Liste des participants
Joseph Avruch, Yves-Alain Barde, John Blenis, David Carling, Marian Carlson, Bruce A. Edgard, Rick Firtel, Kun-Liang Guan, Michael N. Hall (Organisateur), Pierre Léopold, Thomas P. Neufeld, Marius Pende, Matthias Peter, Jacques Pouyssegur, Anne Ridley, David D. Sabatini Jr (Organisateur), Nahum Sonenberg, James Umen, Anders Zetterberg
by Michael Hall and David Sabatini
May 9-15, 2005
Michael Hall (Biozentrum, University of Basel, Basel, Switzerland) began the meeting on May 10th by introducing TOR. TOR is an atypical serine/threonine kinase, the target of the immunosuppressant and anti-cancer drug rapamycin, and a central controller of eukaryotic cell growth. TOR in yeast and mammals is found in two structurally and functionally distinct multiprotein complexes, TORC1 and TORC2. These two types of TOR provide a molecular basis for the complexity of TOR signalling. Yeast TORC1 contains KOG1, LST8 and TOR1 or TOR2, and is rapamycin sensitive. The mammalian counterpart of TORC1, mTORC1, contains raptor (the KOG1 ortholog), mLST8 (GbetaL), and mTOR. TORC1 in yeast and mammals mediates temporal control of cell growth by regulating several cellular processes including translation, transcription, ribosome biogenesis, nutrient transport and autophagy. Recent findings on the mechanism by which TORC1 regulates ribosome biogenesis were presented. TORC1 regulates ribosomal protein (RP) gene expression via the Forkhead-like factor FHL1 and the two co-factors IFH1 (a co-activator) and CRF1 (a co-repressor). TORC1, via PKA, negatively regulates the kinase YAK1 and maintains CRF1 in the cytoplasm. Upon TOR inactivation, activated YAK1 phosphorylates and activates CRF1. Phosphorylated CRF1 accumulates in the nucleus and competes with IFH1 for binding to FHL1 at RP gene promoters, and thereby inhibits transcription of RP genes. This constitutes a signaling mechanism linking an environmental sensor to ribosome biogenesis.
Yeast TORC2 contains AVO1, AVO2, AVO3, BIT61, LST8, and TOR2, and is rapamycin insensitve. mTORC2 contains mAVO3 (rictor), mLST8, and mTOR. TORC2 in yeast and mammals mediates spatial control of cell growth by regulating the actin cytoskeleton. Recent results indicating that TORC2 controls more than just the actin cytoskeleton were presented. TORC2 also regulates transcription. A large part of this transcriptional regulation is the inhibition of calcineurin dependent transcription. TORC2 is also required for sphingolipid synthesis. Thus, TORC2, like TORC1 appears to regulate several cellular processes via different effector pathways.
David Sabatini (Whitehead Institute/MIT, Cambridge, MA, USA) continued on the theme of TOR with a presentation on the regulation of growth by the mTOR pathway. He began with a review of the mammalian TOR pathway and focused on the identification of two distinct mTOR-containing complexes, called mTORC1 and mTORC2. mTORC1 is composed of mTOR, raptor, and mLST8, while mTORC2 consists of mTOR, rictor, and mLST8. The mTORC1 complex phosphorylates S6K1 and the strength of the interaction between the mTOR and raptor components of this complex is regulated by upstream signals such as nutrients. After the introduction, Sabatini concentrated his talk on the potential role of mTORC2 in phosphorylating the hydrophobic motif of Akt/PKB, a cell survival kinase that becomes hyperactive in cells that have lost the PTEN tumor suppressor. He proposed that mTORC2 is the missing ‘PDK2’ for Akt/PKB and presented loss of function results from human and Drosophila cells as well as biochemical data supporting this idea.
