Biochemical Journal

Research article

Stable isotope-labelling analysis of the impact of inhibition of the mammalian target of rapamycin on protein synthesis

Yilin Huo, Valentina Iadevaia, Zhong Yao, Isabelle Kelly, Sabina Cosulich, Sylvie Guichard, Leonard J. Foster, Christopher G. Proud

Abstract

mTORC1 [mTOR (mammalian target of rapamycin) complex 1] regulates diverse cell functions. mTORC1 controls the phosphorylation of several proteins involved in mRNA translation and the translation of specific mRNAs, including those containing a 5′-TOP (5′-terminal oligopyrimidine). To date, most of the proteins encoded by known 5′-TOP mRNAs are proteins involved in mRNA translation, such as ribosomal proteins and elongation factors. Rapamycin inhibits some mTORC1 functions, whereas mTOR-KIs (mTOR kinase inhibitors) interfere with all of them. mTOR-KIs inhibit overall protein synthesis more strongly than rapamycin. To study the effects of rapamycin or mTOR-KIs on synthesis of specific proteins, we applied pSILAC [pulsed SILAC (stable isotope-labelling with amino acids in cell culture)]. Our results reveal, first, that mTOR-KIs and rapamycin differentially affect the synthesis of many proteins. Secondly, mTOR-KIs inhibit the synthesis of proteins encoded by 5′-TOP mRNAs much more strongly than rapamycin does, revealing that these mRNAs are controlled by rapamycin-insensitive outputs from mTOR. Thirdly, the synthesis of certain other proteins shows a similar pattern of inhibition. Some of them appear to be encoded by ‘novel’ 5′-TOP mRNAs; they include proteins which, like known 5′-TOP mRNA-encoded proteins, are involved in protein synthesis, whereas others are enzymes involved in intermediary or anabolic metabolism. These results indicate that mTOR signalling may promote diverse biosynthetic processes through the translational up-regulation of specific mRNAs. Lastly, a SILAC-based approach revealed that, although rapamycin and mTOR-KIs have little effect on general protein stability, they stabilize proteins encoded by 5′-TOP mRNAs.

  • AZD8055
  • eukaryotic initiation factor (eIF)
  • mammalian target of rapamycin (mTOR)
  • mRNA translation
  • protein degradation
  • stable isotope-labelling with amino acids in cell culture (SILAC)

INTRODUCTION

mTOR (mammalian target of rapamycin) plays key roles in cellular regulation, controlling protein synthesis and promoting cell growth, as well as ribosome biogenesis, gene transcription and autophagy [1]. Protein accumulation is the balance of protein synthesis and breakdown. In the present study, we have used recently developed stable isotope-labelling approaches to study the influence of mTOR signalling on both of these processes.

mTOR is a protein kinase which forms two distinct types of multiprotein complex. mTORC (mTOR complex) 1 contains raptor (regulatory associated protein of mTOR), which recruits substrates for phosphorylation by mTOR. Substrates for mTORC1 include the ribosomal protein kinases S6K (S6 kinase) 1/2 and the 4E-BPs [eIF (eukaryotic initiation factor) 4E-binding proteins] [2]. The latter bind to eIF4E, block its interaction with eIF4G, a scaffold protein, and thus prevent it from engaging in translation initiation complexes. mTORC2 contains rictor (rapamycin-insensitive companion of mTOR), not raptor, and phosphorylates a distinct set of proteins including protein kinase B, which plays a key role in oncogenic signalling downstream of the PI3K (phosphoinositide 3-kinase) pathway [3] and is activated in many cancers.

Rapamycin inhibits mTORC1 by binding, together with the immunophilin FKBP12, to mTOR at a site adjacent to its kinase domain. However, rapamycin does not inhibit all of the functions of mTORC1, as indicated by results obtained with mTOR-KIs (mTOR kinase inhibitors) [4,5]. For example, phosphorylation of several sites in 4E-BP1 is insensitive to rapamycin, but is inhibited by mTOR-KIs, consistent with evidence that mTORC1 mediates their phosphorylation in a rapamycin-insensitive manner [6]. This and other factors led to the development of active-site-directed inhibitors of mTOR, which compete with ATP for binding to mTOR's kinase domain, including PP242 [5], Torin1 [4], Ku-0063794 [7] and AZD8055 [8]. The mode of action of these mTOR-KIs thus differs from that of rapamycin, which does not directly inhibit mTOR kinase activity.

In addition to activating S6Ks and relieving the inhibition of eIF4E by 4E-BPs, mTORC1 also promotes translation elongation [9] and the translation of a set of mRNAs characterized by 5′-TOPs (5′-terminal oligopyrimidines). They include those encoding all the cytoplasmic ribosomal proteins and several translation factors [10]. Control of the synthesis of these proteins is especially relevant to understanding cell growth control; first, the Rps (ribosome proteins) are abundant and therefore constitute a substantial proportion of cellular protein content; secondly, their levels determine the cellular capacity for protein synthesis and thus govern rates of cell growth and proliferation. Serum stimulation of starved cells increases the translation of 5′-TOP mRNAs, which shift into larger polysomes; this effect is partially inhibited by rapamycin [1113]. However, the effects of rapamycin on 5′-TOP mRNA translation are modest, suggesting they may be regulated independently of mTORC1 [14]. The PI3K inhibitor LY294002 inhibits 5′-TOP mRNA translation very strongly [15] (more so than rapamycin), suggesting a role for signalling through this pathway in their regulation. Nonetheless, the molecular mechanisms controlling 5′-TOP mRNA translation remain unclear [16].

As mentioned above, we have applied a recently developed approach, whereby newly synthesized proteins are labelled with isotopologues of arginine and lysine, to study the effects of rapamycin and mTOR-KIs on the synthesis of specific proteins. This is termed pSILAC [pulsed SILAC (stable isotope-labelling with amino acids in cell culture)]. It allows the identities and rates of accumulation of specific proteins to be determined, and directly provides quantitative information about the effects of signalling inhibitors on the production of specific proteins. We have also used a SILAC-based approach to study the effects of mTOR signalling on protein turnover.

