Research ArticleCancer

Signaling from mTOR to eIF2α mediates cell migration in response to the chemotherapeutic doxorubicin

See allHide authors and affiliations

Science Signaling  17 Dec 2019:
Vol. 12, Issue 612, eaaw6763
DOI: 10.1126/scisignal.aaw6763

Moved by DNA damage

Doxorubicin is a commonly used therapeutic that kills cancer cells by inducing genotoxic stress. Harvey et al. found that, at clinically relevant doses, doxorubicin promoted cancer cell migration. Doxorubicin inhibited the multiprotein complex mTORC1, which led to the phosphorylation of the translation factor eIF2α. This phosphorylation event would normally be expected to inhibit protein synthesis, but instead, it led to increased migration of doxorubicin-treated cancer cells. Inhibiting eIF2α phosphorylation with an FDA-approved drug restricted doxorubicin-induced cell migration. These results suggest that although inhibiting mTORC1 may shut down protein synthesis in tumors, it may enhance metastasis in response to doxorubicin.

Abstract

After exposure to cytotoxic chemotherapeutics, tumor cells alter their translatome to promote cell survival programs through the regulation of eukaryotic initiation factor 4F (eIF4F) and ternary complex. Compounds that block mTOR signaling and eIF4F complex formation, such as rapamycin and its analogs, have been used in combination therapies to enhance cell killing, although their success has been limited. This is likely because the cross-talk between signaling pathways that coordinate eIF4F regulation with ternary complex formation after treatment with genotoxic therapeutics has not been fully explored. Here, we described a regulatory pathway downstream of p53 in which inhibition of mTOR after DNA damage promoted cross-talk signaling and led to eIF2α phosphorylation. We showed that eIF2α phosphorylation did not inhibit protein synthesis but was instead required for cell migration and that pharmacologically blocking this pathway with either ISRIB or trazodone limited cell migration. These results support the notion that therapeutic targeting of eIF2α signaling could restrict tumor cell metastasis and invasion and could be beneficial to subsets of patients with cancer.

INTRODUCTION

Protein synthesis is an energy-consuming cellular metabolic process (1), and in response to stress or damage, mammalian cells rapidly reprogram protein synthesis to reduce energy demands and adapt to the stress imposed (2). Control of protein synthesis is complex, with many regulatory signaling pathways converging on the factors regulating cap-dependent mRNA translation, which is a three-step process (initiation, elongation, and termination). The process of mRNA translation is predominantly regulated through the modification of the phosphorylation status of eukaryotic initiation factors (eIF) and elongation factors (3).

Regulation of protein synthesis at initiation is controlled by two critical rate-limiting steps; eIF4F assembly and ternary complex formation (3). The eIF4F complex (composed of the cap-binding protein eIF4E, the DEAD-box helicase eIF4A, and the scaffold protein eIF4G) assembles at the 5′ cap structure of mRNA. eIF4E bioavailability, and thus eIF4F complex formation, is regulated by a family of eIF4E-binding proteins (4E-BPs) that compete with eIF4G for a single binding site on eIF4E (4). Through this mechanism, 4E-BPs inhibit cap-dependent translation by preventing the subsequent recruitment of the 43S complex. The capacity of 4E-BPs to bind eIF4E is regulated by mechanistic target of rapamycin complex 1 (mTORC1). In response to growth factors and amino acids, mTORC1 promotes translation initiation through the phosphorylation of 4E-BPs and p70 S6 kinase (S6K) (5). Upon mTORC1 activation, 4E-BPs are hyperphosphorylated, inhibiting binding to eIF4E and thereby enabling eIF4F formation. Conversely, in response to mTORC1 inhibition, 4E-BPs are hypophosphorylated, promoting their binding to eIF4E and inhibition of eIF4F formation (6).

Ternary complex is composed of eIF2 in complex with guanosine 5′-triphosphate (GTP) and the initiator tRNA (tRNAimet) and functions to deliver this tRNA to the 43S complex during translation initiation. After start codon recognition, eIF2 undergoes GTP hydrolysis and is released from the initiation complex as eIF2–guanosine diphosphate (GDP), which is then recycled by the guanine nucleotide exchange factor eIF2B back to eIF2-GTP (3). eIF2 is phosphorylated within the α subunit (eIF2α) in response to a range of stresses, which enhances the affinity of eIF2-GDP for eIF2B, thus decreasing the availability of ternary complex and inhibiting translation initiation (7). Four eIF2 kinases (eIF2Ks) have been identified in mammalian cells: GCN2 (general control nonderepressible-2), PKR (protein kinase double-stranded RNA dependent), PERK [PKR-like endoplasmic reticulum (ER) kinase], and HRI (heme-regulated inhibitor), which phosphorylate eIF2α in response to a range of stress stimuli including amino acid deprivation, ultraviolet (UV)–induced DNA damage, viral infection, ER stress, and heme deficiency (8). In addition to inhibiting global protein synthesis, eIF2α phosphorylation enables the cell to reprogram translation (2) and selectively translate mRNAs containing upstream open reading frames (uORFs) (7), which encode proteins required for the stress response. For example, in response to bulky DNA adduct-inducing agents, including anticancer chemotherapeutics such as cisplatin, protein synthesis is rapidly inhibited through GCN2-dependent phosphorylation of eIF2α (911), whereas the translation of uORF-containing mRNAs, which encode proteins that regulate DNA damage repair pathways, is simultaneously enhanced (10, 11).

The formation of ternary complex and eIF4F were originally thought to be distinct processes that converged to regulate translation initiation. However, it has been shown that ternary complex availability and eIF4F formation are exquisitely coordinated to provide a robust response to cellular stress. For instance, acute activation of mTORC1 inhibits 4E-BPs and enhances eIF4F formation while simultaneously decreasing eIF2α phosphorylation through a mechanism dependent on the phosphorylation of eIF2β to increase ternary complex availability (12). Conversely, catalytic mTORC1 inhibition enhances eIF2α phosphorylation through protein phosphatase 6 (PP6)–dependent activation of GCN2 (13). Furthermore, PERK/GCN2-dependent eIF2α phosphorylation induced by oxidative stress was shown to diminish mTORC1 activity (14), whereas mTORC1 hyperactivation in tuberous sclerosis complex 2 (TSC2)–deficient cells diminishes PERK activity and eIF2α phosphorylation (15). Moreover, Akt has been implicated in maintaining an inhibitory phosphorylation site on PERK, preventing its activation (16). These studies suggest that mTOR and eIF2K signaling are closely intertwined and enable cells to form a coordinated, specific, and robust response to cell stress because it is not beneficial for a cell to expend energy on forming ternary complex when eIF4F is not present to recruit the 43S subunit.

