MDM4 actively restrains cytoplasmic mTORC1 by sensing nutrient availability

HeLa, 293 T, HCT116, MDA-MB231 cells were maintained in DMEM/10% FBS (Life Technologies, USA), p53^−/−Mdm4^−/−MEFs, p53^−/−Mdm2^−/−MEFs, and p53^−/−MEFs in DMEM high glucose/10% FBS (Cambrex). MCF10A cells in MEGM (Lonza, Switzerland). MDM4 and control (CTL) siRNA were by Invitrogen (Stealth RNAi), siRNA for S6K1 were from Ambion. siRNA and plasmids transfection were performed with RNAiMAX and Lipofectamine Plus respectively according to manufacturer’s instructions (Invitrogen). mTOR_1 shRNA was obtained from D. Sabatini through Addgene. Rapamycin (Sigma) was used 50nM unless specifically indicated. Torin2 was used 50nM. For amino acid starvation, cells were incubated for 3 h in amino acid free RPMI (US Biological) supplemented with 10% inactivated FBS, and stimulated with amino acid mixture for the indicated time. For serum and amino acid starvation, cells were incubated in EBSS (Invitrogen) for 50’, and stimulated with amino acids mixture or complete medium for the indicated time.

MCF10A cells were transfected with stealth MDM4-specific (siMDM4-MCF10A) or stealth control RNA (siCTL-MCF10A), and after 48 h were lysed. The proteomic analysis was performed on proteins extracted from cytoplasmic cell lysate of MCF10 cells, through a label-free data-independent differential proteomic analysis by nUPLC-MS^E. Details of the analysis are reported in [7].

For mammosphere formation assay, cell culture dishes have been coated with pHEMA (poly(2-hydroxyethyl methacrylate) 10 mg/ml, dried and rinse with PBS. MDA-MB231 were interfered for siRNA control or siMDM4 for 16 h. Afterwards, cells were detached and seeded at 2000 cells/well in pHEMA coated 6 wells dishes for 72 h in DMEM/F12 supplemented with 2 mM Glutamine, 100U/ml Penicillin/streptomicin, 5%FBS, 20 ng/ml EGF, 0.5 mg/ml Hydrocortisone, 10ug/ml Insulin.

For immunoprecipitation (IP), cells were lysed in CHAPS lysis buffer (40 mM Hepes pH7.4, 120 mM NaCl, 2 mM EDTA, 0.3% CHAPS) containing mix of protease inhibitors (Boehringer), plus 5 mM NaF, 10 mM glycerophosphate and 1 mM Na₃VO₄. For IP lysates were pre-incubated with protein G-Agarose (Pierce) and then with the indicated antibody, under gentle rocking at 4 °C overnight. For Western blot (Wb) cells were lysed in RIPA buffer. Membranes were developed using the enhanced chemiluminescence (ECL Amersham) by chemiluminescence imaging system, Alliance 2.7 (UVITEC Cambridge) and quantified by the software Alliance V_1607. Primary antibodies used: MDM4 BL1258 (Bethyl laboratory), MDM4 C82 (Sigma), MDM4 8C6 (Millipore) p53 FL393 (Santa Cruz), α-tubulin DM1A (Sigma), actin C-40 (Sigma), mTOR (Santa Cruz), mTOR (Cell Signaling), anti-FLAG M2 affinity gel (Sigma), phosphoSer473-AKT (Cell Signaling), phosphor-Thr389-S6K (Cell Signaling), Akt (Cell Signaling), S6K1 (Santa Cruz), Raptor (Cell Signaling), Raptor (Santa Cruz).

Fractionation of lysates into heavy membrane and light membrane/cytosolic fractions was performed according to Menon et al. 2014.

