Research ArticleMOLECULAR BIOLOGY

AKT Promotes rRNA Synthesis and Cooperates with c-MYC to Stimulate Ribosome Biogenesis in Cancer

See allHide authors and affiliations

Science Signaling  30 Aug 2011:
Vol. 4, Issue 188, pp. ra56
DOI: 10.1126/scisignal.2001754

Abstract

Precise regulation of ribosome biogenesis is fundamental to maintain normal cell growth and proliferation, and accelerated ribosome biogenesis is associated with malignant transformation. Here, we show that the kinase AKT regulates ribosome biogenesis at multiple levels to promote ribosomal RNA (rRNA) synthesis. Transcription elongation by RNA polymerase I, which synthesizes rRNA, required continuous AKT-dependent signaling, an effect independent of AKT’s role in activating the translation-promoting complex mTORC1 (mammalian target of rapamycin complex 1). Sustained inhibition of AKT and mTORC1 cooperated to reduce rRNA synthesis and ribosome biogenesis by additionally limiting RNA polymerase I loading and pre-rRNA processing. In the absence of growth factors, constitutively active AKT increased synthesis of rRNA, ribosome biogenesis, and cell growth. Furthermore, AKT cooperated with the transcription factor c-MYC to synergistically activate rRNA synthesis and ribosome biogenesis, defining a network involving AKT, mTORC1, and c-MYC as a master controller of cell growth. Maximal activation of c-MYC–dependent rRNA synthesis in lymphoma cells required AKT activity. Moreover, inhibition of AKT-dependent rRNA transcription was associated with increased lymphoma cell death by apoptosis. These data indicate that decreased ribosome biogenesis is likely to be a fundamental component of the therapeutic response to AKT inhibitors in cancer.

Introduction

Increased protein synthesis and cell growth are critical for tumorigenesis (1, 2). Tumor cells are generally larger than normal cells and contain more and bigger nucleoli, a phenotype indicative of increased ribosome synthesis (3). Furthermore, enlarged nucleoli are predictive of tumor aggressiveness (4, 5). Ribosome synthesis requires the coordinated actions of all three DNA-dependent RNA polymerases (Pol I, II, and III). Transcription of the ~200 copies of the 45S ribosomal RNA (rRNA) genes [rDNA (ribosomal DNA)] by Pol I ultimately gives rise to 18S, 5.8S, and 28S rRNAs after extensive processing during ribosome assembly (68). These mature rRNAs, along with the 5S rRNA transcribed by Pol III, form the nucleic acid backbone of the ribosome. Transcription of these rRNAs can constitute up to 60% of cellular RNA synthesis and is generally considered to be a major limiting step for cellular growth (6). Synthesis of the ribosomal proteins, the other major constituent of functional ribosomes, requires Pol II–mediated transcription. Assembly of the ribosomal proteins with rRNA and its associated processing requires many factors to produce functional ribosomes in the cytoplasm (9). Genetic mutations in key components of the ribosome biogenesis machinery, such as ribosomal proteins S17 and S14, are associated with increased cancer susceptibility (10, 11), and deregulation of key signaling pathway molecules involved in modulation of protein translation is associated with tumorigenesis (12, 13). Three of the major signaling networks that regulate cell growth—PI3K (phosphatidylinositol 3-kinase)–AKT–mTORC1 (mammalian target of rapamycin complex 1), c-MYC, and RAS–ERK (extracellular signal–regulated kinase)—are each deregulated in a large percentage of primary tumors (1416), suggesting that dysregulation of ribosome biogenesis and protein translation may be a common feature in malignant transformation. Indeed, it is becoming increasingly apparent that the downstream actions of these growth pathways on the regulation of ribosome biogenesis and protein translation are essential for their oncogenic effects (1719).

The PI3K-AKT-mTORC1 signaling network plays an essential role in malignant transformation (14, 2023). Gain of function of this pathway has been demonstrated in various cancers, and multiple components of this pathway are either oncogenes or tumor suppressors (14, 2426). Most studies examining the mechanisms by which this signaling network contributes to malignancy have focused on the processes of resistance to cell death, cell cycle progression, angiogenesis, or metabolism (27). However, mTORC1 also coordinates the synthesis of both ribosomal proteins and rRNA to modulate ribosome biogenesis (2831). These observations thus indicate that control of ribosome biogenesis through the PI3K-AKT-mTORC1 pathway is a potential alternative requirement for malignant transformation.

The PI3K-AKT-mTORC1 pathway interacts with c-MYC, which is dysregulated in 15 to 20% of human malignancies by amplification or translocation (22, 23). Part of the mechanism by which c-MYC drives malignancies is also dependent on its ability to promote ribosome biogenesis (17, 18, 3234). The mechanisms by which individual members of the PI3K-AKT-mTORC1 pathway drive ribosome biogenesis, how they interact with c-MYC, and how their dysregulation in the context of ribosome biogenesis contributes to malignancy remain unknown. Here, we show that AKT was required for Pol I–driven rDNA transcription, independent of its ability to activate mTORC1. Furthermore, AKT could increase rDNA transcription and growth in the absence of growth factors to a similar extent as c-MYC. Moreover, AKT and c-MYC cooperated to maximally activate rDNA transcription and cell growth, suggesting that they function through distinct but complementary pathways. AKT activity is increased in more than 50% of B cell lymphomas (23), and in a c-MYC–driven model of B cell lymphoma, we demonstrated that AKT signaling was necessary for ribosome biogenesis and cell proliferation. Together, these data indicate that the PI3K-AKT-mTORC1 pathway and c-MYC can cooperate during malignancy to ensure that ribosome biogenesis remains constitutively activated in tumor cells. Our data suggest that inhibition of ribosome biogenesis may constitute a primary therapeutic effect of targeting PI3K-AKT signaling in cancers, particularly those displaying aberrant signaling through growth pathways such as PI3K-AKT and c-MYC.