During tumorigenesis, the same pathways that limit protein synthesis under stressed conditions permit survival and the increased proliferation of cancer cells in the unfavorable growth conditions within the tumor microenvironment. Therefore, the signaling pathways and proteins that are involved in these processes provide attractive therapeutic avenues for cancer treatments (1719). Initially, mTOR inhibitors such as rapamycin and its analogs (rapalogs), including temsirolimus and everolimus, were tested in various human tumors as both monotherapies and combination treatments with doxorubicin (Adriamycin) (20, 21), although response rates in clinical trials are modest. There has been increased research into the consequences of inhibition of signaling from eIF2Ks to restrict activation of the prosurvival unfolded protein response in tumor cells, with some promising data in this regard (22, 23). However, what is poorly understood and has not been investigated is the extent of cross-talk between mTOR and eIF2K signaling in response to cancer chemotherapeutics and how combination therapies based on the inhibition of these pathways affect cell survival. Here, we set out to address these questions by studying the regulation of protein synthesis in response to doxorubicin-induced DNA damage, and because p53 is pivotal in this response, we chose to use the MCF10A cell line, which is non–tumor derived, and a parallel MCF10A cell line in which p53 is deleted (MCF10Ap53−/−). We identified a DNA damage–specific, cross-talk signaling pathway between mTOR and eIF2 in response to a physiological and clinically relevant stimulus. We showed that doxorubicin suppressed protein synthesis through the p53-dependent inhibition of mTOR signaling and that doxorubicin-induced mTOR inhibition enhanced the phosphorylation of eIF2α through the activation of GCN2 and PERK. Moreover, we showed that this signaling pathway operated in cancer cell lines. Our data suggest that phosphorylation of eIF2α is not necessary for protein synthesis shutdown but, rather, is required for cell migration under stress conditions, an important implication for the use of inhibitors of these pathways in a clinical setting.

RESULTS

Doxorubicin inhibits global protein synthesis through the mTOR axis

Doxorubicin acts as a topoisomerase II poison at concentrations in excess of 400 nM (24, 25) and, at higher doses, can generate DNA damage through indirect mechanisms (25), including the generation of reactive oxygen species (26). Many studies that use cultured cells to investigate the mechanism of action of doxorubicin use concentrations in excess of 1 μM, and although this has provided valuable insights, it should not be considered clinically relevant. Therefore, we chose to study the effects of doxorubicin using a concentration of 500 nM, which did not induce cell death (fig. S1A).

To determine the effect of doxorubicin on global protein synthesis, radiolabeled [35S]-methionine incorporation was used. Protein synthesis was not significantly inhibited until 9 hours after treatment with doxorubicin (Fig. 1A), a considerable delay from the initial recognition of DNA damage, culminating in a total reduction of over 50% at 24 hours. Sucrose density gradient centrifugation was used to measure the distribution of ribosomes between subpolysomes (free ribosomes) and polyribosomes and to determine whether doxorubicin inhibited ribosome recruitment to mRNA. After treatment with doxorubicin for 6 hours, global protein synthesis was not inhibited (Fig. 1A), and this was reflected in the polysome profile analysis, in which polysomes were minimally decreased and subpolysomes were increased (fig. S1B). In contrast, the number of polysomes decreased when protein synthesis was inhibited at 24 hours, suggesting that the inhibition of initiation blocked ribosome loading onto mRNA (Fig. 1B). During a block in translation initiation, ribosomes would ordinarily dissociate from polysomes and accumulate as free ribosomes in subpolysomes. However, ribosomes lost from polysomes did not accumulate within the subpolysomes in response to doxorubicin, suggesting that this treatment did not result in the “classical” initiation block, such as that observed in response to the mTOR inhibitor AZD8055 (fig. S1C). However, these data are consistent with prolonged treatment with doxorubicin resulting in either ribosome degradation or accumulation of ribosomes within nonsoluble stress granules, as has been observed in response to UV-induced DNA damage in yeast (27) and mammalian cells (28).

Fig. 1 Doxorubicin-induced DNA damage inhibited global protein synthesis rates.

(A) MCF10A cells were treated with doxorubicin (500 nM) for the indicated times and pulse-labeled with [35S]-methionine for 30 min. Total counts per minute were normalized to total protein, and values are shown as a fold change relative to untreated control samples for each time point. Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and *P ≤ 0.05, by unpaired Student’s t test (ns, not significant). (B) Comparison of polysome profiles from MCF10A cells untreated or treated continuously with doxorubicin (500 nM) for 24 hours. Cytoplasmic lysates were centrifuged at 38,000 rpm through 10 to 50% sucrose gradients at 4°C for 2 hours, and absorbance was measured at 254 nm (A254) using a flow rate of 1 ml/min. Profiles shown were representative of three independent experiments. (C) Schematic representation of the regulation of translation initiation through mTOR-dependent regulation of eIF4F complex formation and eIF2K-dependnent regulation of ternary complex (TC) formation. (D) MCF10A cells were treated with doxorubicin (500 nM) for the indicated times, lysed, and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments. (E) MCF10A cells were treated with thapsigargin (Tg) (250 nM for 1 hour), with or without ISRIB (200 nM), and pulse-labeled with [35S]-methionine for 30 min. Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05, by unpaired Student’s t test. (F) MCF10A cells were treated with doxorubicin (Dox) (500 nM for 16 hours), with or without ISRIB (200 nM), and pulse-labeled with [35S]-methionine for 30 min. Total counts per minute were normalized to total protein, and values are shown as a fold change relative to untreated control samples for each time point. Error bars represent means ± SD (n = 3 independent experiments); ***P ≤ 0.001 and **P ≤ 0.01, by unpaired Student’s t test. (G) Samples prepared in parallel to (E) and (F) were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments.

mRNA translation initiation is inhibited at the initiation stage by the phosphorylation of 4E-BPs and eIF2, which regulate eIF4F and ternary complex formation, respectively (Fig. 1C). To understand the regulation of eIF2 and 4E-BPs in relation to the inhibition of translation initiation, Western blot analysis was used to determine the cellular signaling response to doxorubicin. The DNA damage response (DDR) was activated within 3 hours of treatment, as shown by the phosphorylation of ataxia telangiectasia-mutated (ATM) and its substrate checkpoint kinase 2 (Chk2) (Fig. 1D). eIF2α was not phosphorylated until 12 hours after treatment with doxorubicin, at a point when protein synthesis had been robustly inhibited. Moreover, eIF2α phosphorylation was preceded by the inhibition of mTOR signaling, and the reduced phosphorylation of 4E-BP1 observed at 6 to 9 hours after treatment (Fig. 1D) correlated with the inhibition of protein synthesis.