Kinase assays were performed as previously described [16] with some modifications. Flag-mTOR immunoprecipitate was washed twice in CHAPS lysis buffer and twice in 25 mM HEPES (pH 7.4), 20 mM potassium chloride. Kinase assays were performed at 30 °C for 20 min in a final volume of 30 μl consisting of mTORC1 kinase buffer (25 mM HEPES [pH 7.4], 50 mM KCl, 10 mM MgCl₂, 250 μM ATP) and inactive GST-S6K1 purified (by GST-Agarose gel, Sigma), from Hela cells transfected with GST-S6K1 plasmid and treated with EBSS and 20 μM LY294002 for 1 h. Reactions were stopped by the addition of sample buffer and boiling for 5 min. When used, 150 ng of GST-MDM4 was added to mTORC1 10 min before the addition of ATP to the kinase assay.

Cell proliferation was determined by Cell Titer Blue colorimetric assay or Cell Live/Dead kit according to the manufacturer’s instructions (Promega and Invitrogen, respectively). Cell cycle profiles and forward scatter determination (FSC-H) were evaluated by fixing cells in cold 70% ethanol for 1 h on ice and staining DNA for 30 min at room temperature with 50 μg/mL propidium iodide (PI) in PBS containing 1 mg/mL RNase A. FSC-H evaluation was performed by previous gating of cells in G1 phase. FACScan flowcytometer (Becton Dickinson, USA) was used and data analysed by CellQuest Software (Becton-Dickinson).

Hela cells were fixed with 4% formaldehyde 5’ 37 °C, permeabilized with TritonX100 0.2% 15’ RT, and blocked with 0.25% BSA. Cells were stained with DAPI and primary antibodies: anti-MDM4 (1:100 Origene 4B5), and anti-mTOR (1:400 Cell Signaling). Cyanine (Cy3)-conjugated and Cyanine (Cy2)-conjugated secondary antibodies were used.

Control (WT) and Mdm4 transgenic (TG) mice [5] were maintained and treated in accordance with the Guidelines on the protection of animals used for scientific purposes (European Directive 63/2010/EU and Italian Law DL116/1992 and DL 26/2014). Relative ethical approval has been obtained by Animal Welfare Body “Fondazione S. Lucia” (Protocol Number: 969/2015-PR). For in vivo assessment of mTOR activity, 13–15 week old male mice were fasted overnight and after 16 h intraperitoneally injected with leucine (120 mg/kg) or saline solution (control) in 0.2 ml volume. Ten minutes after injection, mice were sacrificed, tissues were snap frozen in liquid nitrogen and samples processed in RIPA lysis buffer for Western blot analysis.

The FH1t-UTG Mdm4 3' UTR-GFP lentiviral construct was obtained by Marine’s Lab by cloning shRNA sequence for MDM4-3'UTR (ACAGTCCTTCAGCTATTTCATTTCAAGAGAATGAAATAGCTGAAGGACTGTTTTTT) into the FH1tUTG vector, which constitutively expresses GFP [17]. MCF10A, Hela and 293 T cells were infected with FH1t-UTG Mdm4 3' UTR-GFP lentivirus to generate TET-shMDM4 inducible cell line by doxycycline (DOX).

Survey by Ingenuity Pathways Analysis™ of proteomic profile of immortalized MCF10A breast cell line interfered by siRNA to MDM4 compared to control cells (siMDM4 vs siCTL) revealed that some proteins upregulated by knockdown of MDM4 belong to the function “Regulation of eIF4 and p70S6K Signalling” (Table 1) [7, 18‐20].

To validate these proteomic data, we evaluated the levels of the 70 kDa ribosomal protein S6 kinase, p70S6K1 (hereafter, S6K1) following knockdown (KD) of MDM4 in MCF10A. Since phosphorylation of S6K1 at the threonine 389 is a key signal in the activation of the S6K1 downstream signalling pathway [21], we analysed the levels of both S6K1 and its active form, pS6K1³⁸⁹. In fact, MDM4-KD caused up-regulation of pS6K1 while did not alter the total levels of the protein (Fig. 1a), confirming bioinformatics prediction analysis. Similar results were obtained by constitutive inducible knockdown of MDM4 in MCF10A (Additional file 1: Fig. S1a) and in HeLa cells (Additional file 1: Fig. S1b) that express doxycycline-inducible shRNA targeting a different region of MDM4, thus excluding off target effect of siRNAs. Furthermore, re-expression of MDM4 rescued the increase of pS6K1 levels caused by MDM4-KD (Additional file 1: Fig. S1b), confirming the specific activity of MDM4 on S6K1 phosphorylation.

