Stress and IGF-I Differentially Control Cell Fate through Mammalian Target of Rapamycin (mTOR) and Retinoblastoma Protein (pRB)*

Significant discoveries have recently contributed to our knowledge of intracellular growth factor and nutrient signaling via mTOR (mammalian target of rapamycin). This signaling pathway is essential in cellular metabolism and cell survival by enhancing protein translation through phosphorylation of 4EBP-1 and p70S6K. Growth factors like insulin-like growth factor-I induce mTOR to prevent cell death during cellular stress. Agents targeting mTOR are of major interest as anticancer agents. We show here, using human breast cancer cells, that certain types of stress activate mTOR leading to 4E-BP1 and p70S6K phosphorylation. UV treatment increased phosphorylation of the translation inhibitor eIF2α, suggesting a potential mechanism for UV activation of Akt and mTOR. c-Myc, a survival protein regulated by cap-dependent protein translation, increased with IGF-I treatment, but this response was not inhibited by rapamycin. Additionally, UV treatment potently increased c-Myc degradation, which was reduced by co-treatment with the proteasomal inhibitor, MG-132. Together, these data suggest that protein translation does not strongly mediate cell survival in these models. In contrast, the phosphorylation status of retinoblastoma protein (pRB) was mediated by mTOR through its inhibitory effects on phosphatase activity. This effect was most notable during DNA damage and rapamycin treatment. Hypophosphorylated pRB was susceptible to inactivation by caspase-mediated cleavage, resulting in cell death. Reduction of pRB expression inhibited IGF-I survival effects. Our data support an important role of phosphatases and pRB in IGF-I/mTOR-mediated cell survival. These studies provide new directions in optimizing anticancer efficacy of mTOR inhibitors when used in combination with DNA-damaging agents.

Overexpression of tyrosine kinase receptors (TKRs) has long been appreciated to contribute to tumorigenesis and resistance to treatment. Receptor activation of insulin-like growth factor-I (IGF-I) 2 , insulin, PDGF (platelet-derived growth factor), and some ErbB receptors induce Akt activity via PI3K (phosphatidylinositol 3-kinase). The p110 catalytic subunit of PI3K stimulates the phosphorylation of PI(4,5)P 2 to PI(3,4,5)P 3 activating PDK1 (3-phosphoinositol-dependent kinase-1). PDK1 then enhances the activity of several kinases including Akt, PKC isoforms, SGK (serum and glucocorticoid-induced protein kinase), mTOR (mammalian target of rapamycin), and p70S6K. Loss of PTEN function in cancer cells leads to similar signaling events as activation of TKRs. Moreover, many of IGF-I-mediated functions in breast cancer cells, such as proliferation and survival, are thought to be conveyed through PI3K and Akt.
Given the Akt potency as a survival mediator, much attention has focused on how it conveys this response. Increased protein translation occurs via Akt and its downstream mTOR. mTOR contains an Akt phosphorylation site, but current evidence indicates that Akt induces mTOR activity indirectly by phosphorylating tuberous sclerosis 2 (TSC2) (1). Thus, Akt enhances mTOR activity by releasing the inhibitory effects of the TSC1-TSC2 complex on mTOR. The proteins RAPTOR (regulatory-associated protein of mTOR) and RICTOR (rapamycin-insensitive companion of mTOR), through the mTORC1 and mTORC2 protein complex, respectively, serve as scaffolding proteins for interaction among mTOR and its substrates, p70S6K and 4E-BP1. Both p70S6K and 4E-BP1 contain a TOS (TOR signaling) motif that allows them to bind to RAPTOR; this then recruits mTOR (2). Other components of this pathway induce negative feedback on Akt activity (reviewed in Ref. 3).
Unphosphorylated 4E-BP1 binds eIF4E and inhibits cap-dependent protein translation. mTOR regulates the hierarchical phosphorylation of 4E-BP1, releasing it from eIF4E. Once the 4E-BP1 binding is relieved, eIF4E enhances cap-dependent translation by forming a protein complex called eIF4F (containing eIF4G, eIF4E, and eIF4A proteins). eIF4F promotes translation of proteins like c-Myc, VEGF (vascular endothelial growth factor), IGF-II, cyclin D, and FGF (fibroblast growth factor). This is accomplished first by eIF4E binding to the 7-methyl guanosine cap at the 5Ј-mRNA terminus and subse-quent unwinding and scanning of mRNA. The importance of mTOR-and cap-dependent translation in conveying survival or death depends, in part, on its regulation of c-Myc and the presence or absence of growth factors (4,5). Other mTOR targets include inhibition of phosphatases, such as PP2A (6) or phosphorylation of proteins likes STAT3 and pRB (retinoblastoma protein) in a poorly defined fashion (7)(8)(9)(10)(11). Finally, p53 and p21 Cip1 have been presented as potential IGF-I-and mTORmediated targets of cell fate in response to DNA damage (12)(13)(14). During DNA damage, phosphorylation of p53 Ser-15 induces AMPK (AMP kinase). Activated AMPK down-regulates mTOR function by reducing TSC2s suppression of mTOR. With glucose deprivation, PP2A, which is inhibited by mTOR, later dephosphorylates p53 Ser-15 (12).