Because eIF2α phosphorylation was only observed after mTORC1 inhibition and did not correlate with the initial inhibition of protein synthesis, it was important to establish the role played by eIF2α phosphorylation in this regard. To address this question, we used the integrated stress response inhibitor (ISRIB), which reverses the inhibitory effect of eIF2α phosphorylation on global protein synthesis (29). ISRIB does not inhibit the activity of eIF2Ks or the phosphorylation of eIF2α (29) but suppresses eIF2α phosphorylation–dependent signaling, such as the induction of activating transcription factor 4 (ATF4) expression (30). To demonstrate the functionality of ISRIB in MCF10A cells, thapsigargin was used to induce ER stress and inhibit protein synthesis, and this inhibition was reversed by the addition of ISRIB (Fig. 1E). However, ISRIB did not rescue the doxorubicin-induced decrease in protein synthesis (Fig. 1F), indicating that doxorubicin-induced translation inhibition does not depend on eIF2α phosphorylation. In agreement with a previous study (30), immunoblot analysis showed that cotreatment with ISRIB abolished the increase in ATF4 expression without affecting eIF2α phosphorylation in response to thapsigargin (Fig. 1G). However, ISRIB abolished doxorubicin-induced eIF2α phosphorylation (Fig. 1G), indicating that when cells are subjected to prolonged stress in the presence of ISRIB, the compound may lead to off-target effects or modulate the stress response in an alternative way. Moreover, doxorubicin did not enhance the expression of ATF4 (Fig. 1G and fig. S1D) or other proteins that are up-regulated during the ISR (Fig. 1G). It was unexpected that ATF4 expression was not up-regulated in the presence of high eIF2α phosphorylation; however, this was most likely due to a lack of availability of the ATF4 mRNA because ATF4 expression depends on the coordinated transcriptional and translational regulation during the ISR (31). Regardless of the effects of ISRIB on doxorubicin-induced eIF2α phosphorylation, these data suggested that eIF2α phosphorylation did not make a major contribution to doxorubicin-induced inhibition of protein synthesis, which was most likely regulated by the inhibition of mTORC1 signaling.

p53 is required for doxorubicin-induced mTORC1 inhibition

p53 is an important mediator of the DDR, arresting the cell cycle to enable the repair of DNA or initiating cell death pathways. Moreover, p53 negatively regulates mTOR signaling (Fig. 2A) (32), and mTOR activity is enhanced in p53-deficient tumors (33). Therefore, we investigated the role of p53 in doxorubicin-induced mTORC1 inhibition using an MCF10A p53 knockout cell line (p53−/−).

Fig. 2 Doxorubicin-induced mTOR inhibition depended on p53 activity.

(A) Schematic representation depicting the regulation of mTOR signaling. Blocked arrows indicate the inhibition of the downstream protein, whereas arrows indicate the phosphorylation or activation of the downstream protein. (B) Wild-type p53 MCF10A cells (p53+/+) and p53 knockout (KO) MCF10A cells (p53−/−) were treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments. (C and D) Quantification of p70 S6K phosphorylation at Thr389 (C) and 4E-BP1 phosphorylation at Ser65 (D) from blots shown in (B). Error bars represent means ± SD (n = 3 independent experiments); ***P ≤ 0.001, **P ≤ 0.01, and *P ≤ 0.05, by unpaired Student’s t test. (E) p53+/+ MCF10A cells were transfected with an siRNA directed against p53 (sip53) or a nonspecific scrambled control siRNA (siCt). After two consecutive 24-hour transfections, cells were treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments. (F and G) Quantification of p70 S6K phosphorylation at Thr389 (F) and 4E-BP1 phosphorylation at Ser65 (G) from blots shown in (E). Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and *P ≤ 0.05, by unpaired Student’s t test. (H) Comparison of polysome profiles from MCF10A p53−/− cells untreated or treated continuously with doxorubicin (500 nM) for 24 hours. Cytoplasmic lysates were centrifuged at 38,000 rpm through 10 to 50% sucrose gradients at 4°C for 2 hours, and absorbance was measured at 254 nm using a flow rate of 1 ml/min. Profiles shown were representative of three independent experiments. (I) Polysome/subpolysomal ratio from p53+/+ and p53−/− cells treated with doxorubicin for 24 hours. Ratio was calculated from the area under the curve of polysome profiles separated into polysomal and subpolysomal regions. Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and *P ≤ 0.05, by unpaired Student’s t test. (J) MCF10A p53+/+ and p53−/− cells and p53+/+ cells transfected with siRNA specific for p53 or a control nontargeting siRNA were pulse-labeled with [35S]-methionine for 30 min. Total counts per minute were normalized to total protein, and values are shown as a fold change relative to untreated control samples for each time point. Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and *P ≤ 0.05, by unpaired Student’s t test.

Wild-type MCF10A cells (p53+/+) or cells with genetic ablation of p53 (p53−/−) were treated with doxorubicin or the catalytic mTOR inhibitor AZD8055 for 16 hours (Fig. 2B). As expected, Western blotting showed that p53 was stabilized and the expression of p53 gene targets such as Sestrin2 was enhanced in response to doxorubicin in p53+/+ cells but not in p53−/− cells (Fig. 2B). The doxorubicin-induced decrease in phosphorylated p70 S6K was also abolished in p53−/− cells, and the reduction in the phosphorylation of 4E-BP1 was also relieved (Fig. 2, B to D). Catalytic inhibition of mTOR with AZD8055 robustly inhibited mTOR signaling in both p53+/+ and p53−/− cell lines (Fig. 2, B to D), suggesting that p53 may regulate signaling pathways upstream of mTORC1. This hypothesis was further supported using a small interfering RNA (siRNA) specific to p53 (sip53) to deplete p53 in p53+/+ cells. After treatment with doxorubicin, p53-depleted cells lost the reduction in phosphorylated p70 S6K that occurred in p53+/+ cells transfected with a nontargeting siRNA (Fig. 2, E to G). However, there was minimal change in the phosphorylation of 4E-BP1 (Fig. 2, E to G), suggesting that an additional signaling mechanism may also regulate 4E-BP1 activity. TSC is an important inhibitor of mTORC1 activity (34) that is activated in response to DNA damage (fig. S2A) (35). By depleting TSC2 and hence inactivating TSC, our data showed that doxorubicin-induced mTORC1 inhibition was independent of TSC (fig. S2, B and C).