Results

AKT signaling is necessary for optimal rates of rDNA transcription in exponentially growing cells

To determine the role of AKT in the regulation of ribosome biogenesis and cell growth, we examined the effect of AKT inhibitor 1/2 (AKTi-1/2), a non–ATP (adenosine 5′-triphosphate)–competitive allosteric inhibitor that interacts with the pleckstrin homology domain of AKT (35), on transcription of the 45S ribosomal gene (rDNA), which encodes the 18S, 5.8S, and 28S rRNAs. To determine the effects on rDNA transcription, we measured the abundance of the 5′ external transcribed spacer (5′ETS) pre-rRNA. rRNA is transcribed by Pol I as the 45S precursor, termed pre-rRNA, which is rapidly processed at the externally transcribed spacer (ETS) and internally transcribed spacer (ITS) regions to form the mature rRNAs. Whereas the mature rRNAs have half-lives of hours to days, the processed ETS and ITS have half-lives of minutes; thus, their abundance can be used as a measure of rDNA transcription rates (28, 32). Treatment of exponentially growing human embryonic kidney (HEK) 293 cells with AKTi-1/2 for 30 min blocked AKT activation (as assessed by phosphorylation of Ser473 in AKT) (Fig. 1A and fig. S1) and decreased the abundance of 5′ETS pre-rRNA (Fig. 1B) with minimal effect on mTORC1 activity [as assessed by phosphorylation of Ser240/244 in ribosomal protein S6 (rpS6), a substrate of the mTORC1-dependent rpS6 kinase 1] (Fig. 1A). Conversely, 30-min treatment with rapamycin significantly inhibited mTORC1 activity (Fig. 1A) but did not affect the abundance of 5′ETS pre-rRNA (Fig. 1B). Thus, AKT signaling is required for optimal pre-rRNA synthesis rates in exponentially growing HEK293 cells independently of mTORC1 signaling.

Fig. 1

AKT is necessary for sustained rDNA transcription and rRNA synthesis. Exponentially growing HEK293 cells were treated with 20 nM rapamycin or 5 μM AKTi-1/2. (A) Western blotting with the indicated antibodies. Representative of n = 3 blots. Quantitation in table S1. (B) RPA for 5′ETS pre-rRNA abundance. n = 3 sets of cells. (C) Pulse labeling (left) and quantitation (right). Quantitation of 45S rRNA synthesis and processing in table S2. Representative of n = 2 to 3 sets of cells. (D) qChIP analysis to assess RNA Pol I loading on the 5′ETS and 5.8S region of the 45S rDNA gene. n = 3 sets of cells. *P < 0.05, **P < 0.01, ***P < 0.001 compared to the time-matched control (untreated).

We also examined the effect of sustained treatment with AKTi-1/2 and rapamycin on the abundance of 5′ETS pre-rRNA. The reduced abundance of 5′ETS pre-rRNA after 30 min of AKTi-1/2 treatment was maintained at 3 and 12 hours (Fig. 1B). In contrast, mTORC1 inhibition had no effect on the abundance of 5′ETS transcription at 30 min, but significantly reduced 5′ETS transcription at 3 and 12 hours of treatment (to a similar extent as AKTi-1/2). The repression in response to combined treatment with AKTi-1/2 and rapamycin for 12 hours was greater than with treatment with either agent alone (Fig. 1B). The distinct patterns of temporal response for AKTi-1/2 compared to rapamycin, together with the greater reduction in 5′ETS pre-rRNA abundance by AKTi-1/2 and rapamycin when used in combination, indicate that inhibition of AKT decreases the abundance of 5′ETS pre-rRNA by mTORC1-dependent and -independent signaling mechanisms.

To directly measure synthesis rates of 45S pre-rRNA, we performed [32P]orthophosphate pulse labeling of cells and found that 30 min and 3 hours of AKTi-1/2 treatment inhibited pre-rRNA synthesis (Fig. 1C). Moreover, rapamycin had minimal effects on the synthesis of 45S pre-rRNA after 30 min but more robustly inhibited this process after 3 or 12 hours of treatment (Fig. 1C). Combined rapamycin and AKTi-1/2 treatment for 12 hours resulted in increased repression of the synthesis of 45S pre-rRNA. To measure processing of pre-rRNA into the mature 18S, 5.8S, and 28S rRNAs, we repeated and extended these experiments to include a 3-hour “chase” with nonlabeled medium (in the presence or absence of inhibitors). AKT inhibition significantly reduced the abundance of mature 28S and 18S rRNA at all time points, whereas rapamycin had little effect until 12 hours of treatment (Fig. 1C, bottom panel and graphs), further emphasizing the differences in mechanisms of action for AKT- and mTORC1-mediated regulation of ribosome biogenesis. Notably, the accumulation of processed rRNA species in AKTi-1/2– and rapamycin-treated cells was not stoichiometric. Whereas 28S and 18S rRNA were reduced by AKTi-1/2 treatment for 30 min, 3 hours, or 12 hours, and by rapamycin treatment for 12 hours, 45S rRNA abundance remained unchanged (Fig. 1C), suggesting that AKT and mTORC1 also regulate rRNA processing at steps after 45S synthesis in addition to rDNA transcription. To further delineate the requirement for AKT in rDNA transcription compared to rRNA processing, we used a small-molecule inhibitor of Pol I transcription, CX-5461 (36). Pulse labeling of newly synthesized 45S pre-rRNA with [32P]orthophosphate confirmed that CX-5461 inhibited pre-rRNA synthesis (fig. S2A). We next pulse-labeled cells with [32P]orthophosphate followed by a 1-hour chase with nonlabeled medium before treating with AKTi-1/2 or CX-5461 for 2 hours to examine the effect of the inhibitors on processing of the labeled 45S rRNA (fig. S2B). Consistent with its specific role in inhibition of Pol I transcription but not processing, CX-5461 had no effect on the stoichiometry of the processed 32S, 28S, and 18S rRNAs compared to vehicle-treated cells (fig. S2C). In contrast, treatment with AKTi-1/2 reduced 28S and 18S rRNA abundance and triggered accumulation of 45S, consistent with an additional role for AKT in regulating rRNA processing (fig. S2C).

To determine the step at which AKTi-1/2 repressed 5′ETS pre-rRNA synthesis, we performed Pol I quantitative chromatin immunoprecipitation (qChIP) analysis at the 5′ end of the rDNA gene (5′ETS) and within the transcribed portion of the gene (5.8S) (37). If initiation rates were preferentially repressed by inhibition of AKT, we predicted that Pol I loading across the transcribed portion of the gene should decrease, because the elongating polymerases would not be replaced by newly transcribing complexes at the 5′ end of the gene. Alternatively, if elongation was preferentially repressed, we predicted that Pol I loading would not change or would increase immediately downstream of the transcription start site of the 45S rRNA gene, because newly initiating Pol I complexes would stall as a result of the elongation block. After 30 min of treatment with AKTi-1/2, Pol I loading at the 5′ETS and further along at the 5.8S region of the rDNA gene was not decreased (Fig. 1D) despite reduced 5′ETS pre-rRNA abundance (Fig. 1B) and 45S rRNA synthesis (Fig. 1C). This suggests that steps subsequent to initiation and promoter escape, such as rRNA transcription elongation, were preferentially repressed after 30 min of AKT inhibition. In contrast, after 3 hours of AKT repression, Pol I loading at both the 5′ETS and the 5.8S region was significantly reduced, demonstrating that prolonged AKT inhibition also represses initiation of rDNA transcription and Pol I loading onto the 45S rRNA gene or, alternatively, that a prolonged block to elongation or processing leads to release of Pol I from the rRNA genes.

