Tuberous Sclerosis Complex Proteins 1 and 2 Control Serum-Dependent Translation in a TOP-Dependent and -Independent Manner^{▿ †}

Cell culture, constructs, and drug treatments.

Immortalized littermate pairs of WT and null Tsc1 (Tsc1^−/−p53^+/+ and Tsc1^+/+p53^+/+) and Tsc2 (Tsc2^−/−p53^−/− and Tsc2^+/+p53^−/−) MEFs were generous gifts from D. Kwiatkowski (Brigham and Women's Hospital, Boston, MA) and were described previously (38, 78). Tsc2^+/+p53^−/− cells were used throughout the present study as WT MEFs, Tsc2^−/− TP53^−/− cells were used as Tsc2^−/− MEFs, and Tsc1^−/−p53^+/+ cells were used as Tsc1^−/− MEFs. Cells were grown in a humidifier incubator with 8% CO₂ at 37°C in Dulbecco modified Eagle medium (DMEM) containing 4.5g of glucose (Life Technologies)/liter supplemented with 10% fetal bovine serum. Wortmannin and rapamycin (Calbiochem) were added to the appropriate media at the indicated times at 100 and 50 nM, respectively. For serum and/or nutrient starvation experiments, cells were seeded and maintained in DMEM plus 10% fetal calf serum (FCS). The following day, the cells were washed twice in Dulbecco's phosphate buffered saline (D-PBS) and maintained for the indicated time in DMEM alone, D-PBS alone, or D-PBS plus 10% dialyzed FCS. The human TSC1 full-length cDNA was subcloned into LXSP3 (Puro) retroviral vector. Retroviruses were generated by transient transfection of Phoenix packaging cells. Filtered supernatants containing virus were used to infect Tsc1^−/− MEFs, followed by selection with 3 μg of puromycin/ml. We refer to Tsc1^−/− MEFs expressing empty vector or WT TSC1 as Tsc1^−/− (vector) or Tsc1^−/− (Rev) MEFs, respectively. Tsc2^−/− (vector) and Tsc2^−/− (Rev) MEFs were provided by D. Kwiatkowski and were previously described (78).

Polysomal analysis and RNA preparation.

The 4-h time point was selected for the microarray experiments. Cell extracts were prepared essentially as described previously (70). Briefly, after incubation with cycloheximide (CHX) at 90 μg/ml for 10 min, 2 × 10⁶ cells were washed and treated with trypsin in presence of CHX (90 μg/ml). Cell pellets were washed twice in ice-cold PBS containing 90 μg of CHX/ml, resuspended in 150 μl of cold RSB (10 mM Tris-HCl [pH 7.4], 10 mM NaCl, 15 mM MgCl₂) containing 100 μg of heparin/ml, and lysed in ice-cold lysis buffer (1.2% Triton X-100 and1.2% deoxycholate in RSB) on ice. Nuclei and cell debris were cleared out by centrifugation at 12,000 × g for 5 min in a microcentrifuge at 4°C. The supernatant was diluted with an equal volume of polysomal buffer (25 mM Tris-HCl [pH 7.4], 25 mM NaCl, 25 mM MgCl₂, 0.05% Triton X-100, 0.14 M sucrose, 500 μg of heparin/ml) and layered over 12 ml of a 5 to 56% (wt/wt) sucrose gradient. The gradients were sedimented via centrifugation at 37,000 × g for 150 min at 4°C in a SW40 Ti rotor (Beckman). Twelve fractions of 1 ml each were collected, and RNAs were precipitated by a standard procedure, purified by using an RNeasy minikit (QIAGEN), and quantified by determining the absorbance at 260 nm. For microarrays, fractions 5 to 7 and 9 to 10 from three independent experiments were pooled for polysomal and subpolysomal RNA samples, respectively, and were reprecipitated with LiCl buffer (2 M LiCl, 20 mM Tris-HCl [pH 8.0]). RNA quality was monitored by using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Total RNAs, pooled from three independent experiments, were isolated according to the manufacturer instructions (Agilent Technologies). For Northern blot analysis, RNA was fractionated on formaldehyde-agarose gels and transferred to nylon Hybond N membrane (Amersham Biosciences). Northern blotting was performed essentially as recommended by the manufacturer. The TOP mRNA probe was mouse eEF1A. The non-TOP mRNA analyzed was mouse GAPDH (glyceraldehyde-3-phosphate dehydrogenase). Samples were prepared by using products of reverse transcription-PCR (RT-PCR). Primers were designed according to sequences present in the GenBank/EBI data bank.