Analysis of polysome profiles and quantification of the area under the curve of the polysomes and subpolysomes obtained from p53+/+ cells treated with doxorubicin showed a large reduction in polysomally associated mRNAs (Figs. 1B and 2I) compared to p53−/− cells (Fig. 2, H and I). Moreover, using [35S]-methionine incorporation, doxorubicin-induced inhibition of global protein synthesis was partially restored after knockout or knockdown of p53 (Fig. 2J), supporting the polysome profile data. Together, these data are consistent with the notion that p53 mediates mTORC1 inhibition in response to doxorubicin.

mTOR inhibition enhances eIF2α phosphorylation

Communication between mTOR and eIF2 signaling pathways enables the coordination of ternary complex with eIF4F formation in response to cellular stress or growth stimulation (1215). Our data showed that mTORC1 inhibition preceded the phosphorylation of eIF2α after treatment with doxorubicin, suggesting that mTORC1 could regulate eIF2α phosphorylation in response to DNA damage. To investigate this notion further, both p53+/+ and p53−/− MCF10A cells were treated with either doxorubicin or the catalytic mTORC1/2 inhibitor AZD8055. In p53−/− cells, eIF2α phosphorylation was enhanced in response to catalytic mTOR inhibition, indicating that signaling downstream of mTOR was functional, but was not enhanced in response to doxorubicin (Fig. 3, A and B). These data suggest that p53-dependent mTOR inhibition was critical for the subsequent phosphorylation of eIF2α.

Fig. 3 mTOR inhibition enhanced the phosphorylation of eIF2α through a cross-talk signaling cascade.

(A) Wild-type p53 MCF10A cells (p53+/+), p53 knockout MCF10A cells (p53−/−), and p53+/+ cells transfected with either control nontargeting siRNA (siCt) or siRNA specific for p53 (sip53) were treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments. (B) Quantification of eIF2α phosphorylation at Ser51 from blots shown in (A). Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and *P ≤ 0.05, by unpaired Student’s t test. (C) MCF10A p53+/+ cells were treated with Torin 1 (100 nM), AZD8055 (100 nM), or rapamycin (100 nM) for the indicated times. Cells were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of three independent experiments. (D and E) A549 cells were treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours and analyzed by immunoblotting with the indicated antibodies (D). Blots shown were representative of three independent experiments. Quantification of p70 S6K phosphorylation at Thr389 and eIF2α phosphorylation at Ser51 (E) from blots shown in (D). Error bars represent means ± SD (n = 3 independent experiments); ***P ≤ 0.001, **P ≤ 0.01, and *P ≤ 0.05, by unpaired Student’s t test. (F and G) A549 cells that were transfected with either control nontargeting siRNA (siCt) or siRNA specific for p53 (sip53) were treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies (F). Blots shown were representative of three independent experiments. Quantification of p70 S6K phosphorylation at Thr389 and eIF2α phosphorylation at Ser51 (G) from blots shown in (F). Error bars represent means ± SD (n = 3 independent experiments); ***P ≤ 0.001 and *P ≤ 0.05, by unpaired Student’s t test. SESN2, Sestrin2. (H) BT474 cells were grown at either 37°C (leading to loss of p53 function) or 32°C (restoring p53 function) and treated with doxorubicin (500 nM) or AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies. Blots shown were representative of two independent experiments.

To further explore the correlation between mTOR signaling and eIF2α phosphorylation, MCF10A p53+/+ cells were treated with the mTORC1/2 inhibitors Torin 1 and AZD8055 or the allosteric mTORC1 inhibitor rapamycin. The phosphorylation of eIF2α was enhanced after treatment with all three mTOR inhibitors between 3 and 6 hours (Fig. 3C), which corresponds to the delay observed between mTORC1 inhibition and eIF2α phosphorylation in response to doxorubicin (Fig. 1D). To confirm that doxorubicin inhibited mTORC1 and not mTORC2, mTORC2-dependent phosphorylation of Akt at Ser473 was monitored by immunoblot analysis in MCF10A p53+/+ cells treated with either doxorubicin or the catalytic mTORC1/mTORC2 inhibitor AZD8055. Treatment with AZD8055, but not with doxorubicin, reduced phosphorylation of Akt at Ser473 (fig. S3A), suggesting that doxorubicin inhibited mTORC1 but not mTORC2.

To determine whether mTOR-eIF2 cross-talk signaling was also functional in tumor cells, the mechanism was investigated in a range of cancer cell lines. In A549 lung cancer cells, doxorubicin inhibited mTORC1 signaling and both doxorubicin and AZD8055 significantly enhanced eIF2α phosphorylation (Fig. 3, D and E). Moreover, this mTOR-eIF2 cross-talk signaling mechanism was also present in HeLa cells (fig. S3B). Furthermore, in support of data obtained in MCF10A cells, siRNA knockdown of p53 in A549 cells abolished both doxorubicin-induced mTORC1 inhibition and eIF2α phosphorylation but not AZD8055-induced eIF2α phosphorylation (Fig. 3, F and G).

BT474 breast cancer cells contain a temperature-sensitive p53 mutation within the DNA binding domain that results in impaired p53 function (36). Therefore, to further validate the functional role of p53 and mTOR-eIF2 cross-talk signaling in tumor cells, we incubated cells at 37°C to maintain the inactivating p53 mutation or at 32°C to restore p53 function. As expected, doxorubicin treatment did not inhibit mTORC1 signaling or enhance the phosphorylation of eIF2α at 37°C when the mutation actively impaired p53 function (Fig. 3H). However, when p53 function was restored at 32°C, treatment with doxorubicin inhibited mTORC1 signaling and also appeared to slightly enhance the phosphorylation of eIF2α (Fig. 3H). Together, these data suggest that mTORC1-eIF2 cross-talk signaling is a general mechanism found in many different tissue or cell types and is functional in tumor cells.