In addition to these pharmacological inhibitor studies, we also confirmed our results in cells genetically deficient in one of the three isoforms of AKT (AKT1, 2, or 3). Indeed, mouse embryonic fibroblasts (MEFs) isolated from AKT1 knockout mice, but not those from AKT2 or AKT3 knockout mice, showed significantly reduced 5′ETS pre-rRNA abundance compared to wild-type mice (fig. S3). The reduction in 5′ETS pre-rRNA abundance correlated with reduced total AKT abundance and a modest reduction in the phosphorylation of Ser473 in AKT and of PRAS40 in AKT1 knockout MEFs. This modest reduction in AKT activity may reflect compensatory increases in activity of the two remaining isoforms.

These data demonstrated a requirement for AKT to maintain ongoing rRNA transcription in human and murine cells. Accordingly, we asked whether AKT was also necessary for the reactivation of rDNA transcription in quiescent cells stimulated with growth factors. HEK293 cells were serum-starved for 24 hours, pretreated with AKTi-1/2 or rapamycin for 30 min, and stimulated with 10% serum for 30 min, 3 hours, or 12 hours. Serum stimulation for 30 min had a minimal effect on 5′ETS pre-rRNA abundance (Fig. 2A). However, after 3 hours of serum stimulation, 5′ETS pre-rRNA abundance increased three- to fourfold, which was then reduced by 50% if the cells were pretreated with AKTi-1/2 or rapamycin. Inhibition of AKT had a minimal effect on mTORC1 signaling as indicated by phosphorylation of rpS6 (Fig. 2B), consistent with an mTORC1-independent role for AKT in the control of 45S rRNA synthesis. After 12 hours of inhibitor treatment, 5′ETS pre-rRNA abundance was repressed by both inhibitors to an extent comparable to that induced by serum starvation (Fig. 2A). We also directly examined 45S rRNA synthesis by metabolic labeling (Fig. 2C). Consistent with the 5′ETS pre-rRNA abundance data (Fig. 2A), 45S rRNA synthesis increased after 3 or 12 hours of serum stimulation (Fig. 2C, top panel and graphs). In contrast, AKTi-1/2 repressed the synthesis of 45S rRNA to less than that detected in cells serum-starved for 3 hours, whereas rapamycin caused a 50% reduction in 45S rRNA abundance in serum-stimulated cells (Fig. 2C). The greater effect of AKTi-1/2 on synthesis of mature 45S pre-rRNA (Fig. 2C) compared to that on the abundance of 5′ETS rRNA (Fig. 2A) is consistent with a model whereby AKT signaling is required for transcription elongation or processing (or both) downstream of the 5′ETS region of the rDNA gene. A 3-hour chase of the 30-min pulse-labeled cells demonstrated that the accumulation of the rRNA species in AKTi-1/2– and rapamycin-treated cells was not stoichiometric (Fig. 2C, bottom panel and graphs), which is also consistent with data showing that both AKT and mTORC1 signaling can regulate rRNA processing in addition to rRNA synthesis (Fig. 1). Together, the above studies demonstrate that AKT is required for both maintenance of rDNA transcription in exponentially growing cells and reactivation of ribosome biogenesis in quiescent cells that reenter the cell cycle upon exposure to growth factors, and this regulation is in part independent of mTORC1.

Fig. 2

AKT is required for activation of rDNA transcription and rRNA synthesis. HEK293 cells were serum-starved for 24 hours, pretreated with either 20 nM rapamycin, 5 μM AKTi-1/2, or both for 30 min, and then stimulated with 10% serum for 30 min, 3 hours, or 12 hours. (A) RPA for 5′ETS pre-rRNA abundance. n = 3 experiments. (B) Western blotting with the indicated antibodies. Representative of n = 3 blots. Quantitation in table S3. (C) Pulse labeling (left) and quantitation (right). Quantitation in table S4 for 45S, 28S, and 18S rRNA abundance. n = 2 to 3 experiments. #P < 0.05, **,##P < 0.01 compared to the time-matched control (serum-starved) or serum-stimulated cells, respectively.