Immunoblotting.

Cells were washed twice in cold PBS and harvested in protein lysis buffer containing 1% NP-40, 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 1 mM dithiothreitol, phosphatase cocktail inhibitors (Sigma-Aldrich), and protease inhibitor cocktail (Boehringer Mannheim) (1 pill/10 ml of lysis buffer). Cell lysates were prepared as previously described (70). Equal amounts of total protein (30 μg) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 4 to 20% gradient polyacrylamide gels and transferred to polyvinylidene difluoride membranes. Immunostaining was performed with antibodies generated against phospho-Thr389 S6K, phospho-Ser235/236 S6, phospho-Ser240/244 S6, phospho-Ser422 eIF4B, phospho-Ser65 4E-BP1, phospho-Thr37/46 4E-BP1, phospho-Ser51 eIF2α, S6, and eIF4E (Cell Signaling, Beverley, MA); phospho-Ser246 PRAS40 (Biomol International, Plymouth, PA); eIF4G1 (Abcam, Inc., Cambridge, MA); and beta-actin (Sigma-Aldrich, Saint Louis, MO).

Microarray analysis.

Microarray experiments were performed by using a mouse 22K (G4121A) oligonucleotide array (Agilent Technologies) according to the manufacturer's instructions. Briefly, polysomal, subpolysomal, or total RNA samples were tested on Agilent's Bioanalyzer 2100 for quality, amplified, labeled with cyanine 3 and cyanine 5 fluorescent dyes (2 μg of RNA per labeling reaction), and hybridized onto Agilent's 22K mouse oligonucleotide arrays (1 μg of labeled cRNA per channel for hybridization) using Agilent's reagents and protocols. Microarray data were analyzed with GeneSpring 7.2 software (Silicon Genetics) for the calculation of intensity ratios and dye-swap calculations as normalization methods. Features that showed large variability were filtered out by only including the features that had a standard variation of <0.7 between dye swap experiments across all experimental conditions. This left a subset of 7,055 genes in the data set. The normalized datasets were processed by calculating the simple log₂ intensity ratios comparing polysome to subpolysomal values across control and starved samples. A twofold cutoff was chosen to identify polysome-associated RNAs.

Quantitative PCR analysis.

Real-time PCR was performed in ABI 7500 (Applied Biosystems) as previously described (70). RNA samples from polysome and subpolysome fractions were normalized to GAPDH (the polysome/subpolysome ratio of GAPDH did not vary between the control and starved or treated samples).

Either growth factors or nutrients are sufficient for mTORC1 activity in Tsc1^−/− and Tsc2^−/− MEFs.