In mouse embryonic fibroblasts (MEFs), inhibition of mTORC1 leads to the activation of GCN2 through the removal of an inhibitory phosphorylation site by PP6 (13). Depletion of PP6c with a specific siRNA in human p53+/+ MCF10A cells abolished both doxorubicin- and AZD8055-dependent phosphorylation of eIF2α without affecting mTOR inhibition, as inferred by the level of hypophosphorylated 4E-BP1 (fig. S4A), and suggested that PP6 may mediate the signaling response downstream of mTOR in response to DNA damage, consistent with other studies (13). However, it is challenging to assign a precise role to PP6 in this pathway because of its many substrates (37) and because PP6c knockdown also inhibited global protein synthesis rates under control conditions (fig. S4B).

mTOR-induced eIF2α phosphorylation is mediated by GCN2 and PERK

Having established that DNA damage–dependent eIF2α phosphorylation requires mTORC1 inhibition, it was important to determine the identity of the relevant eIF2K. We therefore used specific siRNAs to deplete the predominant eIF2Ks in MCF10A cells: PERK, GCN2, and PKR (Fig. 4A). AZD8055-induced eIF2α phosphorylation was diminished after depletion of either GCN2 or PERK but not PKR (Fig. 4, B and C). Moreover, combined reduction of GCN2 and PERK resulted in the greatest reduction in eIF2α phosphorylation after treatment with AZD8055, but the simultaneous depletion of GCN2, PERK, and PKR did not impair the phosphorylation of eIF2α further (Fig. 4, D and E). These data are consistent with a previous study (13) showing that GCN2 enhances the phosphorylation of eIF2α after prolonged mTOR inhibition. Doxorubicin-induced eIF2α phosphorylation was diminished after depletion of only GCN2 (Fig. 4, F and G) and was also decreased by combined depletion of PERK and GCN2 (Fig. 4, H and I). In this case, the reduction of eIF2α phosphorylation by AZD8055 and doxorubicin was comparable, suggesting that both AZD8055 and doxorubicin promote cross-talk signaling through a common mechanism.

Fig. 4 Doxorubicin-induced eIF2α phosphorylation was mediated by the eIF2Ks GCN2 and PERK.

(A) Schematic representation of the phosphorylation of eIF2α in response to the activation of the eIF2Ks PERK, GCN2, and PKR in response to the indicated stimulus. (B and C) p53+/+ MCF10A cells were transfected with individual siRNAs directed against the eIF2Ks PERK (1 nM), PKR (1 nM), or GCN2 (2 nM). Forty-eight hours after transfection, cells were treated with AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies (B). Blots shown were representative of three independent experiments. Quantification of eIF2α phosphorylation (Ser51) (C) from (B). Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05, by unpaired Student’s t test. (D and E) p53+/+ MCF10A cells were transfected with combinations of siRNAs directed against the eIF2Ks PERK (1 nM), PKR (1 nM), or GCN2 (2 nM). Forty-eight hours after transfection, cells were treated with AZD8055 (100 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies (D). Blots shown were representative of three independent experiments. Quantification of eIF2α phosphorylation (Ser51) (E) from (D). Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05 and **P ≤ 0.01, by unpaired Student’s t test. (F and G) p53+/+ MCF10A cells were transfected with individual siRNAs directed against the eIF2Ks PERK (1 nM), PKR (1 nM), or GCN2 (2 nM). Forty-eight hours after transfection, cells were treated with doxorubicin (500 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies (F). Blots shown were representative of three independent experiments. Quantification of eIF2α phosphorylation (Ser51) (G) from (F). Error bars represent means ± SD (n = 3 independent experiments), and significance was calculated by unpaired Student’s t test. (H and I) p53+/+ MCF10A cells were transfected with combinations of siRNAs directed against the eIF2Ks PERK (1 nM), PKR (1 nM), or GCN2 (2 nM). Forty-eight hours after transfection, cells were treated with doxorubicin (500 nM) for 16 hours. Cells were lysed and analyzed by immunoblotting with the indicated antibodies (H). Blots shown were representative of three independent experiments. Quantification of eIF2α phosphorylation (Ser51) (I) from (H). Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05 and **P ≤ 0.01, by unpaired Student’s t test.

Cross-talk between mTOR and eIF2 is required for cell migration

Inhibitors that target the eIF2 axis in cancers, either as a single agent or as combination therapies, are currently under development. However, the function of eIF2α phosphorylation in protein synthesis regulation is complex and dependent on cell context. Thus, in advanced prostate cancer, eIF2α phosphorylation with loss of phosphatase and tensin homolog (PTEN) is associated with poor patient survival (23), whereas in malignant melanoma, it is associated with resistance to programmed cell death–1 (PD-1) immunotherapy and increased cell migration (22). We therefore addressed how the cross-talk between eIF2 and mTOR pathways after treatment with the cytotoxic chemotherapeutic doxorubicin could affect cell migration.

Trazodone is an antidepressant that has been repurposed to treat neurodegeneration in mice (38). Similar to ISRIB, trazodone relieves translational repression mediated downstream of eIF2α phosphorylation. However, whereas ISRIB functions by enhancing eIF2B activity, trazodone has been suggested to act more directly to minimize the reduction of ternary complex in response to eIF2α phosphorylation (Fig. 5A) (38). Therefore, we used both chemical compounds to determine the role of eIF2α phosphorylation in the regulation of cell migration in a wound-healing assay. Treatment of MCF10A p53+/+ cells with doxorubicin caused cell cycle arrest at both the G1-S and G2-M checkpoints (Fig. 5B and fig. S5A), and eIF2α phosphorylation was maintained after treatment with doxorubicin for up to 72 hours (fig. S5B). Therefore, MCF10A p53+/+ cells were treated for 24 hours with doxorubicin in combination with ISRIB or trazodone before wounding. Untreated cells started to migrate back into the wound area within 48 hours (Fig. 5C), and the migration of doxorubicin-treated cells was ~70% slower compared to that of untreated cells (Fig. 5, C and D). However, consistent with a role for eIF2 signaling in migration after DNA damage, cells treated with doxorubicin and ISRIB showed a greater inhibition of migration into the wound (reduced by ~85%) compared to untreated cells (Fig. 5, C and D). Moreover, combining doxorubicin and trazodone almost completely abolished cell migration and significantly reduced migration compared to treatment with doxorubicin alone (Fig. 5, C and D). Doxorubicin induced a small but significant amount of cell death after 72 hours (Fig. 5E); however, this small amount of death was unlikely to be sufficient to induce the observed changes in migration. Moreover, doxorubicin-induced cell cycle arrest was maintained after 72 hours in the presence of both ISRIB and trazodone (Fig. 5F), suggesting that observed changes in migration were not dependent on cell cycle regulation.

Fig. 5 Doxorubicin-induced mTOR-eIF2 cross-talk signaling mediates cell migration.

