An mTORC1-Mdm2-Drosha axis for miRNA biogenesis in response to glucose- and amino acid-deprivation

Abstract

mTOR senses nutrient and energy status to regulate cell survival and metabolism in response to environmental changes. Surprisingly, targeted mutation of Tsc1, a negative regulator of mTORC1, caused a broad reduction in miRNAs due to Drosha degradation. Conversely, targeted mutation of Raptor, an essential component of mTORC1, increased miRNA biogenesis. mTOR activation increased expression of Mdm2, which is hereby identified as the necessary and sufficient ubiquitin E3 ligase for Drosha. Drosha was induced by nutrient and energy deprivation and conferred resistance to glucose deprivation. Using a high-throughput screen of a miRNA library, we identified four miRNAs that were necessary and sufficient to protect cells against glucose-deprivation-induced apoptosis. These miRNA was regulated by glucose through the mTORC1-MDM2-DROSHA axis. Taken together, our data reveal an mTOR-Mdm2-Drosha pathway in mammalian cells that broadly regulates miRNA biogenesis as a response to alteration in cellular environment.

Fig. 1

Activation of mTOR negatively regulates miRNA biogenesis (a). miRNA microarray analyses reveal an mTOR-dependent down-regulation of mature and pre-miRNA. Top panel shows data for individual miRNA. The Y-axis shows the log2 ratio of signal, while the X-axis shows the –log10 P value. The lower panel shows the number and percentage of miRNAs that are increased (ratio Y >1.0) or decrease (Y<1.0). Rapamycin (R, 5nM) was added to the media 4 days before harvesting Tsc1^−/− MEFs. Each dot represents the mean value from five repeats of an individual miRNA. Between two independent microarray analyses, some variation in individual miRNAs was observed. However, the overall trend is highly reproducible. (b). ShRNA silencing of TSC1 in HeLa cells decreases miRNA levels. The top panel shows the efficacy of TSC1 reduction with three independent shRNAs (e, f, and kl). The middle panel presents the miRNA microarray analysis data for individual human miRNAs. Data shown are the ratio of scrambled vs. shRNA f-silenced HeLa cells. The bottom panel shows the summary data (n=3). A similar pattern was observed in Fig. S1, Table S3, S4, with DROSHA knockdown. (c).Knockdown of TSC1 in HeLa cells reduces pre- and mature let-7d miRNA, as analyzed by Northern Blot. The data have been repeated once. (d). Deletion of Raptor increases miRNA in Lin^- c-Kit⁺ bone marrow cells. Lin^-c-Kit⁺ HSPCs were isolated from Rptr^f/f;Mx1-Cre⁺ (Rptr^−/−) or Rptr^f/f;Mx1-Cre^- (Rptr^+/+) mice at 10 days after the pIpC treatment. pre-miRNA and miRNA levels were measured by miRNA microarray. (e). Tsc1 deletion broadly reduces miRNAs in HSPCs. FACS-sorted LSK from BM of Tsc1^f/f;Mx1-Cre⁺ (Tsc1^−/−) or Tsc1^f/f;Mx1-Cre^- (Tsc1^+/+) mice at 10 days after the pIpC treatment. In (d) and (e), data shown are ratio of individual precursor (right panel) or mature (left panel) miRNA (Y-axis) vs. logP value of the differences. (n=5).