The second half of his presentation was dedicated to describing a new mechanism of action of rapamycin. He showed data suggesting that when cells are cultured for prolonged periods of time in rapamycin the assembly of mTORC2 is inhibited and that this leads to a suppression of Akt/PKB signaling. Curiously, rapamycin inhibits mTORC2 assembly to different extents in different cells so that rapamycin in not a universal inhibitor of Akt/PKB signaling. The cellular properties that determine if mTORC2 assembly is sensitive to rapamycin are not understood but preliminary work suggested that the presence or absence of PTEN is not important. Sabatini proposed that Akt/PKB inhibition by rapamycin may account for some of the beneficial and toxic effects of the drug in people and that this property of the drug should be considered when using the drug clinically.
John Blenis (Harvard Medical School, Boston, MA, USA) began the afternoon session by reviewing what he discussed 4 years ago at the first TOR meeting at Les Treilles, the molecular basis for activation of an important mTOR effector, the 70kDa S6 kinase or S6K1. He then briefly summarized the tremendous progress made by many of the scientists attending the meeting towards our current understanding of the molecular basis for mTOR signaling and cell growth control by energy sufficiency, nutrients such as branched chain amino acids and integration of signals from the phosphatidylinositol 3-kinase pathway. New data was then provided showing for the first time how Ras activation contributes to mTOR signaling independent of PI3-kinase, for example by tumor promoting phorbol esters. The 90kDa ribosomal S6 kinase or RSK, an enzyme distinct from S6K1, is directly activated by the ERK-MAP kinases. RSK then phosphorylates the tuberous sclerosis tumor suppressor complex protein 2 (TSC2) at sites that disrupt its ability to antagonize mTOR signaling. Thus these phosphorylation events provide a mechanism for cross-talk between the Ras pathway and mTOR signaling, and contribute to activation of mTOR effectors such as S6K1 and eIF4E. The literature was briefly reviewed which demonstrates that the 40S ribosomal protein S6, a well-known target for S6K1 in vivo, is not the critical target involved in cell growth control. He then discussed his labs’ work with a novel S6K1 specific target named SKAR. SKAR is a S6K1 target with homology to a RNA binding protein called Aly/REF. Importantly, RNAi-based evidence was provided suggesting that like S6K1, SKAR is also involved in cell growth control. The Blenis lab is currently determining how SKAR regulates cell growth and how S6K1 regulates SKAR. Finally, Dr. Blenis presented new data showing that the eukaryotic initiation factor 3 complex (eIF3) acts as a molecular scaffold upon which mTOR binds following amino acid and mitogen stimulation. Once bound mTOR then phosphorylates the eIF4E-binding protein 1 (4E-BP1) and S6K1, found associated with the eIF4E/eIF3 translation pre-initiation complex (PIC). This results in their release from eIF4E and eIF3, respectively. Release of 4E-BP1 provides a docking site on the eIF4E m7G-cap binding protein for eIF4G, an essential scaffold protein required for assembly of a functional translation initiation complex. Release of S6K1 from the PIC is necessary for its full activation, which then results in it phosphorylating eIF4B and the resulting binding of eIF4B to the translation initiation complex. eIF4B is an important regulatory subunit of the eIF4A/B RNA helicase, an enzyme that unwinds complex stem-loop structures in the 5’ untranslated region (UTR) of mRNA and thus promotes efficient translation of mRNAs with long, complex 5’ UTRs such as VEGF, cyclin D and c-myc. These data provide the first evidence of how mTOR signaling is activated in response to nutrients and growth factors and also provide a novel and direct connection between S6K1 and translation initiation.
Joseph Avruch (Massachusetts General Hospital, Boston MA, USA) continued the session by focusing on the regulation of mTOR signaling by amino acids and the Rheb GTPase, and the mechanism by which Rheb overexpression overcomes the deactivation of mTOR caused by amino acid withdrawal.