The mTORC1 complex, unlike mTORC2, is sensitive to the mTOR inhibitor rapamycin. Pharmacologic inhibitors of mTOR like rapamycin have provided insight to the importance of mTOR as a cancer therapeutic agent. Through inhibition of mTOR, rapamycin reduces p70S6K activity and cap-dependent translation by preventing phosphorylation of 4E-BP1. Newer rapamycin analogs have been developed for their ease of administration and are currently in cancer clinical trials (3).
Through our search for biologically important Akt substrates in breast cancer models, inhibition of mTOR shows significant impact on IGF-I-mediated survival. Our data presented herein provide insights into the lack of a role for protein translation on cell fate. In contrast, we show that the ability of IGF-I and/or mTOR to regulate cellular fate occurs through phosphatase activity on pRB. Understanding the signaling pathways by which growth factors mediate resistance to various treatment modalities will enhance our ability to improve patient outcomes by utilizing the most effective combinations of therapeutic agents.

EXPERIMENTAL PROCEDURES
Cell Culture-MCF7 and ZR75 cells were provided by C. Kent Osborne (Baylor College, Houston, TX). Both cell lines were maintained in IMEM without phenol red (Mediatech, Herndon, VA) and supplemented with 10% fetal bovine serum (Gemini, Calabasas, CA), antibiotics, glutamine, and insulin. shRNA-mediated knockdown of RB in MCF7 cells was accomplished as described by Bosco et al. (15). HEK 293 cells were cultured in Dulbecco's modified Eagle's medium (Mediatech) and was supplemented with 10% fetal bovine serum and antibiotics. In each experiment, cells were plated in full serum-containing media and cultured overnight at 5% CO 2 and 37°C. Cells were washed twice the next day in warm phosphate-buffered saline (Biofluids, Rockville, MD) followed by overnight culture in serum-free medium (SFM). On the third day, the cells were treated with UV (UV-C, 10 J/m 2 ) in a SpectroLinker UV linker 100 (Spectronics, Westbury, NY) with lids removed and/or IGF-I (50 ng/ml). Cells were harvested as described in figure legends. When applicable, pretreatments with LY294002 (Alexis Biochemicals), rapamycin (Sigma), and rottlerin (Calbiochem, San Diego, CA) were performed 40 min prior to stress treatments. Cells were pretreated with Z-IETD-FMK (BD Pharmingen, San Diego, CA) 30 min prior to stimulation with UV. MG-132 (Calbiochem) was added just prior to UV irradiation of cells.
Antisense Oligonucleotide Transfections-2Ј-Methoxyethylribose chimeric antisense phosphorothioate oligonucleotides (ASO) were synthesized on an automated DNA synthesizer (Applied Biosystems Model 380B) by ISIS Pharmaceuticals, Carlsbad, CA. Cells were transfected and treated in the same fashion described previously (16).
p70S6K in Vitro Assays-MCF7 cells were plated and serumstarved overnight. The following day, cells were treated and lysed as described. Cell lysates were subjected to in vitro kinase assays as described (16).
YO-PRO Assay-Cells were plated at 1.25 ϫ 10 6 /60 mm 2 dish for 24 h. Cells were then washed twice with phosphate-buffered saline and serum-starved for 36 h. Cells were then treated with SFM, doxorubicin, IGF-I, and/or rapamycin as indicated for 14 h. Cells were collected by trypsinizing, washing with phosphate-buffered saline ϩ 1% bovine serum albumin, then resuspending to a concentration of 1 ϫ 10 6 /ml in phosphate-buffered saline ϩ 0.1% bovine serum albumin, and treated with YO-PRO, Invitrogen at 100 M/1 ϫ 10 6 cells. Cell suspensions were analyzed using a Beckman-Coulter FC 500 flow cytometer, per manufacturer's recommendations.

RESULTS
To investigate the potential importance of mTOR in IGF-Iand/or Akt-mediated cell survival we treated serum-starved MCF7 cells with UV irradiation using IGF-I to rescue cells from caspase-mediated PARP cleavage. Cells were also pretreated with rapamycin (Rap), wortmannin (Wort, PI3K inhibitor), or LY294002 (LY, a PI3K inhibitor) prior to IGF-I or UV treatment. Fig. 1A shows that rapamycin, wortmannin, and LY all reversed IGF-I cytoprotection in UV-treated cells. However, in the absence of IGF-I, neither rapamycin, wortmannin, nor LY significantly enhanced PARP cleavage associated with UV treatment, suggesting that inhibition of mTOR by the agents specifically inhibits growth factor responses. To further validate our finding that rapamycin inhibits IGF-I survival in other cell lines, we tested if it reverses IGF-I survival in the ZR75 breast cancer cell line using the UV and the cytotoxic drug doxorubicin (Dox), a DNA-interchelating agent. Again, we observed that addition of IGF-I inhibited UV and doxorubicin induced PARP and cleaved caspase 3 cleavage in ZR75 cells, and rapamycin co-treatment inhibited the IGF-I survival response more strongly with the doxorubicin treatment (Fig. 1B). These data demonstrate that the effect of rapamycin is not unique to one cell line or treatment.
