Consequences of interrupted Rheb-to-AMPK feedback signaling in tuberous sclerosis complex and cancer

Abstract

Rheb is a small GTPase primarily known for activating mammalian target of rapamycin complex 1 (mTORC1) and promoting cell growth in response to insulin and nutrients (amino acids, glucose). Shortage of glucose activates adenosine 5′-monophosphate-activated protein kinase (AMPK), which induces catabolic processes that produce ATP and suppresses energy-consuming anabolic reactions. As part of the latter response, AMPK activates the TSC1-TSC2 tumor suppressor complex, which in turn inhibits Rheb, thereby reducing mTORC1 activity and consequently suppressing protein synthesis. We recently identified an mTORC1-independent Rheb-to-AMPK feedback signaling loop in Tsc2-null in vitro models of Tuberous Sclerosis Complex (TSC). In addition to activating AMPK, Rheb reduced the nuclear levels of the cyclin-dependent kinase inhibitor p27^KIP1 (p27). Importantly, Rheb-mediated repression of p27 correlated with activation of Cdk2 and cell proliferation, and with tumor formation by TSC cells. Considering that AMPK was previously reported to regulate stability and subcellular localization of p27, we hypothesize that Rheb regulates p27 in TSC cells by activating AMPK. This article discusses how Rheb-to-AMPK, and p27 signaling may impact on disease progression and treatment of TSC, including sporadic lymphangioleiomyomatosis (S-LAM) and malignancies.

Introduction

Cells control their growth (i.e., increase in size) and proliferation through complex signaling networks that integrate environmental signals, energy sources and availability of nutrients. Consequently, a fine equilibrium between the signaling activities of the individual components of these networks is essential for proper development and prevention of tumor formation. Considering that cell proliferation is typically dependent on cell growth, it is not surprising that alterations in critical components of the cell growth control machinery frequently contribute to proliferative diseases such as tuberous sclerosis complex (TSC), Peutz-Jeghers syndrome, neurofibromatosis type 1, Cowden disease and cancer.^1,2

Within the signaling network that controls cell growth, the TSC1 and TSC2 tumor suppressor genes are of particular importance, as inactivation of either TSC1, which encodes hamartin (TSC1), or TSC2, which encodes tuberin (TSC2), is directly linked to the development of the autosomal dominant disorder TSC. TSC is characterized by benign tumors known as hamartomas, most typically found in brain, kidney, heart, lung, skin and eye. The most frequent manifestation of TSC in the lung is known as lymphangioleiomyomatosis (TSC-LAM), a disease that also occurs sporadically (S-LAM).²

TSC1 and TSC2 physically associate to function primarily as a complex (TSC1/2) for the control of cell growth and proliferation. TSC1/2 activity is regulated by several signal transduction systems, in particular by the phosphoinositide-3-kinase (PI3K)-Akt pathway, and by signaling mechanisms controlling intracellular energy levels.² The PI3K-Akt pathway becomes activated upon binding of growth and survival factors to their respective cell surface receptors (Fig. 1).³ Thereafter, Akt directly phosphorylates TSC2 to inactivate TSC1/2. The primary function of TSC1/2 is to repress mTOR (mammalian target of rapamycin), a kinase centrally involved in promoting cell growth. We and others previously showed that TSC1/2, through a TSC2 GTPase-activating protein (GAP) activity, inactivates the small GTPase Rheb (Ras Homolog Enriched in Brain). Rheb activates mammalian TOR complex 1 (mTORC1) to allow protein synthesis and cell growth to occur (Fig. 1).^2,4

Figure 1.

Signaling network connecting Rheb and AMPK. The Ras/Raf/MEK/ERK pathway is mainly involved in cell proliferation and the Ras/PI3K/Akt/TSC pathway in survival and cell growth.¹ However, by activating ERK/RSK1 or Akt, both pathways converge to inhibit the TSC1/2 complex. Similarly, high ATP:AMP ratios inhibit AMPK, which may also lead to the inactivation of TSC1/2 and thus activation of Rheb/mTORC1 and consequently the translation of mRNAs important for cell growth. Despite its tumor-suppressive roles, AMPK increases the cytoplasmic levels of p27, thereby potentially promoting cell migration and/or autophagy. Our findings indicate that in normal cells Rheb can negatively regulate its activity by activating AMPK (red arrow) independently of mTORC1.