The results of the present study reveal that rapamycin and mTOR-KIs have distinct effects on protein synthesis: the latter inhibit general protein synthesis and the synthesis of many specific proteins much more strongly than rapamycin does. In particular, although rapamycin inhibited the synthesis of proteins encoded by 5′-TOP mRNAs, mTOR-KIs did so much more strongly, revealing a substantial rapamycin-insensitive component to their control by mTOR. mTOR-KIs also strongly impaired the accumulation of numerous other proteins, some of which may also be encoded by mRNAs containing a 5′-TOP. Several of these are involved in anabolic processes, providing mechanisms by which mTOR signalling can enhance biosynthesis.

Our findings also show that, although mTOR inhibition has much less effect on general protein breakdown than on protein synthesis, it does affect the turnover of certain proteins. In particular, it decreases the breakdown of ribosomal proteins.

EXPERIMENTAL

Cell culture and treatment

HeLa cells were grown in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) FBS (fetal bovine serum), glutamine and 1% penicillin/streptomycin. Inhibitors (AZD8055, PP242 or rapamycin) were added to the cells for the indicated times. For insulin treatment, cells were starved in medium without FBS overnight, prior to addition of inhibitors and, 45 min later, insulin. AZD8055 was kindly provided by AstraZeneca.

Vectors and transfections

The vector for HA (haemagglutinin)-tagged human eIF4E was described previously [17]. Lipofectamine™ 2000 (Invitrogen) was used for transient transfections. The medium was changed after 4–6 h and cells were then maintained in medium containing 10% (v/v) FBS for 2 days.

Cell lysis and Western blotting

Cells were lysed directly on the plate by the addition of ice-cold lysis buffer (50 mM Tris/HCl, 50 mM 2-glycerophosphate, 1 mM EGTA, 1 mM EDTA and 1% (v/v) Triton X-100. 2-Mercaptoethanol, Na3VO4 and protease inhibitor cocktail were added to the lysis buffer just before use. Lysates were centrifuged at 4°C at 16000 g for 10 min and then the supernatant was collected. Protein concentrations were determined by Bradford reagent (Bio-Rad Laboratories). Cell lysate or immunoprecipitated samples were heated at 95°C for 7 min in sample buffer [62.5 mM Tris/HCl, pH 6.8, 7% (w/v) SDS, 20% (w/v) sucrose and 0.01% Bromophenol Blue] and subjected to PAGE and electrophoretic transfer on to nitrocellulose/PVDF membranes. Membranes were then blocked in PBS-Tween 20 containing 5% (w/v) skimmed milk powder for 1 h at room temperature (~20°C). Membranes were probed with the indicated primary antibody overnight at 4°C. After incubation with fluorescently tagged secondary antibody, signals were scanned using a LI-COR Odyssey imaging system.

Antibodies

All of the antibodies were from Cell Signalling Technology, except for anti-4E-BP1 (raised in rabbit [18]), anti-eIF4G2 (a gift from Professor Simon Morley, Department of Biochemistry, University of Sussex, Brighton, U.K.) and anti-α-tubulin (TU-02, Santa Cruz Biotechnology).

Measurements of protein synthesis rates

Cells were incubated in the presence of inhibitors as indicated for 60 min at 37°C. For the final 60 min, [35S]methionine (PerkinElmer) was added to a final concentration of 10 μCi/ml. After incubation, the medium was removed completely and the cells were washed with ice-cold PBS and lysed using a standard procedure. The protein concentrations in the extracts were then quantified using the Bradford method. Samples of lysate were applied to 3MM filter papers (Whatman) and allowed to dry at room temperature. After two brief washes with 5% (w/v) TCA (trichloroacetic acid), one at 100°C and one in acetone, filters were again dried. Incorporated radioactivity was measured by scintillation counting.

Pull-down assays

For anti-HA immunoprecipitations, 10 μl of Protein G beads were added to lysis buffer and washed once. A 1 μg portion of Anti-HA High Affinity (Roche) was then added and incubated for 1 h at 4°C. Cell lysate was then added and incubated for 1–2 h. The beads were then washed twice with lysis buffer and finally resuspended in SDS/PAGE loading buffer.

To pull down eIF4E and associated proteins, cell extracts were incubated with m7GTP–Sepharose (7-methyl GTP–Sepharose 4B) beads (GE Healthcare), which had been diluted 1:1 with underivatized Sepharose CL-4B, for 1 h at 4°C under constant shaking. Beads were washed twice with lysis buffer and bound proteins were eluted in sample buffer as described above.

pSILAC labelling and MS

HeLa cells growing in serum (and treated as described in the Figure legends) were incubated for 6 h with SILAC medium (Thermo Scientific) containing 154 mg/l L-[13C]6,[15N]2-Lys and 89 mg/l L-[13C]6,[15N]4-Arg (in place of normal lysine and arginine). Only serum dialysed against PBS was used (to remove free lysine and arginine). After incubation, the cells were washed twice and then harvested in PBS. After a short (5 min) centrifugation at 600 g and 20°C, the pellets were resuspended in buffer containing 50 mM ammonium bicarbonate/1% (w/v) sodium deoxycholate and immediately heated to 95°C.

To study protein degradation, two sets of HeLa cells were grown in the medium either containing 77 mg/l L-[13C]6,[15N]2-Lys and 44.5 mg/l L- [13C]6,[15N]4-Arg (heavy-labelled) or 75 mg/l L-[2H]4-Lys and 43.5 mg/l L-[13C]6-Arg (medium-labelled). Only dialysed serum was used. Cells were passaged five times in the above medium in order to reach essentially 100% labelling. Then, the heavy-labelled cells were moved into medium containing normal lysine/arginine and treated with the indicated inhibitors. After 6 h of incubation, the cells were lysed using the methods described for pSILAC. Heavy-labelled samples were combined with samples from the untreated medium-labelled cells (equal amounts of protein were used) prior to trypsin digestion and analysis. The medium-labelled sample was used as an internal control in the present study.

After cooling, the protein concentration of the lysates was measured using the BCA (bicinchoninic acid) method, and 25 μg of total protein was divided into aliquots (25 μl) and reduced (0.5 μg of dithiothreitol for 30 min at 37°C), alkylated (2.5 μg of iodoacetamide for 30 min at 37°C) and trypsinized (0.5 μg of trypsin overnight at 37°C) as described previously [19]. After digestion the sample was diluted 2-fold with a solution of 1% (v/v) trifluoroacetic acid, 3% (v/v) acetonitrile, 0.5% acetic acid and the resulting deoxycholate precipitate was pelleted by centrifugation at 16100 g for 5 min at 22°C. The supernatant containing the peptides was then desalted, concentrated and filtered on C18 STop and Go Extraction tips [20], and eluted directly into a 96-well plate.