While rapamycin is recognized as a specific mTOR inhibitor, we validated that mTOR was a rapamycin target using ASOs to reduce mTOR expression. MCF7 cells were either mock-transfected or transfected with mTOR or control (nmer) ASOs. The following day, cells were treated with SFM (serum-free media), UV, or UV plus IGF-I. Fig. 1C shows that mTOR ASO transfection inhibits mTOR expression by ϳ50%, whereas mTOR expression is similar in both mock-transfected and nmer-transfected cells. UV treatment of cells induces PARP cleavage in all three groups. In contrast, IGF-I rescue from UV-induced PARP cleavage is inhibited only in the mTOR ASO-transfected cells. Together, these data support that mTOR mediates IGF-I prosurvival responses in MCF7 breast cancer cells. mTOR conveys much of its cell survival and proliferative responses through phosphorylation of 4E-BP1 and p70S6K. mTOR regulates the hierarchical phosphorylation of Thr-37/ 46, Ser-65, and Thr-70, which are required for the release of 4E-BP1 from eIF4E (17). The PI3K inhibitors LY294002 and wortmannin inhibit phosphorylation of Thr-70 and Thr-36/45 of 4E-BP1 relieving its inhibition of cap-dependent translation (18). 4E-BP1 phosphorylation can be visualized by higher mobility shifts, with sequential shifts noted as ␣, ␤, and ␥, or by using a phospho-Thr-37/46 4E-BP1 antibody.
We set out to determine if 4E-BP1 phosphorylation status correlated with IGF-I response and/or cell survival. We also evaluated if c-Myc abundance would correlate with these cellular responses because it has been shown as an important target of cap-dependent translation and mediates cell survival. Our hypothesis was that IGF-I would induce 4E-BP1 phosphorylation and increase c-Myc protein abundance. If this response is mediated by mTOR, then both LY and rapamycin should reverse these effects. Further, we have previously shown that PKC␦ binds to mTOR to mediate IGF-I effects on IRS-1 (insulin receptor substrate 1) (16). Thus, we also tested the effect of the PKC␦ inhibitor rottlerin (Rott). Finally, if the mTOR phosphorylation of 4E-BP1 enhances cell survival through c-Myc, then UV irradiation should result in 4E-BP1 hypophosphorylation and reduction of c-Myc protein.
To this end, MCF7 cells were serum-starved overnight, then cells were pretreated with the kinase inhibitors SH-5 (an Akt inhibitor), LY, rapamycin, or rottlerin (Rott) alone, as indicated. Kinase inhibitors were also used in combination with UV and/or IGF-I treatment. In Fig. 2A, we observed that serumstarved MCF7 cells (SFM) retain some higher mobility (phosphorylated) forms of 4E-BP1, but treatment with LY or rapamycin showed a clear reduction in the phosphorylated 4E-BP1 bands. SH-5 and rottlerin treatments showed a less pronounced effect on 4E-BP1 phosphorylation. Surprisingly, UV treatment did not lead to 4E-BP1 hypophosphorylation; instead it increased 4E-BP1 phosphorylation shifts. Only when MCF7 cells were co-treated with UV and LY or rapamycin was 4E-BP1 hypophosphorylated. eIF4E abundance was unchanged in all treatment groups. Assessment of c-Myc protein showed that UV irradiation substantially decreased c-Myc abundance. IGF-I treatment alone and in combination with UV treatment increased c-Myc but this response was not inhibited by either LY or rapamycin. PKC␦ inhibition, using rottlerin, further enhances c-Myc expression in the UV-and IGF-I-treated cells but this may be due to its ability to inhibit the UV stress-mediated response (data not shown). Thus, c-Myc levels did not correlate with cell survival in samples containing IGF-I with or without LY or rapamycin, indicating at least under these treatment conditions c-Myc is not highly regulated by PI3K or mTOR activity. Fig. 2B confirms that the pharmacologic agent SH-5 kinase inhibited Ser-473 Akt phosphorylation in these experiments.
The ability of UV irradiation to enhance the phosphorylation of mTOR targets like 4E-BP1 was somewhat unexpected and is not well-documented in the literature. In fact, others have shown that DNA damage inhibits mTOR activity (19). Therefore, we evaluated the same treatments with HEK293 cells. With these cells we observed very similar results where LY and rapamycin inhibited 4E-BP1 phosphorylation (shown by a reduction in the ␤ and ␥ forms of 4E-BP1 and phospho-4E-BP1) (Fig. 2C). UV irradiation increased the abundance of the ␤ form A, MCF-7 breast cancer cells were plated at 70% confluency and serum-starved overnight. As indicated, some samples were pretreated with rapamycin (Rap, 20 nM), LY294002 (LY, 50 M), or wortmannin (Wort, 100 nM) then treated with IGF-I (50 ng/ml) and/or UV (10J/m 2 ). Cell lysates were separated using SDS-PAGE and immunoblotted using a PARP primary antibody. B, ZR75 breast cancer cells were treated as described in A. Lysates were immunoblotted using PARP and cleaved caspase 3 primary antibodies. C, MCF-7 cells were transfected with mTOR AS, nmer, or mock-transfected and allowed to recover for 24 h. The cells were then serumstarved for an additional 16 h. Following serum-starvation, the transfected cells were treated with either UV (10J/m 2 ) Ϯ IGF-I (50 ng/ml) for 6 h. The lysates were subjected to SDS-PAGE and immunoblotted with anti-PARP, mTOR, or tubulin, as indicated. Quantification of mTOR indicated that mTOR ASO reduced mTOR expression ϳ50% (p ϭ 0.055, n ϭ 3).