AKT can increase ribosome biogenesis

We examined whether activation of AKT-mediated signaling could increase rDNA transcription and whether it required growth factors. We performed these studies in immortalized, nontransformed human fibroblasts with relatively low AKT activity during exponential growth. In these cells, AKT activity is reduced by growth factor withdrawal, thus facilitating AKT gain-of-function studies. We generated cell lines from low-passage human fibroblasts immortalized by the expression of human telomerase reverse transcriptase (hTERT), the catalytic subunit of telomerase (BJ-T) (38) that stably expressed constitutively activated myristoylated (Myr) AKT1 and AKT3 (Fig. 3A). Serum starvation induced cell cycle arrest in vector control and MyrAKT1- or MyrAKT3-expressing cells (90% cells in G1 phase) (fig. S4). However, expression of constitutively active AKT1 or AKT3 increased the abundance of 5′ETS pre-rRNA in serum-starved cells by 2.5- and 4-fold, respectively, such that the abundance was similar to that achieved in vector only–transfected cells after serum stimulation (Fig. 3B). Thus, in the absence of growth factor signaling, AKT signaling can increase rDNA transcription (as measured by 5′ETS pre-rRNA synthesis). Moreover, expression of MyrAKT1 or MyrAKT3 augmented the abundance of 5′ETS pre-rRNA twofold in cells stimulated with 10% serum for 12 hours (Fig. 3B), demonstrating that AKT activity is limiting for rDNA transcription even in the presence of other serum-derived growth factors. Pulse-labeling experiments confirmed that enforced AKT expression increased 45S pre-rRNA synthesis (Fig. 3C). Furthermore, a 3-hour chase after the 30-min pulse labeling of cells indicated that expression of constitutively active AKT increased pre-rRNA processing, as indicated by the increased abundance of 18S and 28S rRNA in both the absence and the presence of growth factors (Fig. 3C, bottom panel). Unlike 5′ETS pre-rRNA abundance (Fig. 3B), the extent of the increase in 45S pre-rRNA synthesis in serum-starved MyrAKT1- and MyrAKT3-expressing cells was less than that observed after serum stimulation of empty vector control cells (Fig. 3C). Thus, whereas AKT can increase transcription of the 5′ETS region of the 45S rDNA in the absence of growth factors, signaling through other pathways is required for synthesis of full-length 45S pre-rRNA at rates comparable to those seen in serum-stimulated cells. This suggests that steps subsequent to loading of Pol I onto the 5′ end of the 45S rDNA are rate-limiting for rRNA synthesis. Indeed, qChIP experiments in serum-starved cells demonstrated that enforced expression of constitutively active AKT3 increased Pol I loading at the 5′ETS and 5.8S regions of the gene; however, whereas loading at 5′ETS was insensitive to rapamycin, loading downstream of the 5′ETS region was blocked by rapamycin (Fig. 3D). Consequently, the synthesis of full-length 45S pre-rRNA was also reduced with rapamycin treatment (Fig. 3C). Thus, AKT signaling independent of mTORC1 increases loading of Pol I onto the 5′ end of the 45S gene, but mTORC1 signaling is required for AKT to increase full-length 45S pre-rRNA synthesis under conditions of limited growth factor and nutrient availability. To independently demonstrate that AKT does not require mTORC1 for loading Pol I onto the 5′ end of the rDNA, we determined the effect of small interfering RNA (siRNA)–mediated knockdown of the mTORC1-specific component raptor (which reduces mTORC1 signaling) on AKT-dependent loading of Pol I onto rDNA in MyrAKT3-overexpressing cells (fig. S5). Consistent with the effect of rapamycin, inhibition of mTORC1 signaling by raptor depletion did not decrease, but rather increased, AKT-dependent loading of Pol I at the 5′ end of the 45S gene. In contrast, raptor knockdown resulted in decreased loading of Pol I at the 5.8S region of the gene, downstream of the 5′ETS (fig. S5). These results are consistent with a model in which sustained inhibition of mTORC1 blocks AKT-driven Pol I elongation but not transcription initiation, leading to accumulation of stalled Pol I at the 5′ end of the 45S gene.

Fig. 3

AKT promotes ribosome biogenesis. BJ-T cells were transduced with MSCV, MyrAKT1, or MyrAKT3 retrovirus. Cells were serum-starved for 36 hours and treated with 20 nM rapamycin or 10% FBS for 12 hours (or both). (A) Western blotting. Representative of n = 3 blots. (B) RPA for 5′ETS pre-rRNA abundance. n = 3. (C) Pulse labeling for 45S, 28S, and 18S rRNA abundance. n = 3 sets of cells. Quantitation in table S5. (D) qChIP analysis for RNA Pol I loading on 5′ETS and 5.8S region of rDNA gene. n = 4 sets of cells. (E) RNA or protein cellular content. n = 3 sets of cells. *,#P < 0.05, **P < 0.01, ***P < 0.001 compared to control, serum-starved, or serum-treated cells, respectively.

Forced expression of constitutively active AKT increased not only 45S pre-rRNA synthesis, the abundance of 18S and 28S rRNA, and transcription of 5S rRNA by Pol III in serum-starved cells, but also the number of mature ribosomes in the cytoplasm (fig. S6). Sucrose density gradient analysis under high-salt conditions [which denatures 80S ribosomes into the individual 40S and 60S subunits (39)] revealed that MyrAKT3 overexpression increased the abundance of both ribosomal subunits, which corresponded to increased numbers of 80S ribosomes under native conditions (fig. S6B). Accordingly, MyrAKT3 expression also increased protein content (Fig. 3E) and cell size (fig. S4). Thus, AKT increases ribosome biogenesis and cell growth, which includes the synthesis of rRNA and ribosomal proteins, processing of the 45S rRNA, and assembly of functional ribosomal subunits. Although AKT could stimulate growth in BJ-T cells, it did not affect the net proliferation rate, as assessed by the proportion of cells in S phase or the basal rates of apoptosis (fig. S4).

AKT cooperates with c-MYC to increase ribosome biogenesis

The oncoprotein c-MYC activates several key aspects of ribosome biogenesis including transcription by Pol I of the 45S rRNA gene, Pol II–dependent transcription of ribosomal protein genes, and Pol III–dependent transcription of the 5S rRNA gene (40, 41). Our observations that AKT can increase 45S pre-rRNA synthesis, ribosome biogenesis, and cell growth led us to examine the interaction between these two pathways in the control of ribosome biogenesis in mammalian cells. We generated human fibroblasts expressing an inducible version of c-MYC (c-MYC–ER). These cells had increased 5′ETS pre-rRNA (Fig. 4A) and 45S pre-rRNA synthesis when growth factor availability was reduced (Fig. 4B), consistent with our previous studies (32). Notably, the c-MYC–dependent increases in 5′ETS pre-rRNA abundance (Fig. 4A), 45S pre-rRNA synthesis, and accumulation of mature 28S and 18S rRNA (Fig. 4B) were significantly inhibited within 3 hours of AKTi-1/2 treatment. Thus, AKT signaling is required for c-MYC activation of rDNA transcription. Conversely, inhibition of mTORC1 activity with rapamycin (Fig. 4C) had less effect on ribosome biogenesis (Fig. 4, A and B), consistent with the differences in mechanism of action for AKT and mTORC1 observed in HEK293 cells.

Fig. 4

c-MYC–driven ribosome biogenesis requires AKT. BJ-T cells stably expressing pBABE or c-MYC–ER were cultured in the presence of 0.5% FBS for 36 hours. c-MYC activity was induced for 12 hours with 200 nM 4-hydroxytamoxifen and treated for 3 hours with 20 nM rapamycin or 5 μM AKTi-1/2. (A) RPA for 5′ETS pre-rRNA abundance. n = 3 sets of cells. (B) Pulse labeling (left) and quantitation (right). Quantitation in table S6. n = 2 to 3 sets of cells. (C) Western blotting. Representative of n = 3 blots. Quantitation in table S7. **P < 0.005 compared to control (serum-starved); #P < 0.01 compared to c-MYC–ER.