Peptide mixtures were analysed on a 1100 Series nanoflow HPLC instrument (Agilent) online coupled through a nanoelectrospray ion source (Proxeon) to an LTQ (linear trapping quadrupole)-Orbitrap (ThermoFisher Scientific) tandem mass spectrometer [19]. Briefly, peptides were injected directly on to a reversed phase (3 μm-diameter ReproSil-Pur C18; Dr Maisch HPLC) column manually packed into a 15 cm-long, 75 μm-inner-diameter, fused silica emitter. Peptides were eluted directly into the LTQ-Orbitrap using a linear gradient from 4.8% acetonitrile to 24% acetonitrile in 0.5% acetic acid over 60 min at a flow rate of 200 nl/min. The LTQ-Orbitrap was set to acquire a full-range scan in the Orbitrap, from which the five most abundant multiply charged ions were selected for fragmentation in the LTQ [19].

Polysome analysis and Northern blot analyses

Polyribosomes were resolved from ribosomal subunits and subribosomal material by sucrose-gradient-density centrifugation; nine fractions were collected per gradient. The absorbance at 254 nm was continuously monitored and RNA was isolated from the two fractions from each gradient and processed for Northern blot analysis as described previously [10]. Northern blots were visualized and quantified using a Typhoon phosphorimager (GE Healthcare).

Real-time RT (reverse transcription)–PCR amplification analysis

Total RNA was extracted by the proteinase K method and then subjected to ImProm-II™ Reverse Transcription System (Promega) with oligo(dT)15 and random primers following the manufacturer's protocol. Subsequently, real-time PCR was performed using specific primers (PrimerDesign) for: FASN (fatty acid synthase) (5′-GCCTACACCCAGAGCTACC-3′; 5′-TGTGCTCCATGTCCGTGAA-3′), TKT (transketolase) (5′CTACAGAGAAGGCAGTGGAAC-3′; 5′-ACCTGGAAGTCCTCATTGTTG-3′), NAP (nucleosome assembly protein) 1-like 1 (5′-CCAGACTATGACCCAAAGAAGG-3′; 5′-AAGGCTGTAAGTAAATAAGAGTTGTG -3′), PRDX6 (peroxiredoxin 6) (5′-CCAGCAGAGAAGGATGAAAAGG-3′; 5′-AGTTCCTGCCAGTGGTAGC-3′), NACA (nascent polypeptide-associated complex α subunit) (5′-GCCCCTGAAATCCCAAAGTC-3′; 5′AGGTGCCAATGCTGAAGTATG-3′), RpS19 (5′-CACGATGCCTGGAGTTACTG-3′; 5′-CCAGCTTGACGGTATCCAC-3′), β-actin (5′-CATTGGCAATGAGCGGTTC-3′; 5′-CCACGTCACACTTCATGATGG-3′), LDH-B (lactate dehydrogenase B) (5′-AGAAATGGGAACTGACAATGATAG-3′; 5′-CAGCCACACTTAATCCAATAGC-3′), HSPA8 (heat-shock 70 kDa protein 8) (5′-GCCCGATTTGAAGAACTGAATG-3′; 5′-CCACCAACCAGGACAATATCAT-3′), GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (5′-CTCTTGTGCTCTTGCTGGG-3′; 5′-ACCCACTCCTCCACCTTTG-3′); HnRNP (heterogeneous nuclear ribonucleoprotein) C (5′-AAATCTGATGTGGAGGCAATCTT-3′; 5′-GGGCATTTCTCTCATTAACATACTG-3′); and 18S rRNA (HK-SY-hu-600 from Primer Design). The samples were analysed in triplicate with SYBR GREEN dye (Primer Design mix) on an ABI StepOnePlus qPCR (quantitative PCR) instrument (Applied Biosystems). The comparative CT method was employed to measure amplification of specific mRNAs compared with the total level of 18S. These results were compared with the corresponding results for untreated cells, and the results are expressed as, for example, ‘target mRNA from treated cells/control’.

RESULTS

mTOR-KIs inhibit protein synthesis more strongly than rapamycin

To examine the effects of different mTOR inhibitors on general protein synthesis, we measured the incorporation of [35S]methionine into TCA-insoluble material in control cells or cells pre-treated with rapamycin or each of two distinct mTOR-KIs, PP242 [5] and AZD8055 [8]. Relatively short times (2 or 6 h) were chosen to minimize the influence of effects on the cell cycle or ribosome biogenesis (regulated through mTOR [21]) which, over longer periods, could affect overall protein synthesis. At both times tested, rapamycin only weakly (5–10%) inhibited protein synthesis, whereas AZD8055 (Figures 1A and 1B) and PP242 (Figure 1B; Y. Huo, unpublished work; and [5]) caused substantially stronger inhibition (25–30% for AZD8055; slightly more for PP242). Others observed similar effects using Torin1 in wild-type MEFs (mouse embryonic fibroblasts) and MEFs lacking mTORC2 (rictor−/− MEFs) [4]. Thus the greater effect of mTOR-KIs on general protein synthesis is apparently not due to effects mediated through mTORC2, but probably through rapamycin-insensitive effects of mTORC1.

Figure 1 Effects of rapamycin and AZD8055 on protein synthesis rates in HeLa cells growing in serum

(A) Serum-fed HeLa cells were pre-treated with rapamycin (100 nM) or AZD8055 (100 nM) for 1 h and then incubated with [35S]methionine (10 μCi/ml) for 1 h. Lysate was added to 3MM filter paper and the incorporated radioactivity was measured as described in the Experimental section. Results are means±S.E.M. (n=3). (B) As (A), but treatment and labelling of cells were for 6 h and PP242 was used at 1 μM. (C) Lysates of untreated serum-fed HeLa cells or from cells treated for 6 h with the indicated inhibitors were fractionated on sucrose-density gradients. The positions of the 60S and 80S ribosomal particles and polysomes are shown; absorbance values at 254 nm are in arbitrary units and are on the same scale for each panel.