There is strong evidence indicating that the TSC1/2-Rheb-mTORC1 signaling network promotes tumorigenesis. However, we recently reported that aberrant Rheb activation could also contribute to disease independently of its ability to activate mTORC1—by regulating the activities of adenosine 5′-monophosphate-activated protein kinase (AMPK) and of the cyclin-dependent kinase inhibitor p27^KIP1 (p27).⁵ Below, we will discuss how these findings may have important implications for disease progression and treatments from TSC to cancer.

AMPK: An Ambivalent Role during Tumorigenesis?

The ability to adapt to environmental availability of energy sources is one of the most basic features needed for survival. It involves a multitude of steps, ranging from digestion of food, absorption and cellular uptake of sugars to intracellular metabolism that eventually yields adenosine-5′-triphosphate (ATP). AMPK is a multi-subunit protein complex that responds to increases in the adenosine-5′-monophosphate (AMP): ATP ratio by inducing catabolic processes that produce ATP and by inhibiting energy-consuming anabolic mechanisms, such as protein synthesis.^6–8 Furthermore, AMPK not only acts at the cellular level. In the hypothalamus, it stimulates food intake and thereby systemically increases energy levels.⁹

AMPK-mediated repression of protein synthesis results in the inhibition of cell growth. Repression of protein synthesis is accomplished by AMPK-induced phosphorylation and activation of TSC2 and consequently inhibition of Rheb/mTORC1 signaling.² Interestingly, we recently discovered an mTORC1-independent Rheb-to-AMPK feedback mechanism.⁵ A plausible explanation for its existence is that it provides a means by which Rheb controls its own activity in normal cells. However, in conditions where it is interrupted, Rheb is predicted to be highly active, leading to increased mTORC1 and AMPK activity (Figs. 1 and 2). For instance, such conditions exist in TSC where TSC2 or TSC1 is genetically inactivated.² Rheb-induced activation of mTORC1 facilitates cell growth in TSC and is consistent with tumorigenesis. On the other hand, Rheb-induced activation of AMPK is surprising and in principle incompatible with Rheb's oncogenic activity. In fact, as AMPK inhibits energy-consuming anabolic mechanisms supporting cell growth,^6,7 AMPK plays tumor suppressive roles. Accordingly, the potential tumor suppressor activity of AMPK has been therapeutically exploited in the context of cancer with AMPK-activating drugs such as metformin.¹⁰ However, in support of AMPK's potential tumorigenic role, AMPK activation positively correlated with survival and Cdk2 activation in Tsc2-null cells,¹¹ with proliferation and survival of prostate cancer cells,¹² and with p27-mediated cell survival under metabolic stress.^11,13 Similarly, we showed that Tsc2-null ELT3 cells proliferate in serum-free medium when AMPK is activated. Under these conditions, knockdown of Rheb reduced AMPK activity and cell proliferation, an effect that was paralleled by an increase in both the nuclear and cytoplasmic levels of p27, and occurred independently of mTORC1. Similarly, Rheb knockdown in several human colon cancer cell lines resulted in AMPK deactivation and upregulation of nuclear and cytoplasmic p27.⁵

Figure 2.

Interrupted Rheb-to-AMPK feedback signaling in tumorigenesis—model of a proposed mechanism. In the absence of TSC1/2 activity, constitutive activation of Rheb results in activation of mTORC1 and AMPK. We propose that AMPK activated via Rheb phosphorylates p27, thereby decreasing the nuclear while increasing the cytoplasmic levels of p27. By this mechanism, Rheb could promote cell cycle progression by reducing the nuclear levels of p27 (a), cell migration (b) and/or cell survival through autophagy (c).

p27 in the Cytoplasm is Pro-Tumorigenic

p27 prevents cell cycle progression mainly by binding to and inhibiting the kinase activity of cyclin A-Cdk2 and cyclin E-Cdk2 complexes in the nucleus.^14,15 Due to its function as an inhibitor of cell proliferation, p27 is considered to be a tumor suppressor. However, many studies have emerged indicating that the critical role of p27 in tumorigenesis may extend beyond its ability to regulate Cdks.^14–16 For instance, knock-in mice with a p27 variant that cannot interact with cyclins and CDKs develop tumors in multiple organs mostly at frequencies higher than corresponding p27-null animals.¹⁷ Thus, it seems that p27 suppresses tumorigenesis by inhibiting cyclin-Cdk activity in the nucleus, but carries out potential oncogenic functions in the cytoplasm.