TOR functions in two physically distinct complexes, TORC1 and TORC2; both complexes contain the LST8 polypeptide, however TORC1 contains raptor whereas TORC2 contains AVO3/rictor (and other components, at least in S. cerevisae). The rapamycin/FKBP12 complex is known to associate directly with TORC1 but is unable to bind TORC2, and TORC1 is thought to be responsible exclusively for the phosphorylation of S6 Kinase and 4E-BP at their rapamycin-sensitive sites (e.g., S6K[Thr412]). The TOS motifs of S6K and 4E-BP are necessary for the binding of these mTOR substrates to raptor, and deletion/mutation of the TOS motif of S6 Kinase and 4E-BP reduces the in vivo phosphorylation of these polypeptides at their rapamycin-sensitive sites by >95%. Raptor-independent phosphorylation of S6K(Thr412) becomes evident only when inactivation of the TOS motif is combined with deletion of the S6K carboxyterminal autoinhibitory domain; this doubly mutant S6K remains responsive to insulin much as wildtype S6K, but in contrast to wildtype S6K, the doubly mutant S6K is entirely resistant to inhibition by rapamycin. Amino acid withdrawal, which inhibits wildtype S6K as completely as does rapamycin, has only a modest inhibitory effect on the doubly mutant S6K. Recent results from Ali and Sabatini (JBC 280, 19445, 2005) indicate that the activating input to the doubly mutant S6K is entirely attributable to TORC2. Thus it appears that TORC1 and TORC2 are both highly responsive to insulin; in contrast, amino acid withdrawal causes a profound inhibition of signaling from TORC1, but has only a modest inhibitory effect on signaling from TORC2 (at least toward the S6K double mutant).
Genetic and biochemical evidence have established that the major upstream activating input provided by the insulin/IGF1 receptor pathway to S6K, 4E-BP and cell growth is mediated by PKB, through its negative regulation of the Rheb GAP activity of the Tuberous Sclerosis heterodimer (TSC1/TSC2). The activation state/GTP charging of Rheb in vivo is controlled primarily by the GAP activity of the Tuberous Sclerosis complex and phosphorylation of TSC2 by PKB inhibits TSC-GAP activity. Conversely, AMPK, activated in response to energy deficiency, phosphorylates TSC2 at other sites and stimulates the TSC GAP activity. Murine cells lacking TSC2 exhibit very high levels of Rheb GTP charging and constitutive activation of S6K; the latter is unresponsive to further stimulation by insulin, but retains normal sensitivity to inhibition by rapamycin, establishing the requirement for TORC1 for these effects of TSC2 deletion and Rheb-GTP. The elevated S6K activity of TSC2-deficient cells exhibits significant, although not absolute, resistance to inhibition by amino acid withdrawal, indicating that amino acid control is exerted, at least in part, independently of the TSC complex, and at a site distinct from rapamycin. Transient overexpression of wildtype Rheb in mammalian cells can further activate S6K in nutritionally replete cells and, in a dose-dependent manner, prevent/overcome the dephosphorylation of wildtype S6K and 4E-BP otherwise seen after withdrawal of medium amino acids.
Avruch described recent studies examining how Rheb regulates mTOR signaling and the mechanism by which Rheb rescues mTOR signaling from the effects of amino acid depletion. Mutational analysis indicates that Rheb stimulation of signaling to S6K and 4E-BP requires intact Rheb switch 1 and switch 2 loops, and an ability to bind guanyl nucleotide; Rheb mutants (e.g., S20N or D60I) that are unable to bind guanyl nucleotide are not only devoid of stimulatory activity, but also exhibit a moderate inhibitory effect on coexpressed S6K in amino acid replete medium. The possibility that Rheb-GTP might act directly within the TORC1 complex was suggested by the finding that overexpression of mTOR or LST8 is capable of coprecipitating the endogenous TSC complex. In fact endogenous mTOR and raptor coprecipitate with recombinant Rheb; similarly, recombinant wildtype Rheb coprecipitates coexpressed mTOR, and in much larger amounts than are found with recombinant v12Ha-ras or v12rap1b. Surprisingly, the association of Rheb with mTOR does not require Rheb GTP charging; the Rheb mutants S20N and D60I, which are unable to bind guanyl nucleotides in vivo (or in vitro), bind mTOR more avidly than does wildtype Rheb. Reciprocally, RhebQ64L, a mutant that is nearly 90% GTP charged in vivo, binds mTOR less well than does wildtype Rheb, which exhibits 50-60% GTP charging during transient overexpression.