of 4E-BP1, which was inhibited by LY or rapamycin co-treatment. LY and rapamycin also inhibited 4E-BP1 phosphorylation in the IGF-I-treated samples but not with combined IGF-I and UV treatment. To explore if the IGF-I-mediated responses that we observed in MCF7 cells could be generalized to other breast cancer cell lines, we tested the ability of LY, rapamycin, and/or rottlerin to inhibit EGF (epidermal growth factor) or IGF-I-mediated changes in 4E-BP1 phosphorylation in human ZR75 and T47D breast cancer cells. Indeed, both LY and rapamycin inhibited 4E-BP1 phosphorylation. In these two cell lines, rottlerin showed a stronger ability to reduce 4E-BP1 mobility shifts (Fig. 2D).
To evaluate the functional significance of 4E-BP1 phosphorylation by either IGF-I or UV, treated MCF7 cell lysates were incubated with 7-methyl-GTP-Sepharose beads as an assay for the ability of 4E-BP1 to suppress cap-dependent translation. Using this assay, both IGF-I and UV irradiation alone or in combination decreased 4E-BP1 binding compared with the serum-starved samples. We then tested if loss of binding between 4E-BP1 and eIF4E would enhance eIF4E and eIF4G binding in a treatment-dependent fashion. We observed only a minor increase in binding when co-immunoprecipitating eIF4E with eIF4G after IGF-I and UV treatments (Fig. 3A). Together, these data indicate that cap-dependent translation is not the mechanism by which mTOR mediates IGF-I survival in these cells.
We then evaluated p70S6K activity as another mTOR target that may mediate IGF-I survival. Similar to our results with 4E-BP1, both IGF-I and UV treatment enhanced phosphorylation of Thr-389 p70S6K (Fig. 3B). The Akt inhibitor SH-5 did not significantly alter IGF-I-mediated effects on p70S6K phosphorylation in these cells, indicating that mTOR activity is not directly enhanced by Akt. In contrast, pretreatment with either LY or rapamycin significantly inhibited UV-induced p70S6K phosphorylation, while rottlerin showed comparatively less inhibition of this response. To assure that phosphorylation of p70S6K correlated with increased activity, cell lysates were sub- 20 nM, or rottlerin (Rott) 10 M, as indicated, then exposed to IGF-I (50 ng/ml) Ϯ UV (10 J/m 2 ). Cells were harvested 3.5 h later. Total cell lysates were separated by SDS-PAGE and immunoblotted with primary antibodies as indicated. B, subconfluent MCF7 cells were serum-starved overnight then pretreated with SH-5 for 40 min (as indicated) followed by IGF-I (50 ng/ml) treatment. Cells were harvested 1 h later, and lysates were immunoblotted with pAkt (Ser-473) and tubulin primary antibodies. C, subconfluent T47D and ZR75 cells were serum-starved for 40 h then pretreated with LY 50 M, Rap 20 nM, or Rott 10 M, as indicated, then exposed to IGF-I (50 ng/ml). Cells were harvested 3.5-h later. Total cell lysates were separated by SDS-PAGE and immunoblotted with a primary 4E-BP1 antibody. Tubulin was then immunoblotted to evaluate equal loading of samples.

FIGURE 3. Both UV and IGF-I enhance mTOR-mediated responses on 4E-BP1 on cap-dependent translation and p70S6K activity.
A, MCF7 cells were treated in the same fashion as described in Fig. 1 but harvested 2 and 4 h after treatment. 300 g of total cell lysate were then incubated with 7-methyl GTP-Sepharose beads or immunoprecipitated with eIF4G antibody and protein G-agarose at 4°C, then washed twice with lysis buffer. Beads were boiled and proteins separated using SDS-PAGE. Membranes were then immunoblotted with primary antibodies to 4E-BP1 and eIF4E. B, cells were treated with treatments (as indicated) and harvested after 1 h. Cell lysates were then either separated using SDS-PAGE and immunoblotting using phospho-p70S6K (Thr-389) (n ϭ 3) or p70S6K was immunoprecipitated, washed, and kinase assays performed. Radioactive histone H2B was then quantified and graphed as fold increase over SFM. C, MCF7 cells were plated and serumstarved as in part A but cells were treated with taxol 0.02 M alone or in combination with Rap, Rott, LY, and/or IGF-I, as indicated.
jected to an in vitro p70S6K activity assay. These results correspond closely to those we obtained using the phospho-p70S6K antibody, confirming that UV treatment enhances p70S6K activity approximately 3-5-fold (one-way analysis of variance p Ͻ 0.0006; post-hoc Student's t test, p Ͻ 0.0079, n ϭ 3). Together, these data show that UV irradiation enhances mTOR activity via a PI3K-dependent pathway. Given the similar effects of UV and IGF-I on 4E-BP1 and p70SK, they are unlikely targets of IGF-I-and mTOR-mediated survival responses.