Fig. 2

Tsc1 deletion causes a specific block in pri- to pre miRNA processing by accelerating Drosha degradation. (a). RT-qPCR analyses of five candidates with reduction in both pre- and mature miRNA levels. The data shown are means and SEM of triplicate data from one experiment. These data have been repeated independently 3 times. (b). Defective processing of pri-miRNA in Tsc1^−/− MEF revealed by decreased ratio of relative pre-miRNA over that of pri-miRNA among those that show no reduction in either pre- or mature pre-miRNA. The ratios are calculated by the following formulas: Y= Pre(Tsc1^−/− /Tsc1^+/+)/ Pri(Tsc1^−/− /Tsc1^+/+). The data shown are means and SEM of triplicate data from one experiment. These data have been repeated independently three times. (c) mTOR-dependent reduction of Drosha. Tsc1^−/− (KO) or WT MEFs were treated with rapamycin (R, 5nM) or vehicle (V) for 4 days and lysed for immunoblotting analysis. Data have been reproduced twice. (d). mTOR activation does not reduce transcripts for miRNA processing proteins. Drosha, Dicer, Exportin5, and Ago mRNA levels were measured by RT-PCR. Hprt was used as an endogenous control. Data shown are means and S.D. of triplicate samples from one experiment. The same trends were observed in 3 independent experiments. (e). Immunoblotting of Drosha, Dicer, Exportin5, Ago2, and pS6 (phosphorylated S6 protein, which is used as a marker for mTOR activation) in WT or Tsc1-/- MEF cells. Actin levels were used as a loading control. These data have been repeated once. (f). mTOR activation causes a selective reduction of Drosha in Tsc1^−/− or WT splenocytes and thymocytes isolated from the Tsc1^flox/flox; Mx-Cre mice following polyI:C treatment. Drosha, Dicer, and Ago2 were detected by immunoblotting. (g, h). Proteasome-mediated degradation of Drosha in Tsc1^−/− MEFs. MEFs were incubated in media containing vehicle, cycloheximide (CHX, 100μg/mL), or CHX combined with MG132 (20μM) for indicated times. Drosha levels were analyzed by immunoblotting and quantitated by Li-COR Odyssey Infrared Imaging system. (g). Representative photograph. (h). Quantitative data from an experiment with triplicate samples are presented. The same trend has been reproduced in three independent experiments.

Fig. 3

Mdm2 is regulated by mTOR and regulates Drosha degradation. (a). Tsc1 deletion causes increased expression of Mdm2 in MEFs. (b). Tsc1 deletion increases Mdm2 mRNA by a p53-dependent mechanism. Mdm2 transcripts inTp53^+/+Tsc1^+/+, Tp53^+/+Tsc1^−/−, Tp53^−/−Tsc2^+/+, Tp53^−/−Tsc2^−/− were measured by real-time PCR, using L32 as an internal control. All data were normalized against the mean levels of Tsc1^+/+ cells. Data shown are means and SEM of triplicate samples from one experiment. The same trend has been observed in three independent experiments. (c). Inactivation of the Tsc complex up-regulates Mdm2 and decreases Drosha in MEFs by a Tp53-independent mechanism. Lysates from WT or Tp53^−/− MEFs with or without Tsc2 deletion and/or Mdm2 shRNA were probed for levels of pS6, Drosha, and Mdm2 by immunoblotting. Actin levels were used as a loading control. (d). Mdm2 deletion increases Drosha levels by a Tp53-independent mechanism. Lysates from Tp53^−/− MEFs with or without Mdm2 deletion were probed for Drosha levels by immunoblotting. Actin levels were used as loading control. (e). ShRNA silencing of MDM2 by two independent shRNA clones increases DROSHA levels in 293T cells. Data from lentiviral-mediated gene silencing by two independent shRNA sequences are presented. (f). MDM2 silencing by shRNA HS-5 increases the half-life of DROSHA, as analyzed by immunoblotting and quantitated by Li-COR Odyssey Infrared Imaging system. A representative photograph is presented. As indicated by ACTIN levels, we reduced the amount of total lysates in MDM2-silenced cells to achieve similar initial DROSHA levels for better comparison. The normalized DROSHA levels are presented underneath the ACTIN blot (D/A means Drosha/Actin). (g). Stabilization of DROSHA by MDM2 ShRNA. Quantitative data on the amounts of DROSHA at different time after CHX treatment. The data from an experiment with triplicate samples are shown. These experiments have been repeated at least two times. * P < 0.05, * * P < 0. 01.