The TSC/mTORC1 pathway is important for the regulation of mRNA translation and integrates various signals such as nutrients, growth factors, energy, and stress to regulate cell growth and cell proliferation. To assess the role of TSC1 and TSC2 in mTORC1 signaling, we compared the effects of serum starvation, nutrient starvation, or serum and nutrient starvation on mTORC1 activity in WT, Tsc1^−/−, and Tsc2^−/− MEFs. Throughout the present study, we define mTORC1 activity through both its ability to phosphorylate its defined substrates (S6K [Thr-389] and 4EBP1 [mobility shift and Ser-65]) and its ability to be inhibited by rapamycin. In addition, we used the phosphorylation of RP S6 (at Ser235/236 and Ser240/244) as an additional measure for the activity of S6K. As shown in Fig. 1, under normal growth conditions (DMEM plus 10% FCS) WT MEFs displayed strong mTORC1 activity that is sensitive to rapamycin. However, depletion of either growth factors (FCS) or nutrients (DMEM) in these cells rapidly decreased the mTORC1 signaling. Withdrawal of either growth factors or nutrients also reduced the phosphorylation of 4E-BP1, as indicated by the appearance of faster mobility forms. In contrast, Tsc1^−/− and Tsc2^−/− MEFs maintained elevated mTORC1 activity upon 4 h growth factor or nutrient depletion (Fig. 1). In the absence of nutrients but in the presence of dialyzed FCS (D-PBS+dFCS) in Tsc1^−/− or Tsc2^−/− MEFs, S6K and 4E-BP1 phosphorylation is reduced but with delayed kinetics compared to WT cells. In contrast, the withdrawal of both growth factors and nutrients (D-PBS) dramatically reduced the mTORC1/S6K/4E-BP1 signaling in these cells. Elevated S6K and S6 phosphorylation and 4E-BP1 mobility shift in Tsc1- and Tsc2-null cells was always inhibited by rapamycin, thus confirming a dependence on mTORC1. However, wortmannin was only effective at inhibiting mTORC1 activity induced by growth factors but not nutrients, confirming the importance of PI3K in mitogen-mediated but not nutrient-mediated signaling to mTORC1.

Serum depletion inhibits the assembly of translation initiation complexes.

mTORC1 integrates nutrient and growth factor signals to regulate the translation initiation of mRNAs important for cell growth (54). To determine whether mTORC1 inhibition mediated by serum and/or nutrient withdrawal was correlated with inhibition of translation initiation, we next examined the consequences on the maintenance of cap-dependent translation. mTORC1 regulates cap-dependent translation in part through the phosphorylation of 4E-BP1. Unphosphorylated 4E-BP1 inhibits eIF4E by preventing its binding to eIF4G and subsequent association with capped mRNAs. 4E-BP1 phosphorylation at Ser65 is correlated with the dissociation of 4E-BP1 from eIF4E (46, 75). Active and inactive translational complexes can be determined by measuring the levels of 4E-BP1 associated with eIF4E precipitated through its binding to the capped mRNA mimic methyl⁷-GTP Sepharose. Figure 2 shows that in the presence of growth factors and nutrients (DMEM+FCS) in both WT MEFs and Tsc1^−/− and Tsc2^−/− MEFs, low levels of 4E-BP1 were associated with methyl⁷-GTP Sepharose (Fig. 2, lanes 1, 13, and 25). This closely correlated with the levels of phosphorylation of 4E-BP1 detected with either specific antibodies (Ser65) or with total 4E-BP1 mobility. Deprivation of either growth factors (FCS) or nutrients (DMEM) resulted in increased levels of 4E-BP1 bound to eIF4E in WT MEFs but not in the Tsc-null cells (in Fig. 2, compare lanes 4 and 10 to lanes 16 and 22 and to lanes 28 and 34). However, the withdrawal of both nutrients and growth factors (No Nutr No FCS) strongly increased the amount of 4E-BP1 bound to eIF4E in WT, Tsc1^−/−, and Tsc2^−/− MEFs. Rapamycin strongly enhanced the association of hypophosphorylated 4E-BP1 to methyl⁷-GTP Sepharose under all of the conditions tested in all three MEFs, demonstrating the contribution of mTORC1 in the assembly of translation initiation complexes (Fig. 2, compare lanes 7, 19, and 31). Taken together, these data demonstrate that, in the absence of TSC1 or TSC2, either growth factors or nutrients are sufficient for mTORC1 activation and protein translation initiation.