The functional significance of the Rheb-mTOR interaction was examined by comparison of the protein kinase activity of mTOR polypeptides bound to wildtype and mutant Rheb. Recombinant Rheb coprecipitates an endogenous kinase that phosphorylates S6K at Thr412 as well as 4E-BP1; this Rheb-associated kinase activity is greatly enhanced by coexpression of Rheb with mTOR. In contrast, when assayed at the same polypeptide concentration, the mTOR polypeptides that coprecipitate with the nucleotide-deficient Rheb mutants S20N or D60I, as well as the very small amounts of mTOR recovered with v12Ha-Ras, are essentially devoid of S6K(Thr412)/4E-BP kinase activity. In contrast, mTOR polypeptides bound to RhebQ64L exhibit approximately two-fold higher specific activity as those bound to wildtype Rheb. These findings argue strongly that an association of mTOR with Rheb-GTP is necessary for the acquisition of mTOR kinase activity. It is unclear whether continued association of Rheb is required to sustain mTOR kinase activity.
The Rheb binding site on mTOR maps to the upper lobe of the mTOR catalytic domain. This is a segment whose primary sequence is highly conserved among the Ptd Ins 3’OH kinase-related protein kinases (PIKKs); in fact the homologous segments of ATM and ATR bind recombinant Rheb during transient expression to a similar extent as does that from mTOR. In contrast, Rheb shows no binding to several conventional protein kinase catalytic domains. In addition, recombinant Rheb was found to bind to LST8 and to raptor, the latter through its carboxyterminal beta propellor, WD domain. The binding of Rheb to AVO3/rictor has not yet been assessed. These multiple sites of Rheb binding within TORC1 were confirmed by in vitro binding assays using purified recombinant polypeptides. Those studies confirmed that the ability of Rheb to bind specifically to the mTOR carboxterminal segment does not require Rheb guanyl nucleotide charging, and is actually inhibited by GTP. Rheb also binds specifically to the aminoterminal ras-binding domain of cRaf-1, but in contrast to Ha-ras, whose binding to this segment of Raf is strongly stimulated by GTP charging, Rheb binding to cRaf-1 is not stimulated, and may be moderately inhibited by GTP charging. In parallel experiments, Rheb-GTP was shown to bind to the TSC2 GAP domain much better than Rheb-GDP. Taken together, these results indicate that in contrast to other ras-like GTPases, GTP charging is not required for the binding of Rheb to its effector, mTOR, but is required for Rheb activation of the mTOR kinase.
An important feature of the interaction between Rheb and mTOR is that, as with mTOR signaling to S6K and 4E-BP, the binding of Rheb to endogenous and recombinant mTOR is reversibly inhibited by withdrawal of all extracellular amino acids or just leucine. The effect of amino acid withdrawal is not attributable to changes in Rheb-GTP charging; amino acid withdrawal does not alter the GTP charging of recombinant Rheb. Moreover, the binding of mTOR to Rheb mutants that are unable to bind guanyl nucleotide in vivo is also inhibited by amino withdrawal. The inhibitory effect of amino acid withdrawal is exerted through an action on mTOR, at a site largely distinct from that responsible for the binding of Rheb; deletion of the larger, carboxyterminal lobe of the mTOR catalytic domain eliminates the inhibitory effect of amino acid withdrawal on Rheb binding, without altering Rheb binding per se. The lesser ability of the mTOR catalytic domain to bind Rheb after amino acid withdrawal does not persist after extraction and purification of the mTOR polypeptide. Amino acid withdrawl may generate an inhibitor of the Rheb-mTOR interaction that interferes with the signaling function of TOR complex1.