Rapamycin had little effect on the distribution of ribosomes between polysomes and 80S/ribosomal subunits (Figure 1C), whereas AZD8055 and PP242 caused partial polysome disaggregation and a marked rise in levels of 80S monomers, indicating inhibition of translation initiation (Figure 1C).

pSILAC analysis reveals qualitative differences in the effects of mTOR-KIs and rapamycin on the synthesis of specific proteins

A key issue is whether the greater effects of mTOR-KIs than rapamycin on overall protein synthesis merely reflect a larger effect on all proteins or also differential effects on synthesis of different proteins. To address this, we applied the pSILAC method to tag newly synthesized proteins, allowing us to identify and quantify changes in their accumulation rates. A similar method was independently reported by Selbach and colleagues [22,23], but has not been used previously to evaluate the effects of different signalling pathways.

For pSILAC, cells are incubated with medium containing lysine and arginine isotopologues for short periods (a few hours) to tag newly synthesized proteins (Supplementary Figure S1 at http://www.BiochemJ.org/bj/444/bj4440141add.htm). After lysis, proteins are digested with trypsin and peptides are analysed by MS. Peptides from newly made proteins appear as heavier species than the corresponding untagged peptide (Supplementary Figure S1B). The ratios of the two variants provide information about the relative rates of synthesis of the parent protein and, by analysing multiple peptides from the same protein, accurate information about their rates of accumulation can be derived. Our aim was to assess whether rapamycin and mTOR-KIs exert differential effects on the synthesis of specific proteins, or whether the latter merely more strongly inhibit the synthesis of all proteins. In the present study, we focused on their effects on the most abundant proteins, which contribute most strongly to cell mass (although our approach can be configured to provide proteome-wide analysis by fractionating the samples prior to MS). It should be noted that, since each molecule of a given protein has the same chance of being degraded, differences in degradation rates will not affect the heavy/light peptide ratios.

We have used labelling times up to 6 h because this yields a high-enough proportion of labelled peptides, thus providing accurate and easily quantifiable results. Importantly, treating cells with mTOR-KIs or rapamycin for 2 or 6 h had similar effects on overall protein synthesis (Figures 1A and 1B). mTOR-KIs had particularly pronounced effects on the synthesis of numerous proteins (Figures 2A and 2B), which were larger than their effects on overall protein synthesis (30–40% inhibition, Figures 1A and 1B). The dataset of proteins for which we obtained data in triplicate for rapamycin and at least one mTOR-KI is provided in Supplementary Table S1 at http://www.BiochemJ.org/bj/444/bj4440141add.htm.

Figure 2 pSILAC data for relative rates of synthesis of different categories of proteins

Where indicated, inhibitors (rapamycin or AZD8055, each at 100 nM; PP242 at 1 μM) were added to the medium when the cells were transferred to the medium containing stable isotope-labelled arginine and lysine. After 6 h, cells were lysed, proteins were digested with trypsin, and the resulting peptides were analysed as described in the Experimental section. (A) Degree of inhibition by the three different inhibitors used of labelling of mRNAs encoded by 5′-TOP mRNAs (5′-TOP) or mRNAs not known to belong to this category (NON-TOP). (B) Average relative rate of synthesis for proteins in each category in cells treated with the indicated inhibitors derived from three independent experiments. S.E.M. values are not appropriate as the data refer to many different proteins within each category. (CE) Relative rates of synthesis of selected proteins in each category; results are means±S.E.M. (n=3). In each case, data are derived from three independent experiments.

mTOR-KIs reveal a strong rapamycin-insensitive component to the control of 5′-TOP mRNAs by mTOR

Prominent among the proteins whose synthesis is strongly affected by mTOR-KIs are those encoded by 5′-TOP mRNAs (Figures 2A–2C). There was generally good agreement between the effects of AZD8055 and PP242 on their synthesis (Figure 2C and Supplementary Table S1). PP242 often had a slightly greater effect than AZD8055, perhaps reflecting additional effects of PP242. Figures 2(B) and 2(C) show that the effects of these compounds on the synthesis of selected 5′-TOP mRNA-encoded proteins follows a fairly consistent pattern; AZD8055 or PP242 caused very strong inhibition (often 70–75%), whereas rapamycin had a markedly weaker effect (30–50% inhibition). Neither AZD8055 nor rapamycin affected the overall levels of the mRNAs for a typical 5′-TOP mRNA (RpS19) or a housekeeping protein (β-actin; Supplementary Figure S2 at http://www.BiochemJ.org/bj/444/bj4440141add.htm). For non-5′-TOP mRNAs, inhibition by AZD8055 of >50% was only observed for approximately 15% of proteins, some of which are listed in Table 1. Rapamycin typically inhibited their synthesis by 20%, and frequently by much less, if at all. Figures 2(B) and 2(D) show data for proteins that are not encoded by known 5′-TOP mRNAs. These findings define clear differences between the behaviour of proteins encoded by 5′-TOP mRNAs and most other proteins, and between the effects of rapamycin and mTOR-KIs.

View this table:
Table 1 Proteins whose synthesis is strongly inhibited by mTOR-KIs or rapamycin

The Table shows pSILAC data for selected proteins whose synthesis is strongly inhibited by mTOR inhibitors, together with information about the probable transcriptional start site in human cells.

Nonetheless, mTOR-KIs and rapamycin did strongly inhibit synthesis of several proteins not known to be encoded by 5′-TOP mRNAs (Figure 2E). Indeed, the extent and pattern of inhibition resemble those observed for known 5′-TOP mRNA products (Figure 2C). The DBTSS (Database of Transcriptional Start Sites) (http://dbtss.hgc.jp/) indicates that some of the corresponding mRNAs start with a cytosine codon and contain a 5′-terminal tract of at least seven pyrimidines (e.g. the mRNAs for PRDX6, NAP1L1 or NACA; Table 1a), similar to some other known 5′-TOP mRNAs [10,11]. Analysis by qPCR showed that interfering with mTOR signalling did not affect the overall levels of these mRNAs (Supplementary Figure S2).

The mRNAs for some of the other proteins whose synthesis is strongly decreased by mTOR inhibitors also start with cytosine (e.g. most of those listed in Table 1b); however, this is not followed by an uninterrupted stretch of pyrimidines, as found in 5′-TOP mRNAs. The levels of these mRNAs were not affected by rapamycin or AZD8055, consistent with their synthesis being controlled at the level of translation (Supplementary Figure S2). Interestingly, some of them, similar to other 5′-TOP mRNA-encoded proteins, are involved in protein biosynthesis, e.g. HSPA8, a member of the hsp (heat-shock protein) 70 family, and NACA. Others are enzymes involved in glycolysis, e.g. GAPDH (Figure 2E) and LDH-B (Figure 2E), and thus in anaerobic glucose oxidation, which is enhanced in tumour cells (the Warburg effect [24]). Some transcripts for LDH-B contain a 5′-TOP-like feature (Table 1).