These data were observed also in p53^+/+HCT116 and syngenic p53^−/−HCT116 cells, pointing to a p53-independent effect of MDM4 on S6K1 (Fig. 1b).

Since S6K1 phosphorylation is mostly controlled by the mammalian target of rapamycin complex 1, mTORC1 [22], we analysed whether MDM4 activity is developed through inhibition of this complex. Human embryonic kidney 293 T cells, a mammalian cell line in which mTORC1 activity has been well characterized [23], were transfected with MDM4 siRNA and the levels of pS6K1 analysed in the presence of the mTORC1 inhibitor rapamycin (RAPA). In this cell line too, MDM4-KD increased the levels of pS6K1 but it was ineffective in the presence of RAPA (Fig. 1c). Basal S6K1 phosphorylation was inhibited by RAPA confirming the block of mTORC1 function. Similar results were obtained in MCF10A (Additional file 1: Fig. S1a) and in HeLa cells (Additional file 1: Fig. S1c). These data suggest that MDM4 inhibits mTORC1-mediated S6K1 phosphorylation. Given the inactivation of p53 both in 293 T and in HeLa cells, these data further support the p53-independent activity of MDM4.

The mTORC1-mediated phosphorylation of S6K1 is tightly regulated by nutrient availability and has been particularly well characterized by amino acids signalling [24, 25]. To further analyse the inhibitory function of MDM4 towards mTORC1, the activity of this last was blocked by cell starvation and then re-stimulated by amino acids (aa) addition. Indeed, cell treatment with Earle's Balanced Salt Solution (EBSS) depleted pS6K1 levels that were rescued by addition of amino acids (aa) mixture (Fig. 1d, Additional file 1: Fig. S1d). Under these conditions, MDM4-KD enhanced the increase of pS6K1 caused by aa addition, indicating that MDM4 antagonizes S6K1 phosphorylation by restraining mTORC1 activity (Fig. 1d, Additional file 1: Fig. S1d). Consistently, MDM4-KD was ineffective in the presence of RAPA (Fig. 1d). Similarly, amino acid deprivation restrained mTORC1 activity and the presence of MDM4 reduced the recovery of pS6K1 (Fig. 1e). Conversely, the over-expression of MDM4 strongly decreased the levels of pS6K1 induced by aa supplementation (Fig. 1f), overall indicating that MDM4 inhibits mTORC1 in response to aa depletion. To further confirm that MDM4 effect on pS6K1 are mediated through regulation of mTOR, the knockdown of mTOR prevented the upregulation of pS6K1 by siMDM4 (Fig. 1g). Similar effect were observed by pharmacological inhibition of mTOR with Torin2, a potent ATP-competitive inhibitor [26] although with less efficiency (Additional file 1: Fig. S1e).

To further confirm this MDM4 activity in normal cells, we used the genetic model of Mdm4 knock out in mouse embryo fibroblasts (MEFs) [27]. To exclude the effect of p53, p53 ^−/− Mdm4 ^−/− MEFs were compared to p53 ^−/− MEFs. After cell treatment with EBSS, induction of pS6K1 by growth medium reconstitution was indeed significantly increased in p53 ^−/− Mdm4 ^−/− MEFs compared to p53 ^−/− MEFs (Fig. 2a). In comparison, the p53 ^−/− Mdm2 ^−/− MEFs showed pS6K1 levels similar to those of p53 ^−/− MEFs or even lower (Additional file 1: Fig. S1f), indicating the specificity of the activity of Mdm4 and excluding a general function of Mdm family. Overall, these data demonstrate that MDM4 represses the phosphorylation of S6K1 via inhibition of mTORC1 complex.

