washes with PBS++, and then the slides were incubated with secondary antibodies and phalloidin in 1.5 % BSA (filtered) for 45 min at RT. This was followed by a 5 min wash in PBS++, then a 5 min wash in Hoechst, and then again 5 min with PBS++. Then, these glass slides were placed on standard microscope slides with mounting solution (DAKO, S302380) and dried for 24 hours. Finally, the slides were analyzed by Zeiss Axioimager Z1 Apotome from Zeiss.

Western blotting and ELISA
Cellular proteins were extracted using RIPA lysis buffer (Sigma, R0278) complemented with protease inhibitor cocktail (Complete Mini from Roche Diagnostics, Cat. No. 11 836 153 001) and additional inhibitors (10 mM ortho-phenanthroline, 30 mM N-Ethylmaleimide, 5 mM sodium ortho-vanadate, and 5 mM sodium fluoride). After quantification with a BCA protein assay kit (Pierce, 23225), an equal range of concentration (typically 2.5 ng of protein per sample) was run on a NuPAGE Novex Bis-Tris Gel (Life Technologies, NP0322BOX, EC60252BOX, NP0323BOX) in MES buffer and then transferred onto the nitrocellulose membrane (Amersham™ Protran®, GE Healthcare, 10600001). The membranes were blocked in 5% nonfat milk/TBST for 40 min at 37°C, incubated with primary antibodies in 5% nonfat milk/TBST for 1 hour at RT or overnight at 4°C. This step was followed by incubation with secon dary HRP-conjugated antibodies. Detection was performed with a chemiluminescent reagent depending on the concentration of the target protein (Plus-ECL, Perkin Elmer, NEL105001EA; ECL Prime, GE Healthcare, RPN2232; SuperSignal West Femto Substrate, Thermo Fisher Scientific, 34095). Secreted PSA was analyzed with Anogen Human Free PSA ELISA Kit. A list of antibodies is given in Supplementary  Table S1. siRNA transfection Cells were transfected with siRNA using Lipofectamine® RNAiMAX Transfection Reagent (Invitrogen, 13778) according to the manufacturer's protocol with minor modifications: RNAiMAX was taken 0.75 µl per well of 96-well plate and 386-well plate. Screening of the CRL genes was performed using siRNAs from ON-TARGETplus® SMART pool® siRNA Library-Human Ubiquitin Conjugation Subset 1, complemented with RBX1 and SAG(RBX2) ON-TARGETplus® SMART pool® siRNAs from Dharmacon. Transfection of the SMART pool was done at a final concentration of 20 nM of siRNA and individual siRNAs were added in concentrations of 10 nM, unless otherwise indicated. The optimal knockdown effect was observed with 3-day (LNCaP, PC3) or 5-day (VCaP) siRNA treatment. As controls for transfection AllStars Negative Control siRNA (SI03650318, Qiagen) and AllStars Hs Cell Death siRNA Positive cell death phenotype control (SI04381048, Qiagen) were used. All controls were used in concentrations equal to the concentration of siRNA in the experiment. The siRNA against the ERG gene was prepared by Eurogentec. The sequences of siERG were taken from publication of Tan et al., 2009(Tan et al, 2014: sense -5'-CGACAUCCUUCUCUCACAUAU-3'; antisense -5'-AUGUGAGAGAAGGAUGUCGUG -3'. Three siRNAs for AR were obtained from Dharmacon and used as a pool. siRNA sequences are listed in Supplementary Table S2.

Luciferase Reporter Assays
Cells were grown in white 96-well plates with transparent bottom suitable for luminescence assays (Grenier, 655088) until they reached 70%-80% confluence. Next, cells were co-transfected in triplicates with a transcription factor-specific Firefly luciferase reporter and a constitutively-active Renilla luciferase reference vector (1:10 w/w ratio, 200 ng of total DNA per well) and with 0.5 µl/ well Lipofectamine® 2000 Transfection Reagent (ThemoFisher, 11668019) according to the manufacturer's protocol. The cells were treated with MLN on the day after transfection and analyzed after 24 hours. Luciferase measurements were performed using the Dual-Luciferase Reporter Assay (Promega), according to the manufacturer's instructions using GloMax®-Multi Detection System (Promega). The ratio of Firefly-to Renilla-luciferase activities was calculated. All values were presented as means ± SD. Reporter plasmids are listed in Supplementary Table S3. RNA extraction, RT-PCR, and qPCR RNA was extracted with an RNeasy Mini Kit (QIAGEN, 74104). 1.5 µg RNA was reverse-transcribed in a total volume of 20 µl using a SuperScript® VILO cDNA Synthesis Kit (Life Technologies, 11754050) with random primers according to the manufacturer's protocol. Reverse transcription reactions were diluted to 200 µl of distilled water and further used in concentrations of 2.5 µl per reaction of quantitative PCR (qPCR). qPCR was carried out with a Platinum Quantitative PCR SuperMIX-UDG Kit (Life Technologies, 11730-017) using a StepOnePlus Real-Time PCR system (Applied Biosystems, 4376600). All experiments were run in triplicates, and the results were normalized to 18S rRNA expression. Primer sequences are listed in Supplementary Table S4.