Expression of constitutively active AKT in BJ-T cells did not affect c-MYC mRNA or protein abundance or activity (fig. S7), suggesting that AKT does not regulate c-MYC itself but is most likely required for signaling downstream of c-MYC. c-MYC overexpression repressed endogenous phosphorylation of AKT; however, this did not result in decreased phosphorylation of the AKT substrate PRAS40, indicating that residual AKT activity is sufficient for downstream signaling (Fig. 4C). This repression of AKT activation is not due to decreased AKT abundance. Quantitative reverse transcription–polymerase chain reaction (qRT-PCR) analysis demonstrated no significant change in the abundance of mRNAs encoding AKT1, AKT2, or AKT3 in cells overexpressing c-MYC (fig. S7C). Furthermore, AKT abundance was not significantly changed after expression of c-MYC, and PRAS40 phosphorylation was reduced by AKTi-1/2 treatment (table S7). Thus, low abundance of AKT can facilitate c-MYC activation of rDNA transcription.

Consistent with previous studies (2, 42), c-MYC activated Pol III (as assessed by 5S abundance), an effect that was sensitive to rapamycin (Fig. 4B). However, as was the case for 45S pre-rRNA synthesis by Pol I, 5S synthesis was more potently inhibited by AKTi-1/2 than by rapamycin. These results suggest that c-MYC and AKT work through parallel pathways, or at least target separate Pol I– and Pol III–dependent processes in rRNA synthesis that are necessary for maximal ribosome biogenesis when growth factor availability is limited.

To further explore this possibility, we determined whether AKT and c-MYC cooperate in the stimulation of rDNA transcription. Stable BJ-T cell lines expressing c-MYC–ER, MyrAKT3, or both were generated and examined for 5′ETS and 45S pre-rRNA synthesis and ribosome biogenesis and growth. We used AKT3 because it has the highest activity of the three isoforms (43). These experiments were conducted in the presence of 0.5% serum to minimize apoptosis in serum-starved cells induced by the combination of c-MYC–ER and MyrAKT3 overexpression. However, 0.5% serum alone increased basal 5′ETS and 45S pre-rRNA abundance, thus reducing the apparent increase in rDNA transcription caused by expression of c-MYC or AKT alone (Fig. 5, A and B; compare to Fig. 4, A and B). However, when MyrAKT3 and c-MYC–ER were coexpressed, the increases in 5′ETS pre-rRNA abundance (Fig. 5A) and 45S pre-rRNA and 5S rRNA synthesis (Fig. 5B) were additive. Similarly, combined overexpression of MyrAKT3 and c-MYC–ER produced increases in RNA and protein content per cell greater than that induced by overexpression of either construct alone (Fig. 5C), which is again consistent with the modulation of rDNA transcription by AKT and c-MYC through distinct but cooperative pathways. The additive effect of MyrAKT3 and c-MYC–ER on 45S pre-rRNA synthesis, but not on 5′ETS pre-rRNA abundance, was blocked by rapamycin (Fig. 5, A and B), suggesting that mTORC1 signaling is required for efficient rDNA transcription downstream of the 5′ETS region and is required for the cooperation between c-MYC and AKT to modulate ribosome biogenesis (fig. S8).

Fig. 5

AKT and c-MYC cooperate to stimulate ribosome biogenesis. BJ-T cells stably expressing pBABE or c-MYC–ER were transduced with MSCV- or MyrAKT3-encoding virus. Cells were serum-starved in 0.5% FBS for 36 hours and treated with 20 nM rapamycin for 12 hours. (A) RPA analysis for 5′ETS pre-rRNA abundance. n = 3 sets of cells. (B) Pulse labeling (left) and quantitation (right) to determine rRNA abundance. n = 3 sets of cells. Quantitation in table S8. (C) RNA or protein cellular content. n = 4 to 6 sets of cells. *P < 0.05, **P < 0.01, ***P < 0.001 compared to control (serum-starved empty vector, MyrAKT3, or c-MYC–ER).

AKT is required for c-MYC activation of rDNA transcription in c-MYC–driven malignancy

Because c-MYC drives growth during cancer in various settings, and aberrant translational control is a factor that underlies c-MYC–induced tumorigenesis (18), we examined the extent to which AKT signaling was required for c-MYC–driven ribosome biogenesis in transformed cells. We used a transgenic model in which the c-MYC oncogene is coupled to the immunoglobulin heavy chain enhancer (Eμ), mimicking the translocation between chromosomes 8 and 14 [8;14)] that juxtaposes the c-MYC oncogene to the Eμ region in various human lymphomas and is characteristic of Burkitt lymphoma (44, 45). In this model, overexpression of c-MYC results in increased total cellular RNA synthesis, consistent with a role for c-MYC in promoting rDNA transcription (32, 33, 46). Treatment of cultured exponentially growing Eμ-Myc cells with AKTi-1/2 for 3 hours decreased both 5′ETS pre-rRNA abundance and the synthesis of mature 45S pre-rRNA (Fig. 6, A and B). In contrast, under the same conditions, rapamycin did not significantly alter 5′ETS abundance or 45S rRNA synthesis, although it repressed mTORC1 signaling (fig. S9A). The amount of AKT in Eμ-Myc cells purified from tumor-bearing mice was higher than that in normal B cells purified from control mice (fig. S9B). However, it is not clear whether this correlated with a concomitant increase in AKT activity because the amounts of phosphorylated AKT in B cells and Eμ-Myc cells from tumor-bearing and normal mice were below the detection limit by Western blotting. As with c-MYC–ER fibroblasts, AKTi-1/2 treatment of cultured Eμ-Myc lymphoma cells reduced phosphorylation of PRAS40, reflecting the contribution of basal AKT signaling in these cells (fig. S9A). Sustained treatment (16 hours) of Eμ-Myc cells with AKTi-1/2 resulted in increased cell death, whereas rapamycin did not affect survival (fig. S9C), indicating that in this system, AKT signaling is essential for ribosome biogenesis and survival, independent of mTORC1. The ability of AKTi-1/2 to rapidly repress rDNA transcription by 3 hours suggested that the effect was not an indirect consequence of inducing apoptosis. To confirm this, we used Eμ-Myc cells overexpressing BCL2 (47), an inhibitor of the intrinsic apoptosis pathway (48). As expected, overexpression of BCL2 prevented AKTi-1/2–induced apoptosis (fig. S9C), but did not affect the ability of AKTi-1/2 to repress rDNA transcription (fig. S9C). Moreover, treatment with AKTi-1/2, but not rapamycin, of Eμ-Myc cells overexpressing BCL2 resulted in reduced ribosome biogenesis and cell growth as indicated by reduced RNA and protein content per cell (Fig. 6C). Thus, AKT, but not mTORC1, is required for both high rates of rDNA transcription, growth, and survival of Eμ-Myc lymphoma cells.