To ascertain that MDM4 displays this activity in vivo too, we used a transgenic mouse model overexpressing Mdm4 (TG) [5]. Since mTORC1 activity is strongly regulated in hepatocytes, we analysed pS6K1 levels in the liver of TG and age-matched control (WT) mice. Animals were fasted overnight and after 16 h injected intraperitoneally with the amino acid leucine (Leu), a specific activator of mTORC1 [28]. In control WT mice, Leu increased phosphorylation of S6K1 compared to saline treated mice (Fig. 2b). Of note, such increase was almost abolished in Mdm4 TG mice (Fig. 2b), indicating that the overexpression of Mdm4 restrains mTORC1 activity in vivo too. Evaluation of mTOR levels in these samples did not show significant differences between WT and TG mice thus excluding an impact of Mdm4 on the total amount of the kinase.

Previous data indicate that MDM4 inhibits mTORC1 activity by impairing its ability to phosphorylate the substrate S6K1. To understand whether MDM4 inhibits directly mTOR, we analysed the in vitro kinase function of the complex in presence or absence of MDM4. Flag-mTOR was immunopurified from HeLa cells overexpressing Flag-mTOR and tested in vitro for the phosphorylation of GST-p70S6K1 purified from HeLa cells. Incubation of GST-p70S6K1 with Flag-mTOR increased S6K1 phosphorylation compared to control cells (Fig. 3a). Interestingly, when mTORC1 complex was immunopurified from HeLa cells silenced for MDM4, the levels of phosphorylated GST-p70S6K1 were increased, suggesting that MDM4 directly impairs mTORC1 kinase activity (Fig. 3a). Consistently, pre-incubation of the mTORC1 complex with GST-MDM4 decreased significantly phosphorylation of S6K1 (Fig. 3b), indicating that the presence of MDM4 is sufficient to inhibit mTOR kinase activity.

Since both proteins are mainly cytoplasmic [15, 29, 30], these data prompted us to investigate a possible interaction between MDM4 and mTOR, the kinase effector of the mTORC1 complex. Indeed, overexpressed MDM4 co-immunoprecipitated Flag-mTOR, indicating that the two proteins interact (Fig. 3c). Interestingly, the amount of co-immunoprecipitated mTOR was lower in presence of aa, supporting the inhibitory activity of MDM4 towards mTORC1 under nutrient deprivation (Fig. 3d). Analysis of endogenous proteins confirmed the interaction between MDM4 and mTOR during starvation whereas this was almost undetectable in presence of aa (Fig. 3e). To ascertain whether the binding between the two proteins mediates the MDM4 inhibitory activity, map of the binding of MDM4 to mTOR was performed by using different MDM4 deletion mutants (Fig. 3f) whose cytoplasmic localization was previously reported [29]. The results revealed that the MDM4ΔBD, lacking the aminoacids 1–106 (consisting of the p53 binding domain) was unable to bind mTOR (Fig. 3f and g), indicating that the N-terminal domain of MDM4 is involved in the interaction. Of note, the MDM4ΔBD mutant did not decrease pS6K1 levels compared to the full-length MDM4 (Fig. 3h), indicating that the interaction between MDM4 and mTOR is required for the inhibition of this last. Overall, these data indicate that MDM4 binds mTOR during aa starvation and contributes to silence the kinase activity of the complex.