We believe that the function of p27 in the cytoplasm could explain how Rheb-to-AMPK signaling contributes to tumorigenesis. Previous reports demonstrated that AMPK activation is responsible for the sequestration of p27 in the cytoplasm of Tsc2-null cells.^11,18 Our work similarly shows that AMPK activity and the levels of cytoplasmic p27 are positively correlated in Tsc2-null cells. Additionally, we observed that both AMPK and p27 are regulated by Rheb. Importantly, Rheb regulated both AMPK and p27 independently of mTORC1. Moreover, we found that AICAR (an AMPK activator) could inhibit the effect of Rheb depletion on p27 nuclear accumulation.⁵ Thus, our current hypothesis is that Rheb-to-AMPK signaling is responsible for the stabilization of p27 in the cytoplasm of Tsc2-null cells. How could cytoplasmic p27 promote tumorigenesis? Several lines of evidence indicate that p27's potential oncogenic activity in the cytoplasm seems to correlate with the effect of p27 on promoting cell migration.^14–16 Thus, cytoplasmic p27 could contribute to the metastasization of, in particular, advanced malignancies. In addition, p27 in the cytoplasm could affect cell survival. In fact, depending on the experimental conditions and/or cellular context, p27 exerts pro- or anti-apoptotic effects.¹⁴ The latter is of particular interest in association with tumorigenesis, as it may provide a means by which p27 could prevent cancer cell death. As demonstrated by Liang et al. p27 protected cells from metabolic stress-induced apoptosis.¹³ Intriguingly, metabolic cellular stress induced AMPK resulting in stabilization of p27 by phosphorylation on Thr198, which consequently promoted autophagy-mediated cell survival.¹³

Although there is still no evidence that Rheb could regulate any of the above discussed functions associated with cytoplasmic p27, we are tempted to speculate that a potential connection exists between the Rheb-to-AMPK feedback loop and p27-associated autophagy and tumorigenesis.

Autophagy, Rheb/AMPK/p27 and Energy Metabolism in Tumors

Autophagy serves as an important mechanism that allows energy production and survival under conditions of severe shortage of nutrients. At least in normal cells, it is coupled to a cellular repair and cancer prevention program by which aggregated proteins and damaged organelles are cleared,^19,20 and chromosomal integrity is protected.^21,22 These properties suggest that autophagy reduces age-related cellular dysfunctions including cancer initiation and promotes longevity.^19,23 However, even though autophagy may act tumor-suppressive during early stages of tumor initiation, autophagy-induced cell survival under stress may be a deleterious characteristic of established malignant tumors. For instance, autophagy may antagonize cell killing by chemotherapy.^13,19,20

Chemical stress is not the only negative pressure that cancer cells endure during their fight to survive. Metabolic stress and anoikis, i.e., cell death associated with cell detachment from the extracellular matrix, are additional hurdles to overcome.¹⁹ Metabolic stress in tumors arises in areas of reduced or intermittent blood supply²⁴ and is associated with cancer cell death in combination with severe hypoxia/anoxia, and with autophagy-mediated cell survival in less severe hypoxic environments.²⁵ A key element in the adaptation to low oxygen levels is hypoxia-inducible factor (HIF)-1. However, induction of tumor cell autophagy may occur both through HIF-1-dependent²⁵ and -independent²⁶ mechanisms. In addition to low levels of oxygen, poor tumor vascularization reduces the availability of glucose. Thus, energy-limiting conditions in hypoxic areas may, in principle, activate AMPK. However, as most cancer cells meet their energy requirements predominantly by aerobic glycolisis (“the Warburg Effect”),²⁷ ATP levels are likely not sufficiently low for AMPK to be highly activated. Nevertheless, this model is controversial and does not always apply. For instance, head and neck squamous cell carcinoma cells were reported to respond to hypoxia with an energy-depleting response as demonstrated by low ATP levels and activation of AMPK.²⁸ Furthermore, hypoxia may also activate AMPK without an increase of the AMP:ATP ratio.²⁹ Taken together, it is feasible to speculate that tumor cells adapt to hypoxia by HIF1- and/or AMPK-mediated autophagy. However, it is unclear whether this survival response consistently happens in tumors with or without additional stress such as chemotherapy.