(A) Schematic representation of mTOR-eIF2 cross-talk signaling and the alleviation of eIF2α phosphorylation–dependent translational repression. (B) Cell cycle analysis of wild-type p53 MCF10A cells (p53+/+) treated with doxorubicin (500 nM) for the indicated times. Cells were incubated with EdU (10 μM) for the final 1.5 hours of treatment (to quantify S phase cells), stained with FxCycle violet dye (to quantify cells within G1 and G2 phases), and analyzed by flow cytometry. Data values were an average of three independent experiments, and error bars represent means ± SD. (C) Wound-healing assay after treatment with doxorubicin (500 nM), ISRIB (200 nM), or trazodone (TRZ) (50 μM) for 24 hours. The wound was made with a p200 pipette tip (0 hours), and migration into the wound was analyzed by microscopy after 48 hours. Dashed lines represent the area of the initial wound, and the scale bar represents 500 μM. Images are representative from three independent experiments. (D) Quantification of cell migration into the wounded area from (C) using ImageJ. All values are displayed relative to the migration of the untreated sample. Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05 and **P ≤ 0.01, by unpaired Student’s t test. (E) Quantification of cell death using annexin V–FITC and Draq7 staining of MCF10A p53+/+ cells treated with a combination of doxorubicin (500 nM), ISRIB (200 nM), or trazodone (50 μM) for 72 hours. Error bars represent means ± SD (n = 3 independent experiments); ***P ≤ 0.001 and *P ≤ 0.05, by unpaired Student’s t test. (F) Quantification of cell cycle state using FxCycle staining in MCF10A p53+/+ cells treated with a combination of doxorubicin (500 nM), ISRIB (200 nM), or trazodone (50 μM) for 72 hours. Error bars represent means ± SD (n = 3 independent experiments). (G) Cell migration profiles from the xCELLigence RTCA DP instrument. MCF10A cells were treated for 24 hours with doxorubicin (500 nM) or trazodone (50 μM) before seeding in the upper well of the migration CIM-16 plate, and migration was monitored for 48 hours. Medium containing only 0.1% horse serum (HS) was used as a negative control. Data values were an average of three independent experiments, and error bars represent means ± SD. (H) Quantification of cell migration relative to the untreated sample from (G) at 48 hours. Error bars represent means ± SD (n = 3 independent experiments); **P ≤ 0.01 and ***P ≤ 0.001, by unpaired Student’s t test. (I) Cell migration profiles from the xCELLigence RTCA DP instrument. A549 cells were treated for 24 hours with doxorubicin (500 nM) or trazodone (50 μM) before seeding in the upper well of the migration CIM-16 plate, and migration was monitored for 48 hours. Medium containing only 0.1% FBS was used as a negative control. Data values were an average of three independent experiments, and error bars represent means ± SD. (J) Quantification of cell migration relative to the untreated sample from (I) at 48 hours. Error bars represent means ± SD (n = 3 independent experiments); *P ≤ 0.05 and ***P ≤ 0.001, by unpaired Student’s t test.

Because the inhibition of wound healing was greater using a combination of doxorubicin with trazodone and because trazodone is already U.S. Food and Drug Administration approved with more clinical potential than ISRIB, the migration of cells treated with doxorubicin and trazodone was quantified. MCF10A p53+/+ cells were treated with doxorubicin in combination with trazodone for 24 hours before seeding and migration was quantified for 48 hours. Consistent with the data obtained from the wound-healing assay, doxorubicin significantly reduced cell migration compared to untreated cells (Fig. 5, G and H), and cotreatment of doxorubicin with trazodone again significantly reduced cell migration compared to cells treated with only doxorubicin (Fig. 5H).

Because doxorubicin-induced mTOR-eIF2 cross-talk signaling also occurred in A549 cells (Fig. 3, D and E) and because eIF2α phosphorylation was maintained up to 72 hours after treatment (fig. S5C), the rate of A549 cell migration was quantified. A549 cells treated with doxorubicin showed a greater capacity to migrate than MCF10A cells, albeit to a lesser extent than untreated cells (Fig. 5, I and J). In support of the data obtained with MCF10A cells, cotreating A549 cells with doxorubicin and trazodone significantly inhibited cell migration (Fig. 5, I and J). A549 cells were more sensitive to doxorubicin-induced cell death at 72 hours (fig. S5D); however, the nonsignificant difference in cell death between cells treated with doxorubicin and cotreatment with trazodone was also unlikely to reduce cell migration. Moreover, cell cycle arrest was maintained in A549 cells treated with doxorubicin and cotreatment with trazodone for up to 72 hours (fig. S5, E and F), suggesting that changes in migration did not depend on differential cell cycle regulation. Together, these findings indicate that eIF2α phosphorylation promotes cell migration in response to doxorubicin, perhaps through translational reprogramming that may enable tumor cell adaptation. Thus, these results have implications for survival rate of patients with certain types of cancers and could provide previously unidentified therapeutic avenues to limit cancer migration and metastasis.

DISCUSSION

As part of the response to cell stress, eIF4F complex and ternary complex formation are coordinated to permit reprogramming of the translatome (12), a process that is frequently used by tumor cells to promote survival programs after exposure to cytotoxic chemotherapeutics. mTOR inhibitors including rapamycin, temsirolimus, and everolimus have been used clinically to block translation reprogramming mediated through eIF4F; however, in comparison, little work has been carried out to assess the modulation of ternary complex formation. Here, we described a regulatory pathway downstream of p53, in which cross-talk through mTOR to eIF2α phosphorylation after DNA damage was required for cell migration (Fig. 6). Moreover, blocking this pathway by chemical intervention limited cell migration, supporting the notion that therapeutic targeting of eIF2α signaling to minimize tumor cell metastasis and invasion could be of therapeutic benefit to specific groups of patients with cancer (Fig. 6).

Fig. 6 A model of doxorubicin-induced mTOR-eIF2K cross-talk signaling.

Doxorubicin-induced DNA damage leads to the activation of p53 and subsequent inhibition of mTOR through TSC-independent mechanisms, most likely resulting in the inhibition of global protein synthesis. Inhibition of mTOR activity directly promotes the activation of both GCN2 and PERK, possibly in a PP6-dependent manner, to enhance the phosphorylation of eIF2α and decrease ternary complex availability. Moreover, cross-talk signaling–mediating eIF2α phosphorylation is required for cell migration. Our data suggest that in a tumor setting, this signaling pathway could regulate cell adaptation to promote invasion and metastasis.

In comparison to other studies, such as UVB-induced DNA damage (9, 10), we showed that there was an extensive delay in the reduction of protein synthesis rates after doxorubicin treatment and an increase in eIF2α phosphorylation. In addition, our data showed that the shutdown of mRNA translation occurred through the inhibition of mTORC1, which partially depended on transcriptional pathways induced by p53. In MEFs, p53-dependent Sestrin2 expression regulates the inhibition of mTORC1 signaling and dephosphorylation of 4E-BPs in response to DNA damage induced by etoposide and cisplatin (39). Moreover, depletion of Sestrin1 and Sestrin2 in MCF10A cells abolishes ionizing radiation (IR)–mediated protein synthesis inhibition (40). Here, Sestrin2 expression was enhanced after treatment with doxorubicin but abolished after knockout or depletion of p53, thereby correlating with the relief of mTOR inhibition (Fig. 2, B and E). Moreover, doxorubicin-induced mTORC1 inhibition was independent of TSC, which inhibits mTORC1 by regulating the activity of Ras homolog enriched in brain (RHEB). These data are consistent with the notion that doxorubicin could inhibit mTOR signaling through the expression of Sestrin proteins, which can regulate the translocation of mTOR to the lysosome for activation (41), a possibility that requires further investigation.