The synthesis of TKT, the enzyme which links the PPP (pentose phosphate pathway) to glycolysis, was also strongly suppressed by mTOR-KIs (Figure 2E). This pathway (but not the step catalysed by TKT) plays a key role in anabolism as it generates NADPH (for fatty acid synthesis) and ribonucleotides for RNA synthesis (and thus for ribosome biogenesis). mTOR inhibitors also strongly impaired synthesis of fatty acid synthase, a key enzyme of lipogenesis. RT–qPCR analysis showed that mTOR inhibitors did not change the levels of the mRNAs for FASN or TKT (Supplementary Figure S2). Thus it is likely that their decreased rates of synthesis reflect repression of their translation. Neither of these mRNAs appears to possess a canonical 5′-TOP; other features presumably control their translation. Increased flux through the PPP, glycolysis and lipogenesis, e.g. through increased levels of its enzymes, will promote anabolic pathways that are activated in cancer cells [25].

mTOR inhibitors also strongly impaired (by approximately 60%) the synthesis of some members of the HnRNP family of RNA-binding proteins (e.g. HnRNP C, Figure 2E; and HnRNP F, L, U and G, Supplementary Table S1). The mRNA for HnRNP C does not contain a canonical 5′-TOP (Table 1) but, since its levels were not affected by mTOR inhibition (Supplementary Figure S2), changes in its synthesis are probably due to alterations in translational efficiency. The mRNA for HnRNP A1 is a known 5′-TOP message [26]; the DBTSS database analysis indicates that the mRNAs for HnRNP L and G may also contain a 5′-TOP. In contrast, the mRNA for HnRNP A3 clearly does not contain one.

Analysis of the association of specific mRNAs with polysomes

To study the effects of mTOR inhibitors on the behaviour of specific mRNAs, RNA was extracted from sucrose gradient fractions and analysed by Northern blotting. As expected, treatment with rapamycin, AZD8055 or PP242 had little effect on the distribution of the non-TOP HnRNP A3 mRNA (Figures 3A and 3B).

Figure 3 Distribution of mRNAs of interest between polysomal and non-polysomal fractions

Northern blots were performed on RNA extracted from fractions from sucrose-density gradients from untreated HeLa cells or cells treated for 6 h with rapamycin (100 nM), PP242 (1 μM) or AZD8055 (100 nM), as indicated. (A) Northern blot data for the mRNAs indicated. In (A) the positions of the ribosomal particles and polysomes are noted; (BG) show quantification of the signals as the percentage of the total of each mRNA detected on the blot (in each fraction).

The mTOR-KIs decreased the amount of polysome-associated β-actin mRNA, a non-5′-TOP message (Figures 3A and 3C). In contrast, rapamycin or mTOR-KIs caused a much bigger decrease in polysome-associated RpS19 mRNA, a known 5′-TOP mRNA (Figures 3A and 3D), with the effect of AZD8055 or PP242 being much greater than that of rapamycin, consistent with their larger effects on the synthesis of proteins encoded by 5′-TOP mRNAs (Figures 2A and 2C). Importantly, this reveals a rapamycin-insensitive component to the regulation of the translation of this mRNA by mTOR. The Northern blot analyses indicate that mTOR inhibitors affect the distribution of this and other mRNAs between different ribosomal particles, rather than changing their total levels. The latter was confirmed by qPCR analysis, which showed that mTOR inhibitors did not decrease the levels of these mRNAs (Supplementary Figure S2).

We also examined the effects of mTOR inhibitors on the distribution of the mRNAs for PRDX6 and NAP1L1, proteins whose synthesis is strongly inhibited by rapamycin or mTOR-KIs (Figure 2E). Rapamycin caused a partial shift of PRDX6 and a strong shift of the NAP1L1 mRNAs out of polysomes (Figures 3A, 3E and 3F). mTOR-KIs exerted a much greater effect on the proportion of the PRDX6 mRNA in polysomes (Figure 3E). For NAP1L1, rapamycin already caused a substantial decrease (Figure 3F). These effects are much greater than those observed for the β-actin mRNA and resemble those for RpS19, consistent with the fact that the NAP1L1 and PRDX6 mRNAs contain a pyrimidine-rich tract and could be 5′-TOP mRNAs (Table 1). Rapamycin or mTOR-KIs also caused a marked shift of the GAPDH mRNA out of polysomes (Figures 3A and 3G), indicating that mTOR(C1) signalling controls its translation.

Effects of mTOR-KIs and rapamycin on effectors of mTORC1 linked to mRNA translation

Why do mTOR-KIs have a greater effect than rapamycin on general protein synthesis? This seemed likely to involve known mTORC1 effectors involved in protein synthesis, e.g. the 4E-BPs (and formation of eIF4F complexes containing eIF4G) or the S6Ks (and their substrates linked to mRNA translation; [27]). Treatment of HeLa cells with rapamycin for 60 min strongly inhibited phosphorylation of RpS6 at Ser235/Ser236 (Figure 4A), as did both mTOR-KIs (Figure 4A). eEF2 (eukaryotic elongation factor 2) kinase is phosphorylated and inhibited by S6K1 [28]. Consistent with this, all three compounds increased eEF2 phosphorylation (Figure 4A).

Figure 4 Effects of rapamycin, PP242 and AZD8055 on downstream signalling targets of mTORC1 in serum-fed HeLa cells

(A and B) Serum-fed HeLa cells were treated with the compounds at the indicated concentrations for 1 h. Cell lysates were analysed by Western blotting using the indicated antibodies. Asterisks (*) indicate a non-specific cross-reaction. rapa, rapamycin.

In contrast, rapamycin appeared not to affect phosphorylation of 4E-BP1 at Thr37/Thr46, Ser65 or Thr70 (Figure 4B), whereas AZD8055 and PP242 did inhibit phosphorylation of Thr37/Thr46 and caused a shift to faster-migrating hypophosphorylated forms; differential effects of rapamycin and mTOR-KIs on 4E-BP1 have been reported previously (e.g. [4,29]). However, neither AZD8055 nor PP242 reduced the phosphorylation of 4E-BP1 at Thr70 or eliminated the signal seen with the anti-phospho-Ser65 antibody (Figure 4B). This antibody also recognizes another site in human 4E-BP1, Ser101 [6], which is not affected by rapamycin, and this probably explains the absence of an effect of the inhibitors on the signal seen with this antibody. Phosphorylation of Thr70 is not affected by other manipulations that impair mTOR signalling [6], suggesting that it is not a target for mTOR in vivo.