These findings suggest that the ability of recombinant Rheb, when greatly overexpressed, to rescue S6K and 4E-BP from amino acid /leucine withdrawal, may be attributable simply to the expression of sufficient amounts of GTP-charged Rheb to overcome the inhibition of the Rheb-mTOR interaction engendered by amino acid withdrawal. The relationship of these findings, obtained using purified polypeptides in vitro and overexpressed recombinant Rheb in vivo, to the mode of Rheb action under physiologic conditions, is the focus of ongoing experiments.
In the last talk of the opening day Kun-Liang Guan (University of Michigan, Ann Arbor, MI, USA) discussed the regulation of TSC2 by various signaling pathways. He summarized the literatures on phosphorylation of TSC2 by many kinases, including AKT, RSK, MAPKAPK2, AMPK, and ERK. Dr. Guan presented new data that TSC2 is also regulated by GSK3-dependent phosphorylation. Under energy starvation, AMPK phosphorylates TSC2 and creates priming site for the subsequent TSC2 phosphorylation by GSK3. Phosphorylation of TSC2 by AMPK and GSK3 plays an important role in the coordination between cell growth and cellular energy levels. Dysregulation of the TSC-mTOR pathway results in apoptosis under energy starvation conditions. Glucose deprivation induces apoptosis in TSC2-/- or TSC1-/- cells. Importantly, glucose also induces apoptosis in TSC2-/- cells with re-expression of the AMPK/GSK3 phosphorylation mutant TSC2 but not in those with re-expression of the wild type TSC2. These observations establish the functional significance of TSC2 phosphorylation by AMPK and GSK3. The glucose-deprivation induced apoptosis can be effectively blocked by rapamycin, suggesting that uncontrolled activation of the mTOR pathway contributes to apoptosis under energy starvation. Dr. Guan also discussed TSC2 degradation by the ubiquitination pathway. He showed data that TSC2 associates with HERC1, which is an E3 ubiquitin ligase. TSC1 stabilizes TSC2 by preventing HERC1 to ubiquitinate TSC2. Recent studies have established that high levels of dysregulated mTOR activity are associated with several hamartoma syndromes, including tuberous sclerosis complex caused by mutations in the TSC1 or TSC2 tumor suppressor gene, the Cowden Disease caused by mutation in the PTEN tumor suppressor, and Peutz-Jeghers syndrome caused by mutation of the LKB1 tumor suppressor. Potential connections between dysregulation of mTOR pathway and cellular hypertrophy and tumor growth were discussed.
On the morning of May 11th Anne Ridely (Ludwig Institute of Cancer Research at University College London, London, UK) began the session on Cell Shape by discussing how the Rho family of GTPases play important roles in regulating cell shape and migration through their effects on the actin cytoskeleton and cell adhesion. There are 20 members of the Rho family in mammals, of which the best characterized are Rho, Rac and Cdc42. Rac proteins are believed to be essential for cell migration based on the effects of expressing a dominant negative Rac1 mutant. Rac proteins stimulate actin polymerization at the plasma membrane, driving forward protrusion. Mammals have three closely related Rac isoforms, Rac1, Rac2 and Rac3. Macrophages express Rac1 and Rac2, and surprisingly it was found that macrophages lacking Rac isoforms had altered cell shape but were still able to migrate. Dominant negative Rac1 induced cell rounding and prevented migration of Rac1-deficient macrophages, indicating that it inhibits the activation of other pathways in addition to Rac proteins.
RhoE is an unusual Rho family member in that it binds but does not hydrolyse GTP, and is therefore not regulated by the GTP/GDP switch of GTPases. Increased RhoE expression induced loss of actin stress fibres in cultured fibroblasts and epithelial cells, thereby acting antagonistically to RhoA, which stimulates stress fibre formation. RhoE bound directly to the RhoA-activated serine/threonine kinase ROCK I, preventing it from phosphorylating its targets. RhoE also inhibited cell cycle progression, in part by preventing translation of cyclin D1 mRNA. This response was not mediated by ROCK inhibition, and mechanisms whereby RhoE could affect cyclin D1 expression were discussed. Since RhoE is not a GTPase, its activity must be regulated in other ways. RhoE was found to be phosphorylated by ROCK I on 7 sites near its amino- and carboxy-termini. RhoE phosphorylation enhanced its stability and reduced its association with membranes. RhoE phosphorylation was stimulated by PDGF in fibroblasts, and coincided with its ability to induce loss of stress fibres, suggesting that RhoE activity is primarily regulated by phosphorylation.