Our results demonstrated that doxorubicin-dependent cross-talk signaling from mTOR to eIF2 required GCN2 and PERK and suggest a role for PP6 in this process (Fig. 6). These data are broadly consistent with previous studies, albeit using different inducers of cell stress. For example, catalytic inhibition of mTORC1 increases the activity of PP6, which, through the removal of an inhibitory phosphorylation site on GCN2, enhances eIF2α phosphorylation and induces autophagy (13). Mutations of PP6c that promote its stabilization are linked to therapeutic resistance in melanoma and also induce autophagy (37). Moreover, in melanoma, cell stress caused by the tumor microenvironment promotes an evolutionarily conserved starvation response requiring eIF2α phosphorylation, which results in cell migration and invasion (22). Our data showed that cross-talk signaling after doxorubicin-induced DNA damage resulted in cell migration and that the inhibition of this pathway with either ISRIB or trazodone blocks cell migration, suggesting that these compounds would be of therapeutic benefit to block metastasis.

When treating patients with chemotherapeutics that inhibit protein synthesis, it is essential to understand the degree of cross-talk between mTOR and eIF2K signaling. Moreover, it is also important to examine how cell fate is affected by combination therapies, which include the inhibition of the distinct arms of pathways that signal to protein synthesis. The data obtained thus far suggest that the degree of cross-talk from mTOR to eIF2K signaling depends on various factors including the DNA-damaging agent used and the acquired mutation status of the tumor. Our data suggest that combination therapy, including an agent that blocks cross-talk signaling from mTOR to eIF2α phosphorylation, could potentially reduce migration and metastasis. However, there needs to be careful patient stratification in terms of the DNA-damaging agent used and an understanding of how these signaling pathways are activated in the tumor samples to ensure effectiveness of such treatments.

MATERIALS AND METHODS

Cell culture

MCF10A cells are a nontransformed human breast epithelial cell line. MCF10A cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (1:1) supplemented with horse serum (5%), recombinant human epidermal growth factor (20 ng/ml), human insulin (10 μg/ml), hydrocortisone (500 ng/ml), and cholera toxin (10 ng/ml). p53 wild-type MCF10A (p53+/+) cells were obtained from the American Type Culture Collection (ATCC), and p53 knockout MCF10A (p53−/−) cells were obtained from Sigma-Aldrich. A549, HeLa, and BT474 cell lines were obtained from the ATCC and cultured in DMEM supplemented with 10% fetal bovine serum (FBS).

Treatment of cells

DNA damage was induced by treating cells with doxorubicin (500 nM or 1 μM) (Sigma-Aldrich) for 0 to 48 hours. mTOR was catalytically inhibited by treating cells with AZD8055 (100 nM) (Selleckchem), Torin 1 (100 nM) [Cell Signaling Technology (CST)], or rapamycin (100 nM) (CST) for 0 to 24 hours. Cells were treated with thapsigargin (250 nM) (Sigma-Aldrich) for 1 hour to induce ER stress and treated with ISRIB (200 nM) (Sigma-Aldrich) or trazodone (50 μM) (Sigma-Aldrich) to reverse the effects eIF2α phosphorylation.

Cell transfections

Cells were transfected with siRNAs using Lipofectamine RNAiMAX reagent (Invitrogen) in six-well plates as per the manufacturer’s guidelines for 24 to 72 hours. ON-TARGETplus siRNAs were purchased from Dharmacon: GCN2 (LQ-005314-00-0002), PERK (LQ-004883-00-0002), PKR (LQ-003527-00-0002), PP6c (LQ-009935-00-0002), TSC2 (LQ-003029-00-0002), and nontargeting siRNA (D-001810-01-05). siRNA specific for p53 was purchased from Thermo Fisher Scientific (Silencer Select siRNA, catalog number 4390825).

Western blot analysis

Whole-cell extracts were prepared in lysis buffer [50 mM tris (pH 7.5), 150 mM sodium chloride, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 1× Roche protease inhibitor cocktail, and 1× Roche PhosSTOP phosphatase inhibitor cocktail], and protein concentration of each sample was quantified using a Pierce BCA protein assay kit (Thermo Fisher Scientific). Extracts were diluted in SDS loading buffer [50 mM tris (pH 6.8), 2% SDS, 10% glycerol, 0.1% bromophenol blue, and 50 mM dithiothreitol (DTT)], and total protein (20 to 30 μg) was separated according to mass using SDS–polyacrylamide gel electrophoresis. Proteins were transferred to polyvinylidene difluoride membrane and incubated with the appropriate antibody at the manufacturers’ recommended dilution. Primary antibodies used were eIF2α (CST, no. 9772), p-eIF2α (Ser51) (Abcam, no. 32157), p–4E-BP1 (Ser65) (CST, no. 9456), 4E-BP1 (CST, no. 9644), p–p70 S6K (Thr389) (CST, no. 9205), p70 S6K (CST, no. 2708), p-Akt (Ser473) (CST, no. 4058), Akt (CST, no. 9272), p53 (Dako, no. M7001), Sestrin2 (Abcam, ab178518), p-ATM (Ser1981) (Abcam, no. 81292), ATM (Abcam, no. 32420), p-Chk2 (Thr68) (CST, no. 2661), Chk2 (no. 2662), GCN2 (CST, no. 3302), PKR (CST, no. 12297), PERK (CST, no. 3192), ATF4 (CST, no. 11815), PP6c (Abcam, no. 131335), TSC2 (CST, no. 4308S), BiP (Proteintech, no. 11587-1-AP), C/EBP homologous protein (CHOP) (Proteintech, no. 15204-1-AP), and β-tubulin (CST, no. 2146).

For enhanced chemiluminesence (ECL) detection, membranes were incubated with horseradish peroxidase–conjugated α-mouse (Dako, no. P0447) or α-rabbit (GE Healthcare, no. NA934V) secondary antibodies. Signal was developed by incubating membranes in ECL Prime Solution for 5 min, and x-ray film was exposed to the membrane to visualize luminescence. For fluorescent detection, membranes were incubated with IRDye Light-labeled α-mouse (CST, no. 5257S) or α-rabbit (CST, no. 5366S) secondary antibodies. Fluorescent signal was detected using LI-COR Odyssey Imaging Systems (LI-COR Biosciences), and images were analyzed with LI-COR Image Studio software (package version 5.2.5). Quantification of signal from both ECL- and LI-COR–detected blots was carried out using LI-COR Image Studio. Signal for a phosphorylated protein was normalized to that for the total abundance of the corresponding protein.