AMPK may induce autophagy by reducing mTORC1 activity,³⁰ either via inactivation of TSC2, or by phosphorylation of Raptor, an mTORC1 component.³¹ Furthermore, as previously discussed, AMPK may induce autophagy by regulating levels and cytoplasmic localization of p27.¹³ Moreover, recent reports demonstrate that AMPK may promote autophagy by direct activation of Ulk1.^32,33 Taken together, AMPK has clearly been shown to promote autophagy under certain conditions. Therefore, can Rheb-to-AMPK signaling also promote AMPK-mediated autophagy? In TSC2-proficient cancer cells, we could expect that TSC2-dependent inactivation of Rheb links AMPK to inactivation of mTORC1 and autophagy. However, the scenario seems to be far more complex in tumors with inactivated TSC2. Under serum deprivation, our results obtained with Tsc2-null cells suggest that Rheb activates AMPK and induces cell proliferation by increasing Cdk2 activity through reduction of the nuclear levels of p27. Interestingly, we observed that under serum-free conditions both AMPK and mTORC1 were activated in Tsc2-null ELT3 and MEF cells, and in several human colon cancer cell lines. Furthermore, despite the lack of serum, ELT3 cells proliferated to some extent.⁵ Therefore, although AMPK was activated, we cannot conclude that serum deprivation induced widespread autophagy in Tsc2-null cells, neither by direct inactivation of mTORC1 by AMPK, nor by a Rheb-induced p27-dependent mechanism.⁵ However, Short et al. reported that AMPK inactivation in serum-deprived Tsc2-null cells enhanced their susceptibility to apoptosis.¹¹ Furthermore, p27 phosphorylated at T198 promoted autophagy and knockdown of p27 resulted in apoptosis.¹³ Since T198 of p27 is an AMPK target phosphorylation site,¹² it could be speculated that TSC2-null tumors with intrinsically high Rheb activity might respond to additional stress (such as therapeutic intervention) with more noticeable autophagy than tumors with submaximal Rheb-to-AMPK signaling. For instance, it is known that various chemical stresses also increase p27 levels,^13,34 and p27 plays an anti-apoptotic role in carcinoma and leukemic cell lines following drug treatment.¹⁴ Furthermore, therapy-induced tumor destruction is likely accompanied by extensive hypoxia. As hypoxia is an inducer of AMPK (see above), therapeutic interventions may provide means by which AMPK becomes further activated and thus potentially more readily induces autophagy than if only activated through Rheb. However, data that appear irreconcilable suggest that Rheb inhibits autophagy and sensitizes Tsc2-null cells to ER stress (also known as unfolded protein response)-induced apoptosis.³⁵ A possible explanation for these apparent discrepancies is that the kind of insult or level of stress defines whether Rheb activation will induce p27 localization-dependent cell proliferation, or cell survival via autophagy or apoptosis. Future studies should address in detail how and when different Rheb responses are engaged. Also, considering that cytoplasmic p27 could induce cell migration, it remains to be investigated whether Rheb signaling is able to promote this p27-dependent response.

In summary, we propose a model in which Rheb supports tumor development in addition to its activating effect on mTORC1 through mTORC1-independent mechanisms that involve AMPK and p27 (Fig. 2).

Rheb-to-AMPK Feedback Signaling in Malignant Tumors

In contrast to TSC/LAM, inactivating mutations in TSC1 or TSC2 appear to be rare in sporadic malignant tumors.^36,37 Nevertheless, TSC1 mutations in bladder cancer were described and deletions in the region of the TSC genes in several other sporadic epithelial cancers were found.³⁷ Alteration of TSC1/2 function also occurs by changes in the levels of gene expression, for instance by functional modification of TSC genes via promoter hypermethylation in breast cancer and oral squamous cell carcinoma.^38,39 Loss of TSC2 expression has also been found in endometrial carcinoma.⁴⁰ Furthermore, as Akt, ERK and RSK1 inactivate wild-type TSC2. Thus, oncogenic activation of upstream factors such as Ras or PI3K predicts low TSC1/2 activity.⁴¹