Fig. 6

AKT is required for c-MYC–driven ribosome biogenesis. Wild-type Eμ-Myc B cell lymphoma cells and those overexpressing BCL2 were plated at 1 × 106 cells per well in six-well plates and treated as indicated. (A) qRT-PCR analysis of cells treated with 20 nM rapamycin or 5 μM AKTi-1/2 for 3 hours to determine 5′ETS pre-rRNA abundance. n = 4 sets of cells. (B) Pulse labeling of cells treated with 20 nM rapamycin or 5 μM AKTi-1/2 for 3 hours and quantitation of 45S rRNA abundance. n = 3 sets of cells. (C) RNA or protein content of cells treated with 20 nM rapamycin or 5 μM AKTi-1/2 for 16 hours. n = 4 sets of cells. **P < 0.01, ***P < 0.001 compared to control (vehicle).

Discussion

rRNA synthesis limits the rate of ribosome biogenesis and cell growth and proliferation (4951) and is likely to be a critical process dysregulated during malignant transformation (17, 18). Indeed, many of the oncogenes and tumor suppressors frequently implicated as drivers of transformation [including PI3K, RAS, c-MYC, PTEN (phosphatase and tensin homolog deleted from chromosome 10), retinoblastoma protein, and p53] control ribosome biogenesis at multiple levels (36). Therefore, understanding how signaling pathways cooperate and redundantly control this process in normal and transformed cells is critical for the rational design and implementation of novel therapeutics that selectively target tumor cells.

To dissect the key signaling events in the regulation of ribosome biogenesis in normal and tumor cells, we explored the role of the AKT kinase family in controlling rDNA transcription and the crosstalk of this pathway with another key regulator of ribosome biogenesis, c-MYC. We demonstrated that the AKT family signals through mTORC1-dependent and -independent mechanisms to promote rDNA transcription in mammalian cells and cooperates with c-MYC to promote ribosome biogenesis and cellular growth. Furthermore, AKT signaling was necessary for high rates of rDNA transcription and cell survival during c-MYC–driven lymphomagenesis, suggesting that targeting AKT-dependent regulation of ribosome biogenesis might be a viable therapeutic approach to c-MYC–driven tumor cell growth.

AKT controls Pol I transcription at multiple steps

We demonstrated that AKT signaling regulates rRNA synthesis at multiple levels, including loading of Pol I onto the rDNA (transcription initiation and promoter escape), and subsequent steps, such as transcription elongation and rRNA processing. However, maximal repression of rDNA transcription in response to AKT inhibition occurred within 30 min of treatment (Fig. 1, B and C) without decreasing Pol I loading on the transcribed portion of the 45S gene (Fig. 1D), indicating that continuous AKT signaling preferentially targets steps downstream of Pol I loading such as elongation. These findings are consistent with the work of Stefanovsky and Moss (52), who demonstrated that elongation is a key regulatory point in the synthesis of rRNA and, thus, growth in mammalian cells. However, our findings contrast with other studies that have implicated the Pol I transcription initiation factor Rrn3 in modulating growth factor regulation of rDNA transcription (53). Furthermore, inhibition of AKT resulted in defects in rRNA processing and led to the accumulation of unprocessed 45S rRNA. It is possible that AKT control over rRNA processing, which is coupled to elongation in yeast (54, 55), may be dominant and limiting for transcription elongation and, thus, rRNA synthesis. Further studies are required to examine this possibility.

In contrast to acute inhibition of rDNA transcription, more sustained inhibition of AKT (>3 hours) reduced the enrichment of Pol I at 5′ETS and 5.8S regions on the 45S rRNA gene, suggesting that under these conditions, Pol I loading onto the rDNA was limiting. Thus, the effect of inhibition of AKT signaling on rDNA transcription is biphasic, with an immediate effect on repressing elongation followed by repression of Pol I loading (fig. S8).

AKT regulates rDNA transcription through mTORC-dependent and -independent mechanisms

AKT lies upstream of mTORC1, a signaling complex that regulates ribosome biogenesis and rDNA transcription (28, 29), raising the possibility that AKT might function to modulate rDNA transcription indirectly through mTORC1. However, we showed that acute inhibition of AKT blocks rDNA transcription at the post-initiation level with little effect on mTORC1 signaling, and direct inhibition of mTORC1 for 30 min had no effect on rDNA transcription. Thus, continuous AKT-dependent, but mTORC1-independent, signaling is required to maintain ongoing rRNA synthesis. In contrast to acute inhibition of rDNA transcription, sustained inhibition of AKT (>3 hours) led to reduced Pol I loading at 5′ETS and 5.8S regions on the 45S rRNA gene, effects that correlated with decreased mTORC1 signaling. Moreover, sustained inhibition of mTORC1 signaling also suppressed Pol I loading onto the rRNA genes. Thus, defects in rDNA transcription initiation or promoter release (or both) in response to sustained repression of AKT signaling may be mediated in part though repression of mTORC1 signaling.

AKT signaling promotes rDNA transcription in the absence of growth factors

Consistent with the ability of AKT to control Pol I transcription at multiple levels, ectopic expression of constitutively active AKT increased both loading of Pol I onto rRNA genes and synthesis of full-length 45S pre-rRNA, leading to accumulation of mature cytoplasmic ribosomes. However, AKT signaling did not optimally activate rDNA transcription when mTORC1 signaling was blocked (Fig. 3C). Indeed, although AKT promoted loading of Pol I onto the 5′ end of 45S rDNA in the absence of growth factors and in the presence of rapamycin, mTORC1 signaling was required for optimal synthesis of mature 45S pre-rRNA (Fig. 3D).

We propose a model (fig. S8) in which continuous AKT signaling is required to maintain Pol I transcription elongation and processing but not loading of Pol I onto the 5′ end of the rDNA. However, sustained inhibition of AKT affects both pre- and post-initiation regulation of rDNA transcription, including pre-rRNA processing, in part due to loss of mTORC1 signaling. In the absence of growth factors, AKT signaling independent of mTORC1 promotes loading of Pol I onto rRNA genes but does not increase synthesis of 45S rRNA, suggesting that AKT-dependent activation of mTORC1 is required for post-initiation events such as transcription elongation and rRNA processing.