Serum depletion reduces polysome formation in WT MEFs but not in Tsc1^−/− and Tsc2^−/− MEFs.

mRNA translation is a tightly controlled process taking place on ribosomes. Actively translated mRNAs are distributed to polyribosomes (or polysomes), and inactive mRNAs are associated with subpolysome and monosome fractions. We therefore investigated whether the inhibition of cap-dependent translation upon serum withdrawal had consequences on the distribution of RNA from polysomes to subpolysomes. To address this question, we isolated polysomal mRNA from total cellular mRNA by fractionation through a 5 to 56% (wt/wt) sucrose gradient. To confirm the correct identification of the polysomal peaks, we analyzed the polysomal localization of a well-characterized TOP mRNA, eEF1α, known to be redistributed from polysome complexes into subpolysomal fractions after serum starvation (7). eEF1α demonstrated robust translocation from polysomes to subpolysomes in response to serum deprivation (see Fig. S1 in the supplemental material), suggesting that the polysomal/subpolysomal fractions were accurately isolated in our experiment and therefore suitable for further analysis. We observed a redistribution of polysome-associated RNAs into subpolysomes in response to a 4-h serum starvation in WT MEFs (Fig. 3). However, in Tsc1^−/− and Tsc2^−/− MEFs, serum withdrawal did not significantly affect the polysome-subpolysome profile (Fig. 3). This redistribution of RNA from polysomes to subpolysomes in response to serum starvation in WT MEFs precisely correlated with the effects observed on cap-dependent translation shown in Fig. 2. Altogether, these results are consistent with the regulation of mTORC1 signaling and protein translation showing serum dependence in WT MEFs and serum independence in Tsc1^−/− and Tsc2^−/− MEFs.

Identification of specific mRNAs translationally regulated by FCS in WT, Tsc1^−/−, and Tsc2^−/− MEFs.

To identify specific mRNAs translationally regulated by serum in WT, Tsc1^−/−, and Tsc2^−/− MEFs, we subjected subpolysomal and polysomal RNAs to microarray analysis with Agilent 22k mouse 60-mer oligonucleotide arrays. The experimental approach is shown in Fig. 4A and is described in the figure legend. As shown in Fig. 4B, the withdrawal of 10% FCS in WT MEFs increased the number of RNAs showing a polysome/subpolysome ratio of <0.5, from 38 to 102 RNAs, whereas fewer mRNAs displayed this property in Tsc1^−/− and Tsc2^−/− MEFs. This correlates with the global polysomal profile and the inhibition of the mTORC1 signaling observed in Fig. 3. Figure S2 in the supplemental material illustrates how the nine polysome-associated mRNAs in WT MEFs in presence of DMEM+FCS shifted toward the subpolysome population upon FCS withdrawal and the 102 subpolysome-associated mRNAs of WT MEFs cultured in DMEM shifted toward the polysome population of WT MEFs cultured in DMEM plus FCS.

We next analyzed which mRNAs most dramatically shift between polysomes to subpolysomes upon serum withdrawal in WT, Tsc1^−/−, and Tsc2^−/− MEFs. To investigate this, we compared the ratio of every mRNA in polysome (P) and subpolysome (SP) fractions between control (DMEM+FCS) and serum-starved (DMEM) cells. The ratio of the Cy5 intensity and Cy3 intensity (representing the polysome/subpolysome ratio) reflects its relative polysome distribution. Therefore, mRNAs redistributing from polysomes to subpolysomes upon serum deprivation in WT MEFs that do not change under these conditions in Tsc1^−/− and Tsc2^−/− MEFs likely represent the RNAs translationally regulated by serum in a TSC1- and TSC2-dependent manner. We used an arbitrary cutoff value of twofold redistribution across polysomes under different conditions to designate translational regulation, so the numbers of mRNAs changing are not absolute, but different cutoff values showed similar trends. Figure 4C displays the change in polysome association of each mRNA when cultured in DMEM+FCS compared to DMEM alone. In WT MEFs, 175 mRNAs were preferentially associated with polysomes in the presence of serum, whereas only 40 and 42 mRNAs were more associated with polysomes in Tsc1^−/− and Tsc2^−/− MEFs, respectively, under the same conditions (Fig. 4C). Importantly, 160 of 175 mRNAs were translationally regulated by serum in WT MEFs but not in Tsc1^−/− and Tsc2^−/− MEFs (indicated by the white dots in Fig. 4C), demonstrating that ca. 90% of serum-dependent mRNA translation is mediated through the TSC1/TSC2 complex in MEFs (Fig. 4D).