To study the effects of rapamycin and mTOR-KIs on the association of eIF4G with eIF4E, lysates from HeLa cells treated with AZD8055 or rapamycin were subjected to affinity chromatography on m7GTP–Sepharose, which retains eIF4E and associated proteins. Bound proteins were analysed by SDS/PAGE and Western blotting. As expected, eIF4G and its partner eIF4A were retained on the m7GTP beads, but not on ‘negative control’ Sepharose CL-4B beads (Figure 5A).

Figure 5 Effects of rapamycin and mTOR-KIs on binding of partner proteins to eIF4E

HeLa cells growing in serum were treated, where indicated, with rapamycin (rapa), PP242 (PP) or AZD8055 (AZD) for the times shown (A) or for 1 h (C and D). Samples of lysate were subjected to affinity chromatography on m7GTP–Sepharose. Bound proteins were analysed by Western blotting using the indicated antibodies (A, C and D). Note that in (C), the left-hand three lanes show eIF4G1 and the right-hand three lanes show eIF4G2; similarly in (D), the left-hand panels show 4E-BP1 and the right-hand panels show 4E-BP2. Samples of input cell lysate were also analysed for eIF4E. As a control (Con) for possible non-specific binding of proteins to the matrix, a sample of lysate from untreated cells was mixed with Sepharose CL-4B beads. (B) Quantification of data for eIF4G–eIF4E and 4E-BP1–eIF4E binding from a typical experiment in arbitrary units. (E) The lysates were subjected to affinity chromatography on m7GTP–Sepharose and the bound material was analysed by Western blotting using the indicated antibodies. The positions of endogenous (endog.) and HA-tagged eIF4E are shown.

AZD8055 substantially increased the amount of 4E-BP1 bound to eIF4E and caused a reciprocal decrease in binding to eIF4G1 and eIF4A (Figure 5A; quantified in Figure 5B). Rapamycin only caused a small increase in 4E-BP1 binding, but, surprisingly but reproducibly, also slightly increased eIF4G binding. Levels of eIF4A and eIF4G1 changed similarly, indicating that mTOR signalling does not affect eIF4A/eIF4G binding. Rapamycin and AZD8055 had differing effects on the association of eIF4E with eIF4G2 (Figure 5C). mTOR-KIs also elicited a much greater increase in the association of 4E-BP2 with eIF4E than rapamycin (Figure 5D). The strong inhibitory effect of mTOR-KIs on binding of eIF4G to eIF4E, which is considered a key step in translation initiation, may account for their ability to induce dissociation of polysomes (Figure 1C). Since rapamycin and mTOR-KIs both completely inhibit S6 phosphorylation, the phosphorylation of this and other S6K substrates cannot account for their differential effects on protein synthesis.

Control of 5′-TOP mRNA translation by mTOR is independent of the regulation of eIF4E

It is not clear how mTOR(C1) signalling regulates 5′-TOP mRNA translation. However, inducing expression of eIF4E does favour recruitment of many 5′-TOP mRNAs into polysomes [30].

To test the idea that the dephosphorylation of 4E-BP1, and the consequent inhibition of eIF4E function, caused by mTOR-KIs is responsible for their effects on protein synthesis, we first tried to knock down 4E-BP1 expression using siRNA (small interfering RNA). However, we were unable to eliminate 4E-BP1, and AZD8055 still caused appreciable binding of 4E-BP1 to eIF4E and decreased eIF4G binding (Supplementary Figure S3A at http://www.BiochemJ.org/bj/444/bj4440141add.htm).

We therefore overexpressed eIF4E, arguing that increased levels of eIF4E would buffer non-phosphorylated 4E-BPs, allowing cells to maintain higher levels of eIF4G–eIF4E binding. HA–eIF4E bound to eIF4G1, eIF4G2 and 4E-BP1 (Supplementary Figure S3B), confirming its functionality. Cells expressing HA–eIF4E showed levels of binding of eIF4G to eIF4E similar to those in untreated cells even when exposed to mTOR-KIs (Figure 5E). Thus overexpressing eIF4E eliminates their effect on eIF4G/eIF4E binding.

We also tested whether overexpressing eIF4E could also ‘rescue’ the translation of 5′-TOP mRNAs. Expressing HA–eIF4E promoted assembly of polyribosomes (Figure 6A), but did not overcome the inhibitory effect of mTOR-KIs on total protein synthesis (Figure 6C), indicating that this effect is not only due to decreased binding of eIF4G (and its partners) to eIF4E.

Figure 6 Effect of overexpressing eIF4E on polysome size distribution and protein synthesis

Where indicated, HeLa cells were transfected with a vector encoding HA–eIF4E. Where noted, cells were treated for 6 h with rapamycin (100 nM) or AZD8055 (100 nM). (A) Sucrose-density gradient profiles with the positions of the indicated ribosomal species shown. (B) Western blots indicating the expression levels of eIF4E and HA–eIF4E [developed with anti-(human eIF4E) antibody] and the phosphorylation of S6 at the mTORC1-specific sites, Ser240/Ser244, to verify the efficacy of the mTOR inhibitors used. AZD, AZD8055; Rap, rapamycin. (C) The incorporation of [35S]methionine was determined. The data are derived from at least three independent experiments in each case. P values between the two groups (±eIF4E) were calculated using individual Student's t tests. Results are means±S.E.M. (n≥4).

Increased expression of eIF4E promoted the recruitment into polysomes of the known 5′-TOP mRNA for RpL11 and those encoding PRDX6 and NAP1L1, potential 5′-TOP mRNAs (Figures 7A and 7C–7E; cf. Table 1), as well as the β-actin mRNA (Figures 7A and 7B). Nonetheless, rapamycin or AZD8055 still caused marked shifts of the RpL11, NAP1L1 and PRDX6 mRNAs out of polysomal fractions (Figures 7C–7F), similar to those observed in control cells (cf. Figure 3 and Figures 7A and 7C–7E). Overexpressing eIF4E also did not alter the response of the GAPDH mRNA to mTOR inhibitors (Figures 7A and 7F). Thus factors other than eIF4E and its partners modulate 5′-TOP mRNA translation downstream of mTOR. This is consistent with earlier data that overexpressing eIF4E did not overcome the repression of 5′-TOP mRNA translation seen in quiescent NIH 3T3 cells [31].