Finally, we tested if another cytotoxic treatment like taxol (paclitaxol), which interferes with microtubule depolymerization, increases 4E-BP1 phosphorylation. PARP cleavage was also evaluated to assess if 4E-BP1 phosphorylation status correlated with apoptosis. Taxol treatment of MCF7 cells enhances formation of the ␥ form of 4E-BP1 compared with the serum-starved control cells (Fig. 3C). Interestingly, co-treatment with IGF-I, LY, or rapamycin did not alter the 4E-BP1 phosphorylation state imposed by taxol treatment alone. How-ever, IGF-I is able to rescue MCF7 cells from taxol-induced cell death, measured by a reduction in the PARP cleavage band. Together, these data strongly indicate that 4E-BP1 is not the specific target of IGF-I/mTOR-mediated survival response and that stress treatment can enhance phosphorylation of 4E-BP1. These data further support that 4E-BP1 does not mediate survival in these models.
Recently, other investigators have shown that eIF2␣, an inhibitor of translation by reducing eIF2B, is induced by various types of cellular stress. Interestingly, caspase cleavage of PKR (dsRNA-activated protein kinase) leads to phosphorylation of eIF2␣ Ser-51. In turn, phosphorylation of eIF2␣ enhances PI3K and mTOR activity, leading to p70S6K and 4E-BP1 phosphorylation (20,21). To better understand the mechanism by which stress can induce mTOR in our model, we evaluated phosphorylation of eIF2␣ Ser-51 and Akt Ser-478 by IGF-I or UV treatment and the ability of rapamycin to affect these events. Cells were treated with IGF or UV with or without rapamycin and harvested in a time-dependent fashion. Fig. 4A shows that IGF-I had little effect on eIF2␣ phosphorylation, while rapamycin co-treatment slightly enhanced its phosphorylation at 4 h. On the other hand, UV treatment increased eIF2␣ phosphorylation as early as 30 min after treatment, and it further increased 4 h after treatment (ϳ10fold over SFM). Rapamycin co-treatment slightly enhanced the UV effect only at the 4-h time point. Thus, activation of eIF2␣ by UV treatment and rapamycin may inhibit protein translation despite the noted increases in mTOR activity and phosphorylation of 4E-BP1. We also tested if Akt is deferentially phosphorylated by these same treatments in a time-dependent fashion and to support the ability of phosphorylated eIF2␣ to enhance PI3K-dependent kinases. Akt was rapidly phosphorylated on Ser-473 by IGF-I treatment. This effect was sustained at 4 h and further increased by rapamycin co-treatment at that time point. UV treatment also enhanced phospho-Akt but this effect was only noted at the later time points. This is consistent with eIF2␣ activation occurring prior to activation of Akt. Rapamycin cotreatment did not lead to enhanced phosphorylation compared with UV treatment alone.
Because c-Myc is a well-recognized target of cap-dependent translation and mediates cell survival we evaluated c-Myc pro- , IGF-I (50 ng/ml), and/or rapamycin (20 nM), as indicated. Cells were harvested in a time-dependent fashion and immunoblotted using pEIF2␣ (Ser-51) or pAkt (Ser-473). Membranes were also probed with tubulin primary antibody to assess equal protein loading. B, cells were treated in the same fashion as described in A but instead, membranes were probed with c-Myc or PARP primary antibody. c-Myc bands were quantified and normalized to tubulin-loading controls. Normalized results were graphed using SFM sample as control. C, cells were treated in the same fashion as A using MG-132 (20 M) and/or UV treatment. Samples were analyzed as described in B using c-Myc and PARP primary antibodies for immunoblotting. OCTOBER 17, 2008 • VOLUME 283 • NUMBER 42 tein levels in a treatment-and time-dependent fashion. IGF-I treatment increased c-Myc protein greater than 2-fold over the time course studied. Surprisingly, rapamycin was unable to significantly inhibit the IGF-I-mediated increase in c-Myc protein.

IGF-I and mTOR Regulate Cell Survival through pRB
This effect was also shown in Fig. 2B. On the other hand, UV treatment led to notable decreases in c-Myc protein in a timedependent fashion (ϳ90% versus SFM control), which was unaffected by rapamycin (Fig. 4B).