Serum regulates ribosome biogenesis in a TSC1 and TSC2-dependent manner.

Table 1 lists the top 30 RNAs putatively regulated in a serum-sensitive fashion in a TSC1- and TSC2-dependent manner. Two predominant biological groups were significantly enriched (with P < = 4 × 10⁻¹¹) in the list of 175 serum-regulated genes by gene ontology analysis, namely, protein biosynthesis and ribosome biogenesis. Fifty different RPs and eleven translation initiation and elongation factors were in this group, suggesting that many of the serum-regulated RNAs are involved in ribosome biogenesis or encode components of the translational machinery (see Table S1 in the supplemental material). Some of the mRNAs that we identified have previously been shown to be regulated by serum, amino acids, and/or rapamycin, e.g., eEF1a, eEF1b, and eEF2 and RPs such as Rpl32 and Rps6 (7, 25, 31, 80). However, we also identified many additional regulated mRNAs, and other gene ontology categories were also represented, such as genes involved in metabolism, transport, transcription and development (Fig. 4E; also see Table S1 in the supplemental material).

To confirm our microarray data, we validated some of the genes by TaqMan analysis as represented in Fig. 5 (right panel). Approximately, 80% of the specific probes (of 19 tested) were validated by TaqMan, although three genes (Tnrc6b, Ddx6, and Rpl21) were not. In general, the fold shift from subpolysome to polysome in response to serum measured by TaqMan followed relatively closely the fold shift calculated by the microarray analysis (Fig. 5, left panel). The entire TaqMan results completed for each individual target are shown in Fig. S3 in the supplemental material.

The transcriptional response to serum withdrawal was also analyzed in WT, Tsc1^−/−, and Tsc2^−/− MEFs using the same Agilent 22k microarrays. Although 84 genes were found to be transcriptionally downregulated (>2-fold) upon serum depletion in WT MEFs, only 7 genes overlapped with our list of 175 mRNAs translationally regulated by serum (data not shown). We also analyzed the total mRNA levels of the 19 genes validated in the TaqMan analysis shown in Fig. 5. Overall, the transcriptional profile of these genes was not dramatically affected by serum deprivation, although two genes were downregulated >2-fold (c-Jun and Rpl36a), and one gene was upregulated >2-fold (p57^Kip1) (see Fig. S4 in the supplemental material).

Rapamycin inhibits mTORC1 signaling in WT, Tsc1^−/−, and Tsc2^−/− MEFs.

We next investigated whether inhibition of mTORC1 activity, using the mTORC1 specific inhibitor rapamycin, would affect the same mRNAs as regulated by serum (described above). We performed a similar analysis on WT, Tsc1^−/−, and Tsc2^−/− MEFs cultured in presence of 50 nM rapamycin for 4 h. As shown in Fig. 6A, rapamycin caused inactivation of mTORC1, as judged by dephosphorylation of S6 at Ser235/236 and Ser240/244 and by the appearance of faster-migrating species corresponding to the hypophosphorylated 4E-BP1.