Chemical inhibitor screening
The effect of specific chemical inhibitors on cell viability was analyzed by measuring the number of apoptotic cells as described in "Measurement of apoptosis" section. The treatments were performed in 100 µl triplicates at ~60% of cell confluence. The inhibitors and MLN were diluted in cell medium from DMSO stocks (0.5% DMSO in final solution). Pure 0.5% DMSO was used as a control condition. CellEvent reagent was added during the treatment according to the manufacturer's protocol. At the end of treatment, Hoechst dye was added, and cells were incubated for 30 minutes. The data acquisition and analysis were performed as described above in "Measurement of apoptosis" and "Data quantification" sections. A list of inhibitors is given in Supplementary Table S5 Cell lysates were analyzed by western blotting with protein-specific antibodies and anti-GAPDH for loading control. B, Correlation between cellular level of Cand1 protein (quantified by ImageJ using Supplementary Figure S10A, and normalized to GAPDH) and the sensitivity of a cell line to MLN (IC50 values were estimated using the data shown in Figure 1A). C, CAND1 knockdown potentiates the toxic effect of MLN. VCaP cells were transfected with 5 nM of ON-TARGETplus® SMART pool® CAND1 siRNA or with the same amount of siCTL and treated wi th MLN on the next day. The percentage of apoptotic cells was measured as described in Materials and Methods. The data are shown as a boxplot diagram with P-values compared to control (ctl) condition (* means P < 0.05, *** means P < 0.001, Wilcoxon test). NT-non-transfected control.
Supplementary Figure S12. MLN interaction with autophagy. A, Western blot analysis of autophagy markers LC3 and p62/SQSTM1 in VCaP cells. Cells were treated for 24 h with increasing concentrations of MLN or 100 nM of Rapamycin (Rap, autophagy stimulation control) and analyzed by western blotting on PVDF membrane with protein-specific antibodies and anti-GAPDH for loading control. The plot below shows the relative protein levels normalized first to GAPDH, then to vehicle control (the corresponding values for Rapamycin are shown on the right). B, Western blot analysis of autophagic flux. Cells were first treated with indicated concentrations of MLN or 100 nM of Rapamycin for 24 h followed by the addition of 10 nM Bafilomyicin A1 for 2 h (Baf A, an inhibitor of lysosomal H+-ATPase and degradation pathway). The control cells were not treated with Bafilomyicin A1. Cell lysates were analyzed by western blotting as above. C, Effect of autophagy inhibitors on MLNinduced apoptosis. Cells were grown in 10% ChsM and treated for 5 d with indicated drug concentrations. The percentage of apoptotic cells was measured as described in Materials and Methods. The data are presented as a boxplot diagram along with the corresponding P-values for 500 nM-MLN-points (Wilcoxon test). Figure S12. Recent studies have shown that CRL inhibition by MLN induces autophagy that protects cancer cells from apoptosis; moreover, blocking autophagy markedly enhanced drug efficacy (Zhao et al, 2012). However in our case, examination of autophagy markers revealed that, although MLN increased the level of the lipidated form of LC3, it also resulted in an increa se, but not a degradation, of autophagy substrate p62/SQSTM1 (Supplementary Figure S12A). The la tter suggests that the effect of MLN on autophagy is complex and may include a block of the autopha gosomes at the terminal stages (Klionsky et al, 2016). This is also corroborated by the measurements of autophagic flux (Klionsky et al, 2016), which showed no clear effect of MLN on autophagy markers in the presence of Bafilomycin A1, an inhibitor of lysosomal degradation (Supplementary Figure S12B). Specifically, a similar MLN-Bafilomycin A1 interaction was also observed by Zhao et al. (Zhao et al, 2012). Finally, autophagy inhibitors did not affect significantly VCaP apoptotic response to MLN, thus excluding the major role of autophagy stimulation in the observed MLN phenotypes (Supplementary Figure S12C).

Supplementary Figure S13. MLN interaction with NFkB pathway.
A, MLN induces accumulation of p-p65 in cytoplasmic speckles. Immunofluorescence analysis of p-p65 (phospho-Ser536) in VCaP cells treated with indicated concentrations of MLN. Cells were probed with p-p65-specific antibody (p-p65, green), actin-specific antibody (Actin, red), and Hoechst dye (DNA, blue). Scale bar is 20 m. B, Effect of IKK inhibitors on MLN-induced apoptosis. Cells were treated for 5 d with indicated drug concentrations in 10% StdM. The percentage of apoptotic cells was measured as described in Materials and Methods. The data are presented as a boxplot diagram along with the corresponding P-values for 500 nM-MLN-points (Wilcoxon test). The four-point plot inside the diagram shows IKK-to-IKK selectivity of the drugs (Tian et al, 2015) (see also Supplementary Table S5). Figure S14. Opposite roles of AR and ERG in MLN-induced cell fate. A, Opposite effects of AR and ERG knockdowns on MLNinduced apoptosis. VCaP cells grown in 10% ChSM or 10% StdM were transfected with i ndicated siRNAs and treated with MLN on the next day. The percentage of apoptotic cells was measured as described in Materials and Methods. The data are shown as a boxplot diagram along with the corresponding P-values (Wilcoxon test, see Materials and Methods for more details). B, The same data presented as line charts (mean ± s.d.). C, Effect of dihydrotestosterone (DHT) on MLN-induced apoptosis. VCaP cells grown in 10% ChSM with (DHT) or without (NT) 1 nM DHT were treated with MLN for 5 days. The percenta ge of apoptotic cells was measured as described in Materials and Methods. The data were analyzed as described above. [1] IC50 for cell growth inhibition in sensitive cell lines.

SUPPLEMENTARY TABLES
[2] Values are given for IKK, and for IKK in parenthesis. N.R.-not reported.