Surprisingly, these data were more consistent with a change in the c-Myc half-life via degradation, rather than translation affecting cell fate. Previous work has shown that UV irradiation induces c-Myc proteasomal degradation (22). To evaluate if c-Myc protein is decreasing as a result of enhanced degradation, MCF7 cells were treated with the proteasome inhibitor MG-132 (MG) alone or in the presence of UV irradiation. Cells treated with MG-132 alone showed a greater than 4-fold increase in c-Myc levels over time (Fig. 4C). When cells were treated with MG-132 plus UV, MG-132 enhanced c-Myc levels higher than UV alone and SFM (until the 4-h time point). These data support that UV leads to c-Myc loss due to proteasomal degradation. However, even in the presence of MG-132 c-Myc levels decreased but this did not lead to significant changes in PARP cleavage. This may be consistent with either the inability of the proteasome inhibitor to fully rescue degradation at the concentration used (20 M) or the influence of another pathway affecting c-Myc abundance. Despite increases in c-Myc protein, cell survival was not significantly increased with MG-132 co-treatment of UV-treated cells. Of note, MG-132 can inhibit the proteasomal degradation of a variety of proteins. Therefore, its effect on cellular apoptosis does not specifically address the role of c-Myc under these conditions. We next evaluated pRB as a potential target of mTOR survival responses. pRB plays an important role in cell cycle progression and apoptosis. The regulation of pRB function is complex. pRB is inactivated primarily through phosphorylation by cyclin/cdk (cyclin-dependent kinase) complexes or caspasemediated cleavage (23). Often, growth factors enhance the phosphorylation of nine or more residues on pRB which inactivates its binding to E2F transcription factors, leading to cell cycle transit. Alternatively, cellular stress can lead to pRB cleavage at Asp-886 in the C terminus, resulting in a fragment 5-kDa smaller than the parental pRB (⌬CT pRB). An internal cleavage site has also been described resulting from caspase-3 and/or -7 cleavage at Asp-349, leading to p68 and p48 pRB fragments (24). We have previously shown that while caspase 3 is absent in MCF7 cells, caspase 8 can lead to caspase 7 cleavage resulting in substrate cleavage in a similar manner to cleave caspase 3 substrates (25), which may include pRB. In contrast, pRB can be activated by reduction in phosphorylation through PP2A and PP1 phosphatase activity.
For these studies, we again evaluated the ability of IGF-I to oppositely regulate pRB compared with cellular stress with doxorubicin. In a similar fashion to our earlier experiments, we used rapamycin to inhibit IGF-I-mediated survival in doxorubicin-treated cells. Fig. 5A shows that MCF7 cells treated with rapamycin had primarily hypophosphorylated pRB compared with serum-starved control cells. Doxorubicin treatment leads to hypophosphorylated and cleaved forms of pRB. Addition of IGF-I to doxorubicin-treated cells inhibited the cleaved form of pRB. In contrast, rapamycin reverses the IGF-I-mediated effects on pRB resulting in the predominance of hypophosphorylated and cleaved pRB forms. Fig. 5B further validates these findings when cells were treated with UV, IGF-I, and/or rapamycin in a time-dependent fashion. Unstressed MCF7 cells maintained abundant hyperphosphorylated pRB (as shown by the slower mobility protein species which represents numerous pRB phosphorylation sites). In contrast, UV-stressed MCF7 cells showed a time-and treatment-dependent change in pRB mobility. Notably, UV enhanced hypophosphorylation of pRB and co-treatment with rapamycin enhanced pRB cleavage. The presence of cleaved caspase 7 further supports the close correlation between pRB cleavage and cell death (as indicated by , and/or IGF-I (50 ng/ml), and harvested after 48 h. Cleared whole cell lysates were separated using SDS-PAGE. Immunoblotting was performed using a pRB antibody. B, serum-starved, subconfluent MCF7 cells were treated IGF-I (50 ng/ml), UV (10 J/m 2 ), or rapamycin (20 nM) in a time-dependent fashion. Cleared cell lysates were separated using SDS-PAGE and membranes were probed with pRB, PARP, caspase 7, or tubulin primary antibodies. C, as indicated, MCF7 cells were pretreated with the caspase 8 inhibitor Z-IETD-FMK (IETD) 10 M 30 min prior to UV irradiation. Rap and UV treatment were performed as described in Fig. 1. D, MCF-7 cells were pretreated with rapamycin Ϯ okadaic acid (100 nM). Cells were then exposed to IGF-I Ϯ UV, and harvested 8 h later. pRB, PARP, and tubulin proteins were evaluated in the same fashion as described previously. cleaved caspase 7). PARP cleavage showed the same pattern as cleaved caspase 7 (data not shown).
To ensure that the more rapidly migrating pRB band was indeed a caspase cleavage product, we used the caspase 8 peptide inhibitor IETD to prevent the production of the lower pRB band. Fig. 5C shows that cells treated with UV plus IETD did not produce the most rapidly migrating pRB band. This pattern of pRB cleavage was then compared with PARP cleavage where IETD co-treatment significantly reduced PARP cleavage as well. In the same experiment, rapamycin co-treatment slightly enhanced both pRB and PARP cleavage. Thus, the phosphorylation and cleavage status of pRB closely mirrors treatment effects on cell survival where IGF-I and UV convey diverse responses. Therefore, we pursued more detailed studies involving pRB functional status.
Some mTOR-mediated biological responses occur through its ability to inhibit PP2A phosphatase. Further, PP1 isoforms and PP2A dephosphorylate pRB in response to DNA damage (26). Others have shown that DNA-damaging agents enhance PP2A activity through the production of oxidative stress (27).