Since mTORC1 signaling was maintained in Tsc1^−/− and Tsc2^−/− MEFs upon serum starvation and showed comparable activity to WT MEFs in the presence of serum, we compared the effects of rapamycin on polysomal profiles and microarray analysis under these same conditions to identify specific mRNAs translationally regulated by rapamycin. Rapamycin had a minor effect on polysome distribution in WT MEFs cultured in the presence of FCS (Fig. 6B; also see Fig. S5 in the supplemental material). However, rapamycin caused a more significant decrease in the amount of polysome-associated mRNAs in both Tsc1^−/− and Tsc2^−/− MEFs (fractions 5 to 7 in Fig. 6B). This decrease was also accompanied by a significant increase of the amount of subpolysome-associated RNAs (fractions 9 and 10 in Fig. 6B). We purified polysome-associated mRNAs and subpolysome-associated mRNAs from sucrose gradients and hybridized them to Agilent 22k mouse 60-mer oligonucleotide arrays using a dye reversal as a control (as described in the previous analysis). Figure 7A illustrates that 113 mRNAs showed a >2-fold association with polysomes in WT MEFs in DMEM+FCS compared to treatment with rapamycin. Although the majority of mRNAs repressed by rapamycin were also repressed after serum withdrawal (73 of 113), there were also many mRNAs regulated only by rapamycin but not by serum (40 of 113) (Fig. 7B). Conversely, there were also many mRNAs regulated by serum but not by rapamycin (102 mRNAs found) in WT MEFs, suggesting that these are regulated in an mTORC1-independent manner (Fig. 7B). A total of 86% (88 of 102) of these serum-sensitive and rapamycin-insensitive mRNAs in the WT MEFs are not serum regulated in the Tsc1- and Tsc2-null cells. For these mRNAs, the lack of polysomal redistribution in response to serum withdrawal could either be due to the fact that the mRNA is constitutively associated with polysomes or constitutively associated with subpolysomes. Thirty-three of these eighty-eight mRNAs are constitutively polysome-associated in the Tsc1^−/− and Tsc2^−/− MEFs in absence of serum (indicated as “H” in Table S1 in the supplemental material). In contrast, 85% (62 of 73) of the serum- and rapamycin-sensitive mRNAs are constitutively polysome associated in the Tsc1- and Tsc2-null cells in the absence of serum (indicated as “H” in Table S2 in the supplemental material). Of the 33 constitutively polysome-associated mRNAs in the Tsc1^−/− and Tsc2^−/− MEFs, only 7 were resistant to rapamycin in both WT and null MEFs (indicated by an asterisk in Table S1 in the supplemental material), suggesting that only a small number of mRNAs are translated in a TSC1- and TSC2-dependent but rapamycin-independent manner.

Rapamycin represses the translation of specific mRNAs.

To examine whether a specific translational signature would emerge across the different conditions, we performed hierarchical clustering on all genes regulated by serum and/or rapamycin in all three MEFs. The cluster algorithm separates the experimental conditions into two distinct groups that tightly correlated with the mTORC1 activity status of these samples (Fig. 7C). One cluster (cluster L) is represented by a set of mRNAs affected by low mTORC1 activity (see the figure legend for details); all rapamycin-treated samples, as well as serum-starved WT MEFs, are grouped in this category. The other cluster (cluster H) is represented by high mTORC1 activity (Fig. 7C). Both Tsc2^−/− and Tsc1^−/− MEFs displayed very closely related translational profiles when cultured either in the presence or in the absence of FCS, whereas WT MEFs grown in DMEM+FCS defined a distinct group within the high mTORC1 activity cluster. A similar cluster pattern was observed when all filtered mRNAs from the gene list present in the microarray were used to perform this analysis (data not shown).