AKT can maintain ribosome biogenesis and growth after withdrawal of growth factors

In growth factor–deprived cells, ectopic expression of constitutively active forms of AKT resulted in not only high rates of rDNA transcription but also accumulation of mature ribosomes in the cytoplasm and protein synthesis. These findings identify AKT as a master controller of ribosome biogenesis (similar to c-MYC) and suggest that aberrant AKT signaling may contribute to dysregulation of ribosome biogenesis and growth associated with cancer (1, 22, 36). Indeed, ectopic expression of AKT resulted in increased rDNA transcription under relatively robust growth conditions, suggesting that AKT activity is limiting for ribosome biogenesis. Thus, constitutive activation of signaling through the PI3K-AKT pathway, which frequently occurs in tumor cells, may be associated with increased ribosome biogenesis.

c-MYC and AKT cooperate to promote rDNA transcription in normal and malignant cells

c-MYC is a well-characterized central regulator of ribosome biogenesis and can regulate the major steps in the formation of functional ribosomes. However, the mechanism by which other growth signaling pathways interact with c-MYC to cooperate or redundantly regulate rDNA transcription has not been previously studied. Here, we show that c-MYC and AKT cooperate to promote rDNA transcription and ribosome biogenesis, suggesting that PI3K-AKT-mTORC1 and c-MYC are key regulators of ribosome synthesis in mammalian cells. Such cooperation is consistent with the observations that the c-MYC and PI3K pathways synergize during other processes such as regulation of the cell cycle (56, 57) and apoptosis (58, 59). c-MYC and AKT cooperatively regulate fibroblast proliferation and transformation through phosphorylation of FoxO transcription factors by AKT, thereby alleviating FoxO-mediated inhibition of the transcription of c-MYC target genes (60). However, this does not appear to be the mechanism for coordinate control of rDNA transcription, because overexpression of AKT did not affect the ability of c-MYC to transcriptionally activate its target genes (fig. S7). Conversely, ectopic expression of c-MYC did not stimulate AKT activity, suggesting that these two pathways function in parallel (Fig. 4C).

c-MYC is dysregulated in ~15 to 20% of human cancers. The cooperation between c-MYC and AKT in the regulation of rDNA transcription and ribosome biogenesis raises the possibility that AKT may control growth in tumors driven by c-MYC. Consistent with this concept, we found that acute repression of AKT, but not inhibition of mTORC1, led to repression of rDNA transcription in a c-MYC–driven model of murine B cell lymphoma (Eμ-Myc) (Fig. 6). Thus, consistent with the data obtained in HEK293 cells and fibroblasts, AKT signaling is required for optimal rates of rDNA transcription in B cell lymphoma, independent of mTORC1 activity. Concentrations of AKTi-1/2 that inhibited rDNA transcription within 30 min also led to apoptotic cell death within 12 to 24 hours. The decrease in rDNA transcription elicited by inhibition of AKT was not due to indirect effects of apoptosis, because cells overexpressing BCL2, an inhibitor of the intrinsic apoptosis pathway, also showed decreased rDNA transcription in response to AKTi-1/2. One intriguing possibility is that apoptosis elicited by inhibition of AKT is, at least in part, a consequence of reduced rDNA transcription. Disruption of rDNA transcription leads to nucleolar stress, activation of p53, and apoptosis in various cell types (61, 62). Moreover, c-MYC regulates p53 abundance in Eμ-Myc lymphoma cells through overproduction of ribosomal proteins that suppress the activity of p53 ubiquitin ligase MDM2 (63).

Together, our findings show that AKT cooperates with c-MYC to control ribosome biogenesis in mammalian cells and that malignant diseases characterized by unrestrained cellular growth may be vulnerable to therapeutic strategies that target AKT.

Materials and Methods

Statistical analysis

In experiments with n > 3, the graphs represent mean ± SEM and were analyzed using one-way analysis of variance (ANOVA) with Tukey’s post-test. In experiments with n < 3, the graphs represent mean ± SD. Key mean values and statistical analysis are tabulated in tables S1 to S10.

Plasmid constructs

Constitutively active mutants of AKT were generated by N-terminal addition of a myristoylated sequence (64) to hemagglutinin (HA)–tagged wild-type AKT1 and AKT3 by PCR (table S11) and cloned into pcDNA3. These were subcloned into the murine stem cell virus [MSCV–GFP (green fluorescent protein) vector (Clontech). pBABE Myc-ER is described by Wall et al. (65).

Cell lines and culture

HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS). Human foreskin fibroblast cells expressing hTERT (BJ-T) (38) were cultured in a 1:4 (v/v) mix of DMEM and Medium 199 with 15% FBS and l-glutamine (3.75 mM). Eμ-Myc B cell lymphoma 4242 cells were cultured in Anne Kelso DMEM with 10% FBS, 1% (v/v) penicillin/streptomycin/glutamate, 0.1 mM l-asparagine, and 55 μM β-mercaptoethanol. All media were supplemented with 1% antibiotic-antimycotic (Gibco). Cell lines were maintained in 5% CO2 at 37°C.

BJ-T cell lines stably expressing pBABE and c-MYC–ER were generated by infection with high-titer retrovirus (Supplementary Materials) and selected with puromycin (2 μg/ml; Sigma). c-MYC–ER was activated with 200 nM tamoxifen for 12 hours.

BJ-T cells transiently expressing MSCV, MyrAKT1, or MyrAKT3 were generated by infecting the appropriate stable cell line with retroviral media (Supplementary Materials) three times over 2 days. Experiments were carried out within 2 weeks of infection.

Treatments

Exponentially growing cells were plated and incubated for 48 hours before harvesting. Cells were serum-starved by washing twice with phosphate-buffered saline (PBS), and serum starvation medium was added for 24 to 36 hours. Serum starvation medium contained 0.5% bovine serum albumin (BSA) for the HEK293 and BJ-T cells or 0.5% FBS for the BJ-T Myc-ER cells. Cells were treated with a combination of vehicle, serum, AKTi-1/2 (Calbiochem), or rapamycin (Calbiochem) as indicated in the figure legends.

Cell number determination

Cell number was determined with the Beckman Coulter Z2.

Protein extracts and immunoblotting

Protein extracts were prepared (43) [concentration was determined by DC assay (Bio-Rad)], separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE), and immunoblotted with the appropriate antibodies. Proteins were visualized by Western Lightning Chemiluminescence Plus (Perkin Elmer) and detected by exposure to x-ray film (Hyperfilm ECL film or Kodak). Band densities were quantified with ImageQuant Software (GE Healthcare).