Richard Firtel (UCSD, San Diego, CA, USA) completed the Cell Shape session by discussing chemotaxis. Chemotaxis, or directed cell movement in response to a chemoattractant gradient, is a basic property of many eukaryotic cells including leukocytes and Dictyostelium cells. Cells are able to respond to relatively shallow (less than 2%) differences in the concentration of chemoattractant between the front and the back of the cell. Part of the meeting focused on understanding the mechanisms by which cells are able to polarize (extend themselves in the direction of the chemoattractant gradient) and initiate directed cell movement. Rick Firtel described studies in his laboratory on Dictyostelium cells that have been aimed at understanding how these processes are regulated. To respond to shallow chemoattractant gradients, cells must be able to amplify the shallow extracellular chemoattractant gradient into a very steep intracellular gradient of signaling molecules that leads to directed F-actin polymerization at the side of the cell closest to the chemoattractant source while the posterior of the cell must be able to retract itself, causing the release of cell-cell contacts at the posterior, a process regulated by myosin II. In neutrophils and Dictyostelium cells, chemotaxis is regulated through serpentine, G-protein-coupled receptors. Firtel reviewed studies in which his group demonstrated that phosphatidylinositol 3-kinase (PI3K) is both a major component of the directional sensing machinery and critical for F-actin polymerization. PI3K, in turn, is regulated via Ras through a Ras-GTP binding site on PI3K. When Ras function is abrogated through a combination of knockout mutations of Ras exchange factors and expression of dominant negative Ras, cells lose the ability to directionally sense, indicating that Ras is an essential component of a cell’s ability to respond to chemoattractant gradients. Firtel then focused on how a new, stable leading edge is formed. He showed that cells amplify small differences in the response between the front and back of the cell through a series of positive feedback loops. By examining mutant strains and using GFP reporters to examine the spatial and temporal responses of signaling molecules in cells, Firtel’s group showed that unstimulated cells have basal levels of PI3K and a Rac exchange factor, RacGEF1, which is required for Rac activation and F-actin polymerization. An initial signal activates Ras and, through Ras, PI3K and RacGEF1. These processes trigger initial production of PI(3, 4, 5)P3 and recruitment of PH domain-containing proteins and the activation of Rac and F-actin polymerization, respectively. The F-actin recruits additional PI3K and RacGEF1, leading to increased PI3K activity and Rac-GTP and, through these, additional F-actin. These feedback loops continue establishing a robust response, a process in Dictyostelium that takes only 5-8 seconds. They discovered that these same pathways and positive feedback loops also regulate random cell movement in which their activation does not require a ligand or heterotrimeric G proteins. Blocking any component of this regulatory network leads to a cessation of pseudopod extension and the loss of the ability to localize “leading edge components” to the cell cortex. Firtel also showed a linkage between the TOR Complex 2 (TORC2) and cell polarity and chemotaxis in Dictyostelium. Previous studies in his and Devreotes’ laboratories in the 1990s identified proteins that are orthologs of yeast AVO1 and AVO3/Rictor, components of yeast and mammalian TORC2, respectively. They revealed through mass spectrometry analyses and coimmunoprecipitation the presence of a TORC2 in Dictyostelium and demonstrated that null mutations in TORC2 components result in cell polarity and chemotaxis defects and a loss of robust chemoattractant-stimulated activation of Akt and an Akt-related enzyme, PKBR1, which parallels the role of TORC2 in other systems and the findings from the Hall and Sabatini labs. Furthermore, Dictyostelium cells form a multicellular organism through chemoattractant-mediated aggregation. To accomplish this, cells signal to each other through a relay mechanism in which the chemoattractant, cAMP, activates adenylyl cyclase through the G-protein-coupled chemoattractant receptor, leading to further cAMP production. In association with Carole Parent’s laboratory, Firtel’s group discovered that TORC2 is required for the chemoattractant activation of adenylyl cyclase. Their data suggest that TORC2 functions as an integrator of aggregation and multicellular development in this organism through the regulation of a signal relay as well as chemotaxis.