In addition, microarray analysis of the Tsc1^−/− and Tsc2^−/− MEFs cultured in DMEM (in the absence of FCS) showed that 341 and 219 mRNAs, respectively, shift >2-fold to the subpolysome fractions in response to rapamycin treatment. The Venn diagram illustrated in Fig. 7D shows that most of the 113 mRNAs suppressed by rapamycin in WT MEFs were also regulated in Tsc1^−/− and Tsc2^−/− MEFs. Although many mRNAs were coregulated by rapamycin across the three different MEFs, there were also mRNAs that appear to be regulated uniquely in cells in one or two genotypes. Of the 76 mRNAs regulated by rapamycin in all three cell types, 41 encode for RPs, and 9 encode for translation initiation or elongation factors, therefore enriching the class of genes involved in ribosome biogenesis to 66% in this group (Fig. 7E). Tables S2 and S3 in the supplemental material display the identity of mRNAs translationally repressed by rapamycin in WT, Tsc1^−/−, and Tsc2^−/− MEFs. Moreover, we showed that there are more mRNAs that shift from polysome to subpolysome fractions in response to rapamycin in the serum-starved Tsc-null MEFs than in the serum-fed WT MEFs (Fig. 6B and 7D). Therefore, we believe that the overlap between the Tsc1^−/− and Tsc2^−/− MEFs in this group (Fig. 7D) represents a group of high confidence mTORC1-specific translational targets since these are not complicated by serum effects. The identities of the additional 82 genes are provided in Table S4 in the supplemental material.

We confirmed our microarray data using TaqMan analysis as shown in Fig. 8A; see also Fig. S6 in the supplemental material. We found that the correlation between the two analyses was generally consistent except for a few mRNAs such as Rpl4, Arid5b, and Ddx6. Although the regulation by rapamycin measured by TaqMan was often weaker than the regulation observed by microarray, the general trend was similar, with some mRNAs being regulated only by serum and not by rapamycin, some being regulated by rapamycin and not serum, and some being regulated by both.

Since we have identified a subset of mRNAs that were redistributed from the polysome to the subpolysome upon serum starvation or upon rapamycin treatment, we assessed whether such translational regulation would be detectable at the protein level. To address this, we examined protein expression by SDS-PAGE with specific antibodies based on their availability. Four proteins were tested and analyzed by Western blotting. To minimize effects of protein stability confounding our analyses, we starved cells of serum overnight and restimulated them with 10% FCS in the absence or presence of rapamycin at the indicated times (Fig. 8B; see also Fig. S7 in the supplemental material). Consistent with the microarray results, steady-state levels of eIF4B, p57^Kip2, c-Jun, and Tpt1 were all increased by FCS in WT MEFs. eIF4B, p57^Kip2, and Tpt1 expression required 4 h of serum stimulation to show increases in protein levels, and this increase was prevented by rapamycin. These proteins do not show increased levels in response to serum in Tsc1^−/− and Tsc2^−/− MEFs, but their expression is still decreased in response to rapamycin treatment, a finding consistent with the polysome analysis. Longer treatment (8 to 24 h) with rapamycin is required to visualize decreased protein levels, possibly due to the long half-lives of these proteins. c-Jun was regulated differently in that its expression was increased earlier after serum stimulation in WT, Tsc1^−/−, and Tsc2^−/− MEFs and was not inhibited by rapamycin in WT MEFs. c-Jun mRNA was also transcriptionally regulated by serum in the MEFs of all three genotypes (see Fig. S4 in the supplemental material), indicating that transcriptional regulation appears to be dominant over translational regulation for this protein. Conversely, p57^kip2 mRNA levels were actually increased upon serum withdrawal, and yet the p57^kip2 protein levels increased upon serum stimulation in a rapamycin-dependent manner, showing that in this case translation is dominant over transcription.

Re-expression of TSC1 or TSC2 in Tsc-deficient cells restores mTORC1 regulation by serum and the translation of specific mRNAs.