Fig. 4

MDM2 is the necessary and sufficient E3 ligase for DROSHA. (a). Co-immunoprecipitation of MDM2 with Myc-DROSHA. 293T cells were transiently transfected with pcDNA4/TO/MycDrosha together or separately with pCMV-MDM2. Twenty hours after transfection, cells were treated with MG132 (10 μM) for 8 hours. Lysates of transfected 293T cells were subjected to immunoprecipitation with an anti-Myc antibody and immunoblotting with anti-MDM2 and anti-Myc antibodies. (b). Co-precipitation of endogenous DROSHA and MDM2. 293T cells expressing shRNA against MDM2, DROSHA, or Scramble control were treated with MG132 (10 μM) for 8 hours. Lysates were crosslinked with dithiobis[succinimidyl propionate] (DSP) before immunoprecipitating with an anti-MDM2 antibody. The immunoprecipitates were analyzed by immunoblotting with anti-DROSHA and anti-MDM2 antibodies. MDM2 or DROSHA knockdown (KD) cells were used as negative control. (c). MDM2 is the E3 ligase for DROSHA. In vitro ubiquitinylation assay was performed with Myc-DROSHA immunoprecipitated from Myc-DROSHA transfected Cos7 cells. The immunoprecipitates were incubated with a cocktail containing MDM2 and biotinylated ubiquitin (biotin-Ub) as well as E1 and E2 for 60 min at 37°C. The ubiquitinylated DROSHA was analyzed by immunoblot with streptavidin-HRP. Anti-Myc antibody was used to detect immunoprecipitated Myc-DROSHA. (d). MDM2 causes ubiquitinylation of DROSHA. Myc-tagged DROSHA, MDM2 or MDM2C464A (catalytic mutant), and HA-tagged Ub were co-transfected into Cos7 cells. After 24 hours, cells were harvested. Drosha was immunoprecipitated with an anti-Myc antibody. Then, immunoprecipitates were analyzed by immunoblot with anti-HA antibody first and then reblotted with an anti-Myc antibody. MDM2, DROSHA and HA-Ub protein expression were analyzed in a portion of lysate before immunoprecipitation. (e). ShRNA silencing of MDM2 decreases DROSHA ubiquitinylation, ShRNA clone HS-5 was used. The upper panel shows the isolation and detection of ubiquitinylated DROSHA using His-tag based purification and elution, while the lower panel shows the efficacy of MDM2 silencing. (f). The ubiquitinylation of endogenous DROSHA is also reduced by MDM2 silencing. As in (e), except that no exogenous DROSHA cDNA were transfected so that only endogenous DROSHA was measured. All data have been repeated at least twice.

Fig. 5

The mTOR-MDM2-DROSHA pathway in response to nutrient and energy deprivation. (a). Amino acid deprivation increases DROSHA levels. 293T cells were starved with amino acid-free media for 24 hours, and then cultured in the presence or absence of an amino acid solution for 4 hours. The levels of pS6, MDM2, and DROSHA were determined by immunoblotting. (b). In the absence of p53, amino acids cause a reduction in Drosha by mTOR and Mdm2-dependent mechanisms. Tp53^−/− or Tp53^−/−Mdm2^−/− MEFs were cultured with either amino acid-free media or media supplemented with amino acids for 2 hours in the presence of either vehicle or rapamycin (1 μg /ml). Drosha levels were determined by immunoblotting. (c). mTOR activation through TSC1 gene silencing induces MDM2 while reducing DROSHA in A549 cell lines. The cell lysates were subjected to immunoblotting with indicated antibodies. Phospho-S6 (pS6) was used to monitor mTOR activity. Actin was used as a loading control. (d). Glucose-deprivation increases DROSHA levels. The human lung cancer cell line A549 was cultured in glucose-containing (G+) or glucose-free (G-) media for 8 hours. The levels of pS6 and DROSHA were determined byimmunoblotting. (e). Mdm2 is up-regulated and rapidly decayed after mTOR activation. A549 cells were cultured in glucose-containing (G+) or glucose-free (G-) media together with either vehicle or MG132 (10 μM) for 8 hours. The level of MDM2 was determined by immunoblotting. The data in this figure have been reproduced 2–3 times.