On the afternoon of May 11th Anders Zetterberg (Karolinska Institutet, Cancer Center Karolinska, Stockholm, Sweden) initiated the session on growth control and cell cycle checkpoints in G1. Cell growth (increase in cell mass) and cell cycle progression are two loosely co-ordinated processes, which are under separate control and can be dissociated from each other. Growth control represents the interaction between the cell and its environment, while cell cycle control reflects the co-ordination between cell size, DNA replication and cell division (checkpoints). Mitogenic signals are transduced into the cell by a number of different interacting pathways, converging upon several key regulatory molecules that form a link between growth control and cell cycle control. Tumour cells exhibit defects in both growth control and in cell cycle control. Defects in growth control lead to the abnormal accumulation of cells through uncontrolled proliferation, whereas defects in cell cycle checkpoints lead to the genetic instability seen as aneuploidy and chromosomal rearrangements.
One fundamental control point in the cell cycle of mammalian cells is the restriction point (R), which operates stringently in normal cells but is defective in tumour cells. R is defined as the point in G1, after which the cell can complete a division cycle in the absence of growth factors. In cycling cells R divides G1 into two physiologically different intervals, a post-mitotic interval of G1 (G1-pm) of relatively constant length (3-4 hours), from which the cells rapidly exit to Go in the absence of growth factors, and a pre-S-phase interval of G1 (G1-ps) of variable length (1-10 hours), from which the cells can enter S in the absence of growth factors. Passage through R is not directly followed by entry into S after a set time interval. Some cells enter S shortly (within 1 hour) after passage through R, while other cells remain in G1 (in G1-ps) for many hours (up to 10 hours) after passage through R before entering into S. Passage through R and entry into S thus involve at least two separate checkpoints.
We have studied physiological and molecular aspects of R, focusing on the relationship between R and the phosphorylation of the retinoblastoma protein (pRb-105). Individual cells in asynchronously growing, unperturbed cell populations were analysed by a combination of time-lapse video-microscopy, quantitative cytometry and immunocytochemistry. This single cell approach, enabling the precise timing of transition events in G1, revealed that passage through R and Rb-phosphorylation represent two different checkpoints in G1. We used four independent criteria for pRb-phosphorylation, namely
(1) phosphorylation of cyclin D/CDK4-specifik epitopes in pRb, (2) changes in anchorage of pRb to nuclear structures, (3) release of E2F-1 from pRb and (4) expression of cyclin E. All of these four events were found to occur after passage through R, usually 1 to 2 hours before entry into S-phase. Furthermore, passage through R could occur in the presence of the CDK-inhibitor Roscovitine. Thus, R is located upstream of pRb-phosphorylation in G1, and these two checkpoints do not represent the same molecular mechanism, as has been previously suggested. Although the molecular mechanism underlying the control of R is still unclear, data will be presented showing a relationship between anchorage dependence, passage through R and ability to growth in cell size.