The following antibodies were used: p-AKTSer473 (Cell Signaling Technology 9271), p-rpS6Ser240/244 (Cell Signaling Technology 2217), p-PRAS40 (Cell Signaling Technology 2640), actin (MP Biomedicals 691002), total AKT, 12CA5 (Pearson Lab), c-MYC (Santa Cruz Biotechnology 764), and secondary horseradish peroxidase (HRP)–conjugated antibodies (rabbit: 172-1019; mouse: 172-1011, Bio-Rad).

RNA extraction and expression analysis

Cells were lysed in 4 M guanidine thiocyanate, 25 mM sodium citrate (pH 7), 0.5% (w/v) sarcosyl, and 0.1 M β-mercaptoethanol, and RNA was extracted with phenol and chloroform by means of standard methods. RNA was spiked with a 32P-labeled RNA probe to allow for calculation of RNA recovery. RNA was quantitated and normalized to equal RNA or cell number. First-strand complementary DNA (cDNA) was synthesized using random hexamer primers (Promega) and Superscript III (Invitrogen) per the manufacturer’s instructions. qRT-PCR was performed as described for ChIP with primers listed in table S12. Samples were quantitated with the ΔΔCT method (32).

Ribonuclease protection assay

Ribonuclease (RNase) protection assay (RPA) was performed as described by Lister et al. (66). Specifically, 32P-labeled antisense probes (Supplementary Materials) were generated corresponding to the 5′ETS of the human 45S rRNA precursor (table S12). To measure rDNA transcription, we hybridized RNA from equal numbers of cells (~20,000 HEK293 cells or ~30,000 to 50,000 BJ-T cells) to the [32P]uridine triphosphate–labeled RNA probe followed by RNase digestion (RNase A: Boehringer Mannheim; RNase T1: Roche). RNA hybrids were resolved by electrophoresis on a 5% nondenaturing acrylamide gel and detected with the Storm 820 Phosphorimager; intensities were quantitated with ImageQuant software (GE Healthcare).

Pulse labeling and analysis

rRNA abundances were determined by pulse labeling as described by Stefanovsky et al. (67). Independent of the treatment time with inhibitors, 2 hours before the cells were to be harvested or “chased,” they were washed and incubated in phosphate-free DMEM (Gibco) supplemented with dialyzed FBS for 1.5 hours followed by a 30-min label with 0.5 mCi [32P]orthophosphate (MP Biomedicals). Cells were harvested or washed and chased by addition of complete medium (including inhibitors) for the time periods indicated, and RNA was extracted as above. Equal RNA (5 to 10 μg) or RNA from equal cell number (~300,000 cells) was separated on a 1.2% Mops formaldehyde gel. Ethidium bromide–stained RNA was visualized with Gene Genius Bioimaging System (Syngene). The gel was dried (Model 583 Gel Drier) and exposed to a phosphorimager screen overnight. Bands corresponding to rRNAs were visualized with the Storm 820 Phosphorimager and intensities were quantitated with ImageQuant software (GE Healthcare).

qChIP

qChIP was performed as described previously (32, 37, 68). Cells were cross-linked with 0.6% formaldehyde, and assays were performed with 4 × 106 to 6 × 106 cells and 8 μl of immune sera. Samples were analyzed in triplicate with the FAST SYBR Green dye on the ABI StepOnePlus (Applied Biosystems). The percentage of total DNA bound was calculated using unprecipitated input samples as a reference for all qRT-PCRs. Primer sets are listed in table S12.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/4/188/ra56/DC1

Materials and Methods

Fig. S1. AKTi-1/2 inhibits AKT activity in HEK293 and BJ-T cells.

Fig. S2. AKTi-1/2 represses rDNA transcription and rRNA processing.

Fig. S3. Knockout of AKT isoforms reduces rDNA transcription.

Fig. S4. Overexpression of MyrAKT3 increases cell size and serum starvation induces cell cycle arrest without changing apoptosis rates.

Fig. S5. Knockdown of raptor mimics the effect of rapamycin on RNA Pol I loading.

Fig. S6. Overexpression of MyrAKT3 increases the abundance of the ribosomal subunits 40S, 60S, and 80S.

Fig. S7. AKT does not modulate c-MYC abundance or activity and c-MYC does not modulate AKT abundance.

Fig. S8. Model of regulation of 45S synthesis by AKT, mTORC1, and c-MYC.

Fig. S9. Cells isolated from Eμ-Myc B cell lymphomas respond to inhibitors of AKT and mTORC1.

Table S1. Quantitation for Fig. 1A.

Table S2. Quantitation for Fig. 1C.

Table S3. Quantitation for Fig. 2B.

Table S4. Quantitation for Fig. 2C.

Table S5. Quantitation for Fig. 3C.

Table S6. Quantitation for Fig. 4B.

Table S7. Quantitation for Fig. 4C.

Table S8. Quantitation for Fig. 5B.

Table S9. Quantitation for fig. S3A.

Table S10. Quantitation for fig. S9A.

Table S11. Cloning primers.

Table S12. RT-PCR primers.

References

References and Notes

  1. Acknowledgments: We thank S. Jane and L. Cerruti for assistance with retrovirus production and A. George and A. Lesmani for technical assistance. CX-5461 was a gift from Cylene Pharmaceuticals. Funding: This work was supported by grants and fellowships from the National Health and Medical Research Council (NHMRC) of Australia to R.D.H. (NHMRC #166908, #400120, and #509088) and to R.B.P. (NHMRC #509087 and #400116), Cancer Council Victoria to R.B.P., and Swiss National Science Foundation to M.N.H. R.W.J. is an NHMRC Principal Research Fellow. M.W. is funded by a Clinical Research Fellowship from the Victorian Cancer Agency. This work is supported by NHMRC program and project grants plus grants from the Victorian Cancer Agency and Australian Rotary Health. Author contributions: J.C.C. and K.M.H. performed the experiments and analyzed the data. K.R., P.Y.N., A.P., R.S.L., S.H., M.V.A., M.B., K.J., and K.E.S. performed the experiments. M.W., G.P., B.A.H., and M.N.H. analyzed the data. R.W.J. and G.A.M. designed the experiments and analyzed the data. R.D.H. designed the experiments, analyzed the data, and wrote the paper. R.B.P. designed and performed the experiments, analyzed the data, and wrote the paper. Competing interests: R.D.H. is a paid consultant for Cylene Pharmaceuticals. R.W.J. is a paid consultant for Merck and Novartis. G.A.M. is a noncompensated consultant for Pfizer and Novartis.
View Abstract

Navigate This Article