Because we used nonisogenic Tsc-deficient cells in our study, we were concerned that the effects we observed were independent of their genetic status. To address this issue, we reintroduced WT TSC1 into the Tsc1^−/− MEFs and WT TSC2 into the Tsc2^−/− MEFs using retroviral expression. TSC1 deficiency reduces the steady-state levels of TSC2 (10, 78), and this was rescued upon re-expression of TSC1 (Fig. 9A). Restoration of TSC1 and TSC2 expression into TSC-deficient cells also resensitized mTORC1 activity to serum withdrawal, suggesting that most of the serum effects signal through TSC1/TSC2 in our model system. Upstream signaling to mTORC1 was barely affected in the reconstituted cells, as shown by the similar effect of serum depletion on the phosphorylation of PRAS40, a PKB target known to be regulated by growth factors (37, 60). Consistent with the restoration of mTORC1 regulation by serum, the global polysomal profile of these revertant cells displayed redistribution from polysomes to subpolysomes upon serum depletion to an extent similar to that of WT MEFs (Fig. 9B). We also performed TaqMan analysis on the mRNAs previously identified as being TSC dependent in the Tsc (Vector) and Tsc (Rev) cells (Fig. 9C). Most of the mRNAs tested displayed a significant redistribution from polysomes to subpolysomes upon serum starvation in revertant cells (Rev) compared to the control cells (Vector), suggesting that these mRNAs are regulated by serum in a TSC1/TSC2-dependent manner.

Serum regulates ribosome biogenesis and a subset of TOP mRNAs in a TSC1- and TSC2-dependent manner.

Translation efficiency is determined at the posttranscriptional level, and this regulation is mostly mediated by cis-acting elements located in the 5′- and 3′UTRs (52). Table 1 shows that serum starvation in WT MEFs strongly affected mRNAs encoding RPs, suggesting that a common motif in their 5′UTRs may be functionally relevant. A common feature of RNAs encoding components of the translational machinery is the presence in their 5′UTRs of a TOP motif. This motif is currently known to be associated with mTORC1-dependent translation regulation (25). However, there are mRNAs regulated by mTORC1 that do not contain TOP motifs (48, 76), and also mRNAs containing this motif that are not regulated by mTORC1 (17). Therefore, to examine whether the 175 serum-regulated mRNAs contained a TOP motif or a specific functional sequence within their 5′UTRs, we retrieved from current databases all 5′UTR sequences, when available, of these mRNAs. The structural feature of 5′TOP motifs is defined as initiating with a C residue followed by a stretch of 4 to 14 uninterrupted pyrimidine residues. In our study, although 33% of the serum-translationally regulated mRNAs were TOP containing mRNAs, the majority do not contain this motif, suggesting that motifs other than TOP are important for translational regulation of these mRNAs (Fig. 10A, top left pie chart). In comparison, the frequency of TOP motifs in mRNAs that are not regulated by mTORC1 was 10% (data not shown). Rapamycin-sensitive mRNAs consisted of 47% TOP mRNAs, but the highest proportion of 5′TOP mRNAs (55%) were those regulated by both serum and rapamycin (Fig. 10A, bottom middle pie chart). Among the 102 mRNAs regulated only by serum and not by rapamycin (Fig. 10A, bottom left pie chart), only 18% were TOP mRNAs, and the mRNAs regulated by rapamycin only and not by serum (40 mRNAs) showed a slight enrichment for TOP motifs (33%). Altogether, this suggests that 5′TOP motifs are greatly enriched in the group of mRNAs that are regulated by both growth factors and mTORC1. However, it also suggests that non-TOP containing mRNAs are regulated by TSC1 and TSC2 but in a mTORC1-independent manner. Interestingly, we also found that the mRNAs regulated by both serum and rapamycin contain a short 5′UTR and 3′UTR (medians of 41 and 64 nucleotides, respectively), whereas mRNAs regulated by either serum alone or rapamycin alone have a longer 5′UTR (medians of 119 and 100, respectively) and 3′UTR (medians of 477 and 531, respectively) (Fig. 10B). These 5′ and 3′UTR lengths are similar to those of 160 randomly chosen nonregulated mRNAs (98 and 755 nucleotides, respectively). Altogether, this suggests that this is relevant for the function of these mRNAs and might be implicated in growth-associated translational regulation.