Taken together, we could predict that the Rheb-to-AMPK feedback functions similarly in TSC/LAM and malignant tumors with loss of TSC1/2 function. Regarding those malignant tumors with low TSC1/2 activity due to upstream activation of oncogenic signaling, it is possible that AMPK activation by Rheb may counteract some of the TSC1/2-repression. Such an effect may occur particularly in cells with minimal Wnt signaling activity, as GSK3, a negatively regulated component of the canonical Wnt pathway, enhances AMPK's activation of TSC2.⁴² However, through siRNA-mediated knockdown of Rheb, we demonstrated that, similar to Tsc2-null ELT3 and MEF cells, Rheb activates AMPK and increases the levels of cytoplasmic p27 in SW620, SW1116 and COLO 320 HSR human colon cancer cell lines under serum deprivation. Activating KRAS mutations in the SW620 and SW1116 lines are likely responsible for both low TSC1/2 activity (Fig. 1) and high Rheb-to-AMPK signaling activity.⁵

Therapeutic Targeting of Rheb and mTOR in TSC, S-LAM and Malignant Tumors

Even though Rheb plays oncogenic roles through the activation of mTORC1 in many types of cancer cells, its potential tumorigenic activity associated with AMPK and p27 is likely restricted to a smaller subset of cells and may become only relevant in the context of treatment and/or metabolic/hypoxic stress. Furthermore, we cannot ignore an extensive body of evidence suggesting that AMPK acts strongly as a tumor suppressor in various circumstances. For instance, in addition to repressing mTORC1 signaling via TSC2 activation or inhibition of Raptor (Fig. 1), AMPK may antagonize tumorigenesis by activating p53 and consequentially p21^WAF1, a cyclin-dependent kinase inhibitor.⁴³ Therefore, in such tumors, treatment strategies specifically targeting Rheb alone may be less favorable than those including additional pharmacologically stimulation of AMPK. However, the situation may be different in other malignant tumors and, especially, in TSC and S-LAM. Although our results await confirmation in in vivo models of the disease, we propose that therapeutic targeting of Rheb in TSC, S-LAM and some malignant tumors might be more favorable than targeting mTORC1. In fact, we showed that Rheb activates AMPK in Tsc2-null cells without obvious involvement of mTOR, in particular of mTORC1.⁵ Thus, mTOR/mTORC1 inhibitors, such as those under clinical and pre-clinical evaluation^44,45 are expected to only quench part of Rheb's tumorigenic activity without reducing AMPK's and p27's potential oncogenic functions. In this context, results from a recent clinical trial with rapamycin in TSC and S-LAM are consistent with tumor re-growth once treatment is discontinued.⁴⁶ Thus, it would be important to evaluate whether AMPK and p27 are responsible for the survival of tumor cells during the treatment. Alternatively, considering that high Rheb activity sensitizes cells to ER stress-induced apoptosis, it has been suggested that ER stress agents could be useful to kill tumor cells with low or absent TSC1/2 activity.³⁵

If the molecular signaling network driving tumorigenesis was less complex and the development of different human tumors was driven by similar molecular mechanisms, pre-clinical data obtained with one system would likely apply to another, thus facilitating the development of therapies applicable to a variety of human tumors, including TSC, S-LAM and malignancies. At this time, however, our current knowledge of the complex interplay between AMPK-regulated energy homeostasis and signaling pathways stimulating cell grow/proliferation is incomplete. Therefore, for a solid design of therapeutic strategies, more pre-clinical research is needed to understand Rheb-to-AMPK activation and contribution during tumorigenesis. Furthermore, the identification of a functional link between Rheb, AMPK and p27 in patient samples is challenging and remains to be shown.

Extra View to : Lacher MD, Pincheira R, Zhu Z, Camoretti-Mercado B, Matli M, Warren RS, Castro AF. Rheb activates AMPK and reduces p27Kip1 levels in Tsc2-null cells via mTORC1-independent mechanisms: implications for cell proliferation and tumorigenesis. Oncogene. 2010;29:6543–6556. doi: 10.1038/onc.2010.393.