Depletion of amino acids induces re-localization of mTORC1 from lysosomal membranes to the cytosolic compartment and this correlates with decreased mTORC1 activity. MDM4 is mainly a cytoplasmic protein [29]. We therefore analysed in which compartment the interaction between MDM4 and mTOR occurs. Immunofluorescence showed that under starvation, overexpressed MDM4 and mTOR signals localize in the cytoplasm, whereas their signals are mostly independent when amino acids are not limiting in the culture medium (Fig. 4a). Particularly, upon amino acid supplementation mTOR assumed the characteristic punctate pattern, i.e. the lysosomal-active state [31], while MDM4 signal was not altered. Overall, these data suggest that MDM4 interacts with the soluble cytoplasmic pool of mTOR, contributing to keep it inactive. Accordingly, silencing of MDM4 significantly increased the percentage of cells with punctuated mTOR (Fig. 4b and c). This occurs both in EBSS conditions where a low fraction of mTOR is present at the lysosomes, as well as under aa treatment. Furthermore, fractionation of HeLa cell lysate [32] showed increased levels of mTOR in the lysosome-enriched fraction (hm, the heavy membrane) in siMDM4 compared to control cells (siCTL) and concomitantly reduced levels in the light membrane/cytosolic fraction (lm/cyt) (Fig. 4d). Similar results were obtained in 293 T cells (Additional file 1: Fig. S2). These data confirm the role of MDM4 as mTOR-cytoplasmic anchor and indicate that the balance between MDM4 and mTOR levels is important to determine the fraction of lysosomal active mTORC1.

The main functions of p70S6K1 are the stimulation of protein synthesis and the control of cell size and growth. To evaluate the impact of MDM4 towards these cell features, we analysed cell size by flow-cytometer. The results showed an increase of the mean cell size (FSC-H) in siMDM4 compared to siCTL p53^+/+HCT116 cells (Fig. 5a) whereas the coefficient of variation of the FSC-H distribution was very similar in the two populations (CV_siMDM4, 19,13 vs CV_siCTL 21,45). Similar results were observed in p53^−/−HCT116 (Fig. 5a), thus excluding that these effects are due to MDM4 activity towards its main target p53. These data were further confirmed in DOX-inducible MDM4-KD MCF10A (Fig. 5b), in HeLa cells (Additional file 1: Fig. S3a), and in p53 ^−/− Mdm4 ^−/− MEFs compared to p53 ^−/− MEFs (Fig. 5c), indicating the overall ability of MDM4 to control cell size. Importantly, the MDM4-dependent increase of cell size (FSC-H) is completely abolished upon RAPA treatment (Fig. 5d), indicating that these effects are mediated by the impairment of mTORC1 activity. Additionally, silencing of MDM4 significantly increased cell viability and cell number, whereas it was ineffective in the presence of RAPA (Fig. 5e and f, Additional file 1: Fig. S3b). Of note, the interference of S6K1 abrogated the effects of MDM4 on cell viability, confirming that MDM4 activity is mediated by this kinase (Fig. 5f). Overall, these data indicate that MDM4 controls mTORC1 activities towards cell size and growth.

Mammalian TORC1 activities are instrumental to tumor growth. Consistently, many human tumors are characterized by increased activity and/or levels of mTOR signalling [13, 33, 34]. Given the inhibitory function of MDM4 towards mTORC1, we investigated a possible relationship between MDM4 and mTOR levels in human tumors by interrogating the Atlas database [35]. MDM4 expression has been associated with low risk/good prognosis in breast cancers independently of p53 [36] and high mTOR mRNA expression has been reported in breast tumors as well [37]. For this reason, we analysed MDM4 and mTOR in the database of human breast cancer. Interestingly, regression analysis showed a significant inverse correlation between mTOR and MDM4 mRNA levels (Fig. 6a and b). This is highly significant in the tumors lacking wild-type p53 (Fig. 6a, r ² = 0.11) suggesting the possible development of MDM4 anti-tumor activities especially in the absence of functioning p53. As control, no correlation was observed in normal breast tissues (data not shown). To support these data and ascertain the anti-oncogenic properties of MDM4, mammosphere forming assay was performed in breast tumor cell line MDA-MB-231, carrying R280K mutant p53 [38]. MDM4-KD caused increased pS6K1 levels in this cell line too (Fig. 6c). Interestingly, cells depleted for MDM4 showed an increased ability to form mammospheres compared to control cells (Fig. 6d and e) thus confirming the anti-oncogenic properties of MDM4. Rapamycin strongly reduced mammosphere formation indicating the sensitivity of these cells to mTORC1 inhibition and most importantly, abolished the effects of MDM4 depletion (Fig. 6d and e), supporting the functional link between mTOR and MDM4 in the maintenance of mTOR oncogenic properties.