Figure 7 Overexpression of eIF4E does not prevent the repression of 5′-TOP mRNA translation by mTOR inhibitors

Northern blots were performed on RNA extracted from fractions from sucrose-density gradients from control HeLa cells, HeLa cells overexpressing HA–eIF4E or equivalent cells treated for 6 h with rapamycin (100 nM) or AZD8055 (100 nM), as indicated. (A) Northern blot data for the mRNAs indicated. The positions of the ribosomal particles and polysomes are noted. (BF) Quantification of the signals for each fraction as a percentage of the total of each mRNA detected on the blot.

Effects of mTOR-KIs and rapamycin on protein breakdown

Steady-state levels of specific proteins, and thus the cellular proteome, will be determined by their rates of degradation as well as synthesis. We used a modified SILAC approach to study the effects of mTOR inhibition on the degradation of specific proteins: cells were first grown for 12 generations in medium containing ‘heavy’ isotopically labelled arginine/lysine and then transferred to fresh medium containing normal (12C/14N) arginine/lysine in the presence or absence of AZD8055 or rapamycin. To avoid possible artefacts arising from the observation, as shown above, that mTOR inhibition has differential effects on synthesis of different proteins, samples of lysates from these cells were combined with equal amounts of lysate from cells maintained for the same time in medium containing ‘medium’ tagged versions of arginine/lysine. In this ‘SILAC–chase’ approach, the ratio of heavy/medium versions of given peptides yields information on the relative decay rates of pre-existing proteins from the drug-treated cells, which are tagged with ‘heavy’ lysine/arginine.

From the heavy/medium isotope ratios from multiple peptides derived from almost 400 proteins for which data were obtained in triplicate for all three conditions, the average protein half-lives were calculated as approximately 19 h. Rapamycin had only a modest effect on this (Table 2). AZD8055 extended the average to 22 h. Thus, in general, inhibition of mTOR signalling appears to stabilize proteins, even though inhibition of mTOR can stimulate autophagy [32]. Very few proteins showed decreased half-lives in response to mTOR inhibitors.

View this table:
Table 2 Effects of mTOR inhibition on protein degradation

Data for degradation rates were obtained by the ‘SILAC–chase’ method described in the Experimental section, for untreated cells, or cells treated with rapamycin or AZD8055. These data were converted into protein half-lives for all proteins (or those within a given category) for which triplicate data were available for all three conditions. Results are means±S.E.M.

Proteins encoded by 5′-TOP mRNAs generally have slightly longer half-lives than average (22.7 h compared with 18.9 h; Table 2). Interestingly, rapamycin substantially increased the half-lives of this set of proteins (mainly Rps), from 22.7 to 27.2 h, and AZD8055 did so to a considerably greater extent (to 32 h; Table 2). The number of proteins in this set which are not Rps (e.g. cytoplasmic elongation factors) was too small to ascertain whether mTOR inhibition affects their degradation similarly to that of Rps.

DISCUSSION

In the present paper we describe the first detailed study comparing the effects of inhibiting mTOR signalling on protein synthesis and degradation, including the effects of a recently reported specific mTOR-KI, AZD8055 [8]. AZD8055 inhibited ongoing protein synthesis more strongly than rapamycin, consistent with recent data for PP242 [5] and Torin1 [4]. In principle, this difference could reflect a role for mTORC2 in controlling general protein synthesis. However, the larger effects of Torin1 on protein synthesis are apparently not mediated through mTORC2, since its effects were almost identical in rictor−/− and control MEFs [4]. The greater effects of mTOR-KIs thus seem to reflect roles either for rapamycin-insensitive functions of mTORC1 or a (hypothetical) mTORC3.

mTOR-KIs inhibit the phosphorylation of 4E-BP1 much more strongly than rapamycin, consistent with previous findings [46,33]. Consequently, decreased binding of eIF4G to eIF4E was only observed after treatment with PP242 or AZD8055. Puzzlingly, rapamycin often caused a modest increase in this, for reasons which are unclear and require further study. This observation is potentially important with respect to the application of rapamycin as an anticancer therapy, since increased eIF4G–eIF4E binding corresponds to activation of eIF4E.

To study the effects of different mTOR inhibition on the synthesis of specific proteins, we employed a recently developed pSILAC approach, which offers several advantages over previous methods for studying changes in the synthesis of specific proteins. One widely used method is to label newly made proteins with a radioactive amino acid (often [35S]methionine), resolve them by two-dimensional gel electrophoresis and assess changes in labelling from the intensity of individual ‘spots’. This is laborious and relatively low throughput. Furthermore, it can often be hard to identify specific proteins, and the occurrence of multiple differentially charged (e.g. phosphorylated) species of the same protein can cause considerable confusion. It is also hard to detect, let alone quantify, low-abundance proteins using the [35S]methionine-labelling approach, whereas such proteins can be detected in pSILAC studies if complex samples are fractionated prior to MS analysis.

A second method to study changes in mRNA translation employs sucrose-density-gradient centrifugation to resolve active polysomal mRNAs from translationally inactive ones, and then the mRNA populations are analysed by DNA microarray [34,35] or quantitative real-time PCR. This approach suffers from potential drawbacks that are avoided by the pSILAC approach. Above all, this method does not provide information on the actual rates of synthesis of individual proteins, only on the behaviour of their mRNAs. It may fail to reflect changes where an mRNA merely shifts between polysomes of different sizes rather than into/out of polysomes. Changes in synthesis rates may be misreported where an mRNA's translation is inhibited at the level of elongation (which would result in its accumulation in polyribosomes) or in certain types of microRNA-mediated translational inhibition [36]. Because it reflects actual rates of synthesis of individual proteins, the pSILAC method will be valuable in studying the control of the synthesis of specific proteins in a wide range of situations.