To study the role of mTOR and phosphatases on pRB function and cell survival we tested if the phosphatase inhibitor okadaic acid (OA) would reverse the rapamycin pro-apoptotic effect in UV-treated cells. As shown previously, UV treated cells showed PARP cleavage along with both hypophosphorylated and cleaved pRB (Fig. 5D). Again, rapamycin co-treatment enhanced PARP and pRB cleavage. Addition of okadaic acid to UV-treated cells inhibited both PARP and pRB cleavage, resulting in more hyperphosphorylated pRB compared with UV irradiation alone. Rapamycin inhibited okadaic acid cytoprotective effect, as shown by both increased PARP and pRB cleavage. Interestingly, the effects of both okadaic acid and rapamycin on pRB appear to depend upon the presence of UV, because they show little effect on pRB or PARP when used alone. These data are consistent with UV inducing phosphatase activity, which dephosphorylates pRB, and that rapamycin further enhances phosphatase activity on pRB. These data support that pRB is an important target by which rapamycin and DNA-damaging agents induce cell death.
We set out to determine if pRB directly mediates cell fate or is simply a target of growth factor signaling or caspase cleavage. For these experiments, we used MCF7 cells transfected with shRNA targeting RB (15). Fig. 6A shows that knockdown of pRB is highly efficient. These cells along with the empty vector transfected clone were exposed to doxorubicin, IGF-I and/or rapamycin treatment, as designated. In general, these cells differed slightly in their sensitivity to both doxorubicin and IGF-I and required shorter treatment durations than the cells used in the previous data herein. Fig. 6B shows that the control cells underwent apoptosis after exposure to doxorubicin, while addition of IGF-I reduces PARP cleavage, consistent with the survival effect observed previously. In contrast, shRB MCF7 cells were less protected by IGF-I when exposed to doxorubicin (post-hoc, Student's t test p ϭ 0.05, n ϭ 3). For this reason, the doxorubicin, IGF-I, and rapamycin combination treatment was eliminated in these sets of experiments. Instead, cells were treated with doxorubicin plus rapamycin to determine if rapamycin increases doxorubicin-mediated apoptosis in the absence of pRB. In this apoptosis assay, rapamycin co-treatment in shRB cells had no additional apoptotic effect and a difference in these treatments compared with control cells did not reach statistical significance (post-hoc, Student's t test p ϭ 0.20). For unknown reasons, we also observed that the doxorubicin and doxorubicin plus rapamycin treatment groups have less parental PARP. Overall, these data confirm that pRB itself is an important mediator of IGF-I/mTOR survival responses.
Other studies have previously proposed that p53 and p21 Cip1 are important targets for rapamycin response (12)(13)(14). DNAdamaging treatments enhance p53 function, in part, through phosphorylation of p53 Ser-15, which increases p53 stability and function. Enhanced p53 function then leads to increased p21 Cip1 expression. P53 has also been reported to suppress mTOR activity (19) suggesting that if expression of pRB influences p53 response this effect may alter IGF-I survival responses through mTOR. Therefore, we addressed if pRB status regulates p53 or p21 Cip1 responses to the various treatments. While p21 Cip1 levels are slightly higher in the doxorubicin-treated shRB MCF7 cells, the overall pattern of p53 or  OCTOBER 17, 2008 • VOLUME 283 • NUMBER 42 p21 Cip1 response is not significantly altered compared with the control MCF7s (Fig. 6B).

IGF-I and mTOR Regulate Cell Survival through pRB
Finally, we wanted to determine if the differences in the PARP parental (105 kDa) protein seen in Fig. 6B correlated with any changes in cellular fate using an independent apoptosis assay. We used the YO-PRO dye to detect early changes in membrane permeability of apoptotic cells. Fig. 6C shows similar results as the PARP cleavage assay. IGF-I provided control MCF7 cells with cytoprotection from doxorubicin-induced cell death, whereas the shRB MCF7 cells experienced less reduction in apoptosis when IGF-I is added along with doxorubicin (posthoc, Student's t test, p Ͻ 0.016, n ϭ 4). Using this assay, we did not observe an increase in apoptosis when rapamycin was added to doxorubicin treatment compared with doxorubicin treatment alone. This may suggest another target (other than pRB) for rapamycin-mediated cell death in the absence of growth factor signaling. However, IGF-I protective effects in these cells do appear to be dependent upon pRB.

DISCUSSION
Recently there has been significant interest in better understanding mTOR signaling and identifying the targets by which rapamycin analogs convey their anticancer activity to predict which tumors are likely to respond. mTOR targets that have received the most study include those that regulate protein translation, 4E-BP1 and p70S6K (reviewed in Ref. 28). However, these proteins or their targets, including c-Myc or cyclin D, may not be very useful biomarkers (29). We think the reason may be 2-fold. First, stress treatments may enhance mTOR regulation of 4E-BP1 and p70S6K but result in cell death rather than cell survival. Second, 4E-BP1 and p70S6K may not be the primary targets by which rapamycin analogs mediate their anticancer effects. If this is indeed the case, then their activation status will not predict the efficacy of mTOR inhibitors.