Sucrose density centrifugation

Sucrose gradients were used to separate subpolysomal and polysomal ribosomes. Ten to 50% (w/v) sucrose gradients were prepared in gradient buffer [300 mM NaCl2, 15 mM MgCl2, 15 mM tris-HCl (pH 7.5), 1 mM DTT, and cycloheximide (0.1 mg/ml)]. Cells were seeded on a 15-cm dish, washed in phosphate-buffered saline (PBS)–cycloheximide (100 μg/ml), and scraped into lysis buffer [300 mM NaCl2, 15 mM MgCl2, 15 mM tris-HCl (pH 7.5), 1 mM DTT, 0.2 M sucrose, cycloheximide (0.1 mg/ml), 0.5% IGEPAL, and 5 μl of RNasin per 1 ml]. Lysates were incubated on ice for 3 min before pelleting cells at 1300g for 5 min. The supernatant was layered onto the gradient and centrifuged at 38,000 rpm (acceleration, 9; deceleration, 6) for 2 hours at 4°C using a Beckman Coulter ultracentrifuge. Gradients were fractionated using a gradient fractionation system (Presearch Ltd.), and fractions were collected at 1-min intervals using a Foxy Jr. collection system (Presearch Ltd.), at a flow rate of 1 ml/min. Absorbance was measured constantly at 254 nm using a UA-6 UV-visible detector (Presearch Ltd.). The relative rate of translation was estimated by calculating a ratio of polysomes/subpolysomes by measuring the area under the curve within the subpolysomes (fractions 1 to 5) and the polysomes (fractions 6 to 10).

Radioisotope incorporation

For [35S]-methionine incorporation, cells were incubated with radiolabeled [35S]-methionine for 30 min at normal cell culture conditions. Cells were washed with PBS and lysed with passive lysis buffer (Promega). For [3]H-uridine incorporation, cells were incubated with radiolabeled [3]H-uridine for 30 min at normal growth conditions. Cells were washed with PBS and lysed with whole-cell lysis buffer. Protein was precipitated using trichloroacetic acid at a final concentration of 25%, and protein was captured on glass fiber filter paper (GE Healthcare). Captured protein was washed with 70% industrial methylated sprit (IMS) and acetone before the addition of 2 ml of Ecoscint scintillation cocktail (National Diagnostics), and radioisotope incorporation was quantified using a Wallac WinSpectral 1414 liquid scintillation counter. Each sample was carried out in triplicate, and spectral counts per minute were normalized to the total amount of protein for each sample.

Flow cytometry

All treatments were carried out using a six-well plate, and all flow cytometry data were acquired using BD FACSAria II, BD FACSCanto II, and BD LSRFortessa (BD Biosciences). A total of 10,000 counts were acquired for each experimental condition, and all flow cytometry data were analyzed with FlowJo data analysis software (version 10.1) (FlowJo LLC, Ashland, USA).

Cell death was quantified by measuring annexin V–fluorescein isothiocyanate (FITC) binding to externalized phosphatidylserine (with a 488-nM laser) and Draq7 uptake in the cell (with a 561-nM laser). Cells were collected and washed in PBS before resuspension in annexin buffer (BD Biosciences) and incubation with annexin V and Draq7 for 20 min before analysis.

For cell cycle analysis, cells were incubated with the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU) (20 μM) for 90 min to quantify cells within S phase and FxCycle violet stain (4′,6-diamidino-2-phenylindole, dihydrochloride), which binds to double-stranded DNA, to determine populations of cells within G1 and G2 phases of the cell cycle. Cells were collected and fixed in ice-cold 70% ethanol. EdU incorporation within fixed cells was quantified using Click-iT EdU Flow Cytometry Assay Alexa Fluor 647 azide (Life Technologies), following the manufacturer’s protocol, and detected with a 640-nM laser. To additionally quantify the population of cells within G1 and G2 stages of the cell cycle, FxCycle violet stain (Invitrogen) was incubated with EdU-incorporated cells for 16 hours at 4°C and detected with a 405-nM laser.

Cell migration and scratch assay

Cell migration assays were performed in a standard CO2 incubator using the xCELLigence RTCA DP instrument (ACEA Biosciences) according to the manufacturer’s instructions. Briefly, 3 × 104 MCF10A cells and 2.5 × 104 A549 cells were seeded in the upper chamber of a 16-well migration plate (CIM-16 plate) in 100 μl of media containing only 0.1% serum. Cells migrate to a lower chamber containing 160 μl of supplemented media through a microporous membrane. Microelectrode sensors are embedded on the underside of the microporous membrane. As cells move across the microelectrodes into the lower chamber, they generate impedance measurements that enable label-free quantification of cell migration. Cells were treated with the indicated compounds for 24 hours before seeding, and migration was continuously monitored in real time for 48 hours.

For wound-healing assays, cells were grown to confluence and treated with the indicated compounds for 24 hours before a scratch was made with a p200 pipette tip. The migration of cells back into the wounded area was monitored using phase contrast optics and a Zeiss Axiovert 200 M microscope after 48 hours. A percentage of wound closure (42) was calculated by quantifying the area of the wounds at 0 and 48 hours using ImageJ.

Statistical analysis

All statistical analysis testing for significance used an unpaired Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001 were considered to be statistically significant from three independent experiments.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/612/eaaw6763/DC1

Fig. S1. Doxorubicin inhibited protein synthesis and induced cell death in MCF10A cells.

Fig. S2. Doxorubicin-induced mTORC1 inhibition was not dependent on TSC activity.

Fig. S3. Doxorubicin inhibits mTORC1 signaling.

Fig. S4. Knockdown of PP6 reduced protein synthesis rates.

Fig. S5. eIF2α-mediated cell migration was not caused by changes in cell death or cell cycle regulation.

REFERENCES AND NOTES

Acknowledgments: We thank L. Pinon and D. Read for assistance with flow cytometry and microscopy analysis. We also thank M. Pizzinga and M. Bushell for the critical reading of the manuscript. Funding: R.F.H. was supported by MRC studentship and Wellcome Trust (grant number 110071/Z/15/Z). A.E.W., T.A.A.P., and M.S. were supported by MRC Programme funding (MC_UP_A600_1023). Author contributions: R.F.H. performed the experiments. R.F.H., T.A.A.P., M.S., and A.E.W. designed the experiments and interpreted results. R.F.H. and A.E.W. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Signaling

Navigate This Article