James Umen (The Salk Institute, La Jolla, CA, USA) continued the discussion on cell cycle and size control with a presentation on cell size control in Chlamydomonas reinhardtii. Maintaining cell-size homeostasis is fundamental for the growth and development of all organisms. The Chlamydomonas reinhardtii cell cycle has a long G1 period during which growth is unrestricted and cells can enlarge by up to ~30 fold. The G1 period is followed by a rapid succession of S and M phases whose number is determined by mother cell size and which produce 2n uniform-sized daughters. Previous studies by Umen have shown that a deletion of MAT3 that encodes a retinoblastoma (RB) tumor suppressor homologue in Chlamydomonas reinhardtii results in tiny daughter cells with defects in size-dependent cell cycle progression. Screens for additional size control mutants have identified genes that act upstream of MAT3 in a linear pathway that functions as a cell size sensor. In mat3 mutants expression of critical cell cycle targets is thought to become deregulated. A suppressor screen has identified mutations that partially or completely block the unregulated cell division that occurs in mat3 mutants. The strongest suppressors in this screen were loss of function mutations in DP1 and E2F1, single copy genes that encode homologs of DP/E2F transcription factors. Although the e2f1 and dp1 mutations suppress mat3, they display asymmetric phenotypes with respect to cell size and cell cycle control. The dp1 mutations are much more severe, resulting in a large size phenotype, whereas the e2f1 mutations seem to restore wild type size control. This paradoxical result might be explained if the DP1 protein is still partially active in the e2f mutant background, but not vice versa. Interestingly, at least one partial suppressor of mat3, smt7, interacts genetically with e2f1 to give a large-cell phenotype.
Chlamydomonas has many multicellular relatives, including Volvox carteri. Volvox has evolved a germ/soma dichotomy that involves the production of large reproductive cells and small somatic cells during embryogenesis. Interestingly, cell size is the trigger that leads to the germ/soma differentiation decision: Large cells always become germ cells and small cells always become somatic. This observation suggests that the Chlamydomonas size-sensing pathway has been coopted by Volvox to mediate germ/soma differentiation.
On the morning of May 12th the session on AMPK and nutrient sensing began with a presentation by David Carling (MRC, London, UK). He introduced the AMP-activated protein kinase (AMPK) cascade and emphasized its role as a system for responding to changes in ATP levels within the cell. Depletion of ATP causes a rise in AMP and the resulting increase in the AMP:ATP ratio activates AMPK. Once activated AMPK has diverse actions, but the overall effect is to switch-off energy utilising pathways and switch-on energy generating pathways. In this way, AMPK acts to coordinate cellular energy needs with metabolism. Dr. Carling briefly showed some published data demonstrating that AMPK phosphorylates mTOR on threonine 2446 in response to nutrient deprivation. Phosphorylation at this site appears to antagonise phosphorylation of serine 2448 (John Blenis had previously shown data that indicates that S6KI phosphorylates serine 2448). For the major part of his talk, Dr. Carling focussed on the regulation of AMPK by phosphorylation. AMPK is activated by phosphorylation of threonine 172 within the activation loop segment of the catalytic subunit. Recent work by a number of groups has identified LKB1 as an upstream kinase. He pointed out that inactivating mutations in LKB1 lead to a rare hereditary form of cancer, termed Peutz-Jeghers syndrome, raising the intriguing possibility that AMPK may be itself involved in cell proliferation. Unpublished data was shown which indicates that in addition to LKB1, AMPK is also activated by calcium calmodulin-dependent protein kinase kinase (CaMKK). Based on these new findings, Dr. Carling suggested that AMPK may play a role in the calcium signalling and that AMPK may be regulated by different upstream signalling pathways involving either LKB1 or CaMKK. Finally, Dr. Carling showed some results describing the effect of mutations within the 2 subunit of AMPK. Naturally occurring mutations in human 2 lead to cardiac hypertrophy associated with Wolff-Parkinson-White syndrome. Interestingly, these mutations also lead to a dramatic increase in glycogen storage in heart. An increase in skeletal muscle glycogen is seen in a breed of pig harbouring a mutation in the 3 subunit of AMPK. Although the mechanism underlying the development of the cardiac disease remains unclear, Dr. Carling reported that studies on transgenic mouse models seemed to suggest that the accumulation of glycogen caused a secondary effect leading to inhibition of total AMPK activity and that this may then lead to hypertrophy, possibly via a mechanism involving decreased phosphorylation of mTOR on threonine 2446 and increased mTOR activity.
Before the excursion of the afternoon of May 12th, Marian Carlson (Columbia University, New York, NY, USA) kept the focus on nutrient sensing with the last presentation of the day, a discussion on the Snf1/AMP-activated protein kinase (AMPK) cascade. The Snf1/AMPK family is widely conserved