Beyond its major role in translation, a few studies have reported that 4E-BP1 conveys cellular stress responses while others have reported that DNA damage decreases phosphorylation of 4E-BP1 (30). Houghton and co-workers (31) recently reported that 4E-BP1 activates JNK (c-Jun N-terminal kinase, a stress-induced kinase). These investigators also reported that rapamycin preferentially induces cell death in rabdomyosarcoma cells lacking wild-type p53 by activation of ASK1 (apoptosis signal-regulating kinase 1) and JNK, whereas IGF-I treatment inhibits rapamycin-induced cell death independently of Akt. Moreover, they showed that rapamycin requires 4E-BP1 to induce apoptosis (32). Interestingly, in cells expressing p53, p21 Cip1 inhibited the ability of ASK1 to induce JNK. In trying to test these observations in our model, we observed stress-induced phosphorylation of 4E-BP1. Thus, our results seem to vary somewhat with those of Houghton and co-workers (31). Further, we observe that cells expressing wild-type p53 (MCF-7 breast cancer cells) are also sensitive to rapamycin-induced apoptosis after prolonged treatment (data not shown), and that IGF-I conveys its survival response through mTOR, which is reversed by rapamycin. We have not yet evaluated the importance of JNK in rapamycin-mediated apoptosis. With regard to the importance of 4E-BP1 in cell fate, both stress and IGF-I enhanced 4E-BP1 phosphorylation. Thus, we do not observe any correlation between 4E-BP1 phosphorylation and cell survival. This does not eliminate a role for protein translation affecting cell survival through other mechanisms. For example, we showed that UV, not IGF-I, induced the phosphorylation of eIF2␣. This event may induce PI3K-induced pathways such as Akt and mTOR in response to UV treatment but still inhibit protein synthesis through mechanisms other than phosphorylation of 4E-BP1.
c-Myc is thought to be a major contributor to mTOR-mediated survival via up-regulation of cap-dependent translation. However, c-Myc protein stability also influences cell fate. While our model shows that IGF-I increases c-Myc protein, rapamycin was unable to reverse the effect. In contrast, UV treatment led to c-Myc rapid proteasomal degradation. Together, our data do not support a role for the mTOR up-regulation of capdependent c-Myc protein translation to enhance cell survival. Whether phosphorylation of eIF2␣ inhibits c-Myc translation is a topic for future studies. Degradation of c-Myc via the ubiquitin pathway is also likely to play a significant effect on cell death.
Our data support an integral role for pRB as a mediator of mTOR/IGF-I survival in breast cancer cells. pRB plays an important role in cell cycle progression and apoptosis. In our studies, the phosphorylation status of pRB closely reflects cellular fate. pRB function is inactivated through phosphorylation. Mitogen stimulation of cells results in cyclin A/cdk2 and cyclin E/cdk2 phosphorylation of pRB and pRB disassociation from E2F isoforms (E2F1, -2, -3, and -4). Down-regulation of pRB phosphorylation is mediated through phosphatases like PP1 and PP2A, which increases its binding to E2F proteins. pRB is also a target of stress-induced caspases. Interestingly, the ⌬CT pRB fragment retains its E2F binding ability, whereas the p68 and p48 fragments cannot bind to E2F proteins. Our experiments show that rapamycin treatment reduces pRB phosphorylation particularly when combined with DNA-damaging agents, making it susceptible to caspase cleavage through unclear mechanisms.
We also observed that rapamycin can reverse the protective effect of okadaic acid (a nonspecific phosphatase inhibitor) indicating that PP1, PP2A, or another phosphatase targeting pRB may be a key factor in rapamycin response. Recently, Huang et al. (32) have reported that rapamycin inhibits protein phosphatase 5 activity on ASK1, supporting the idea that rapamycin enhances cell death through inhibition of one or more phosphatases. Kong et al. (33) also reported that PP2A dephosphorylates Ser-15 on p53. It is possible that mTOR suppression of PP2A prolongs p53 function. If rapamycin inhibits this response then apoptosis or a lack of DNA damage repair may ensue. Phosphatases can regulate specific phosphorylation sites on pRB as suggested by others during cellular stress (34,35). Thus, many of the mTOR targets of cell survival are mediated by phosphatases supporting a common mechanism targeting important regulators of apoptosis such as ASK1, p53, and pRB.
The loss of pRB function in tumorigenesis has been appreciated for many years. Interestingly, recent pRB literature indicates that loss of pRB function enhances the efficacy of several types of anticancer agents including tamoxifen, microtubuleinterfering agents, and DNA-damaging agents (15,36,37) in breast, lung, and prostate cancer models. Our data are consistent with these findings and further support the concept that rapamycin can further regulate pRB function to offset growth factor-mediated cell survival.
Given the significant interest in using rapamycin analogs for the treatment of various cancers and the need to predict, which tumors will respond to the agents through the use of biomarkers, our findings provide key information. First, we show that the phosphorylation status of p70S6K may not be a reliable biomarker for rapamycin efficacy because both UV and anisomycin stress treatments can also induce p70S6K phosphorylation but have opposite effects on cellular fate (data within and Ref. 16). Secondly, we suggest that 4E-BP1 is not the key mediator of the rapamycin ability to inhibit growth factor responses and cause tumor cell death. Both of these findings are based on the fact that mTOR activity can be enhanced by cell stress. We believe that our data will direct attention to new rapamycin targets such as pRB and phosphatases in mediating its antitumor functions. Finally, given that our data show pRB is the primary target of rapamycin, then combining rapamycin with specific classes of anticancer agents should enhance treatment responses.