Our pSILAC data reveal that mTOR inhibitors have a particularly strong effect on the synthesis of proteins encoded by 5′-TOP mRNAs. Translation of these mRNAs is known to be impaired by rapamycin, although the effects are small, especially in cells maintained in serum; the present data reveal that mTOR-KIs exert a much stronger inhibitory effect under such conditions, demonstrating that rapamycin-insensitive effects of mTOR play a role here [14]. The PI3K inhibitor LY294002 has a much greater inhibitory effect than rapamycin on 5′-TOP mRNA translation [15]. Since it also interferes with mTOR activity [37], its marked effects on 5′-TOP mRNA translation may reflect interference with rapamycin-insensitive outputs from mTOR, rather than PI3K. Importantly, our data imply that mTORC1 regulates 5′-TOP mRNA translation through a mechanism(s) that is only partially inhibited by rapamycin, but more strongly by mTOR-KIs. Given that many 5′-TOP mRNAs encode ribosomal proteins or translation factors [10,11], the strong inhibition of their synthesis by mTOR-KIs will reduce general protein synthesis rates and cell growth/proliferation.

The synthesis of certain other proteins showed features reminiscent of the behaviour of proteins encoded by 5′-TOP mRNAs, i.e. strong inhibition by mTOR-KIs and relatively strong effects of rapamycin. In several cases, the 5′-UTRs (untranslated regions) of the corresponding mRNAs show two features that are shared with 5′-TOP mRNAs, i.e. the first base is a C followed by a stretch of pyrimidines [11]. Further experiments will be required to confirm that these mRNAs are 5′-TOP mRNAs. As noted above, several of these mRNAs encode proteins involved in protein synthesis or in producing the energy or ribonucleotides needed for rRNA or protein production. Impairment of their synthesis by mTOR-KIs will probably impair cell growth and proliferation. Importantly, these results suggest that the set of 5′-TOP mRNAs includes those for proteins involved in anabolic pathways in addition to protein synthesis.

Our results clearly show that overexpressing eIF4E in HeLa cells increases the proportion of ribosomes in polysomes and the polysomal association of 5′-TOP mRNAs (results not shown), in agreement with previous work [30]. Overexpressing eIF4E also prevents the inhibition of eIF4F formation, which is otherwise caused by mTOR-KIs. However, it did not prevent the inhibitory effects of mTOR inhibition on polysomal association of 5′-TOP mRNAs. Similar behaviour was observed both for established 5′-TOP mRNAs and the putative ‘novel’ 5′-TOP mRNAs described in the present study. Thus eIF4E–eIF4G association is insufficient to maintain 5′-TOP mRNA translation. Very recent data indicate that the RNA-binding proteins TIA-1 and TIAR play a role in the translational control of 5′-TOP mRNAs by amino acids [38], but it is unclear whether they are involved in their regulation by mTOR. Our finding that mTOR-KIs strongly impair the synthesis of additional proteins (e.g. enzymes of intermediary metabolism) has important implications for understanding the impact of mTOR signalling on the proteome and cellular metabolism.

The protein degradation data are important for three main reasons. First, they reveal that mTOR signalling influences the levels of most proteins by modulating rates of synthesis rather than degradation. Secondly, they show that although mTOR inhibition slows the rate of protein synthesis, it does not trigger general protein decay. Indeed, thirdly, mTOR inhibition actually extends the half-lives of certain proteins, in particular Rps. The slower removal of these abundant and essential ‘housekeeping’ proteins, under conditions of mTOR inhibition (e.g. nutrient or energy shortage) makes good physiological sense as it allows cells to conserve resources by retaining existing these proteins rather than degrading them and having to resynthesize them, expensively, when conditions improve.

While the present paper was in preparation, another report described an alternative SILAC-based procedure for studying protein dynamics [39]. This revealed that abundant proteins have longer than average half-lives. It also showed that cytoplasmic Rps are longer lived than their nucleolar counterparts, probably because the latter are not yet assembled into ribosomes. The inhibition of ribosome biogenesis (and thus the synthesis of new, unassembled, Rps) caused by inhibiting mTOR may contribute to extending Rp half-life. However, given that unassembled Rps are only a small proportion of the total, this affect is unlikely to explain the marked increase in the half-lives of Rps. The results of the present study thus indicate that, under the conditions used in the present study, mTOR inhibition does not activate ribophagy [40]. Thus although inhibition of mTOR impairs ribosomal protein synthesis, it also inhibits their degradation.

AUTHOR CONTRIBUTION

Yilin Huo and Valentina Iadevaia conducted most of the experiments. Zhong Yao helped to develop the pSILAC method and Isabelle Kelly conducted MS analyses. Sabina Cosulich and Sylvie Guichard provided valuable input into the design and interpretation of the experiments. Leonard Foster and Christopher Proud conceived the pSILAC approach and helped analyse the MS data. Christopher Proud and Leonard Foster designed the overall study and wrote the paper.

FUNDING

This study was supported by funding from the European Union [grant number 229604], the Ajinomoto Amino Acid Research Program and AstraZeneca U.K.

Acknowledgments

We are grateful to Dr Justin Kenney (Centre for Biological Sciences, University of Southampton, Southampton, U.K.) for helpful discussions.

Abbreviations: 4E-BP, eIF4E-binding protein; eEF2, eukaryotic elongation factor 2; eIF, eukaryotic initiation factor; FASN, fatty acid synthase; FBS, fetal bovine serum; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HA, haemagglutinin; HnRNP, heterogeneous nuclear ribonucleoprotein; HSPA8, heat-shock 70 kDa protein 8; LDH-B, lactate dehydrogenase B; LTQ, linear trapping quadrupole; m7GTP–Sepharose, 7-methyl GTP–Sepharose 4B; MEF, mouse embryonic fibroblast; mTOR, mammalian target of rapamycin; mTORC, mTOR complex; mTOR-KI, mTOR kinase inhibitor; NACA, nascent polypeptide-associated complex α subunit; NAP1L1, NAP (nucleosome assembly protein) 1-like 1; PI3K, phosphoinositide 3-kinase; PPP, pentose phosphate pathway; PRDX6, peroxiredoxin 6; qPCR, quantitative PCR; raptor, regulatory associated protein of mTOR; rictor, rapamycin-insensitive companion of mTOR; Rp, ribosome protein; RT, reverse transcription; S6K, S6 kinase; SILAC, stable isotope-labelling with amino acids in cell culture; pSILAC, pulsed SILAC; TCA, trichloroacetic acid; TKT, transketolase; 5′-TOP, 5′-terminal oligopyrimidine

References

View Abstract