Fig. 6

A critical role for DROSHA in cell survival under conditions of glucose deprivation. (a-c) DROSHA knockdown sensitizes A549 cells to glucose starvation-induced cell death. Immunoblottingin (a) shows the efficiency of DROSHA knockdown in A549 cells. Photographs in (b) are bright field images of Scramble or DROSHA shRNA -transduced A549 cells after overnight culture in the presence or absence of glucose; bar graph in (c) depicts means and S.D. of % of Annexin V⁺ cells in triplicate culture. **P<0.001 when % apoptosis in Scr and shRNA under glucose starvation conditions were compared. (d). DICER knockdown in A549 cells shows similar effect as DROSHA knockdown. Western blot in (insert) shows the efficiency of DICER knockdown in A549 cells. Bar graph depicts means and S.D. of % of Annexin V+ cells in triplicate culture. ***P<0.001 when % apoptosis in Scr and shRNA under glucose containing or starvation conditions were compared. (e, f). Rapamycin (100nM) rescues TSC1 knockdown cells (e) but not DROSHA knockdown cells (f) from starvation-induced cell death. Bar graph depicts means and S.D. of % of Annexin V⁺ cells in triplicate culture. N.S. p>0.05, ***P<0.001 when % apoptosis in vehicle and rapamycin groups under glucose starvation conditions were compared. The data in this figure have been reproduced 2–3 times.

Fig. 7

miRNAs that protect host from apoptosis under glucose deprivation is regulated by glucose, Tsc1, Mdm2, and Drosha. (a). Diagram of high throughput screening of an miRNA mimic library for miRNA that rescue Drosha-deficient cells upon glucose deprivation. miRNA mimics were transfected individually into A549 cells. After 24 hours of culture glucose sufficient medium, the transfected cells were maintained in glucose-free media for 24 hours and inspected by microscopy to identify those miRNAs that can rescue Drosha-deficient cells. (b) Bright field image of Mock- and mi-R376b-3p transduced Drosha-silenced A549 cells. (c). Top panel shows DROSHA protein level in mimic transfected A549 cells. The normalized DROSHA levels from two different experiments are presented underneath the ACTIN blot (D/A, Drosha/Actin). All data were normalized against the levels of DROSHA in A549 Scramble control. Bottom panel shows scoring and validating the miRNA mimics that rescue A549 from apoptosis. The bar graph shows means and SEM of Annexin V⁺ cells. **P<0. 01 when % apoptosis in Scrambled (NC) and the miR mimics under glucose starvation conditions were compared. (d). Confirmation of endogenous miRNA in protecting A549 cells from energy deprivation-induced cell death. A549 cells were transfected with either individual or pooled anti-miR specific for miRNA that rescued Drosha-deficient cells from energy deprivation-induced apoptosis. *P<0.05, **P<0.01 when % apoptosis in Scr (NC) and anti-miR under glucose starvation conditions were compared. The transfectants were cultured in glucose-free medium overnight and analyzed for apoptosis by flow cytometry. The bar graph shows means and SEM of Annexin V⁺ cells of triplicate samples from one experiment. (e). miRNAs that rescued glucose deprivation-induced death are regulated by glucose, TSC, MDM2, and DROSHA. Taqman probes designed for 3 miRNAs were used to compare their levels after the A549 cells were cultured for 8 hours in glucose-free or –sufficient media. To test their regulation by TSC, DROSHA, or MDM2, we compared A549 cells transduced with either scrambled or shRNA specific for the TSC1, DROSHA, or MDM2 genes. Left panel, relative miRNA levels were compared between cells cultured in glucose sufficient (G+, defined as 1.0) and glucose-free (G-) media. *P<0.05, **P<0.01. Right three panels, comparisons were made between Scr (Ctrl, defined as 1.0) and shRNAs-transduced cells that were cultured in glucose-sufficient medium. The data in this figure have been reproduced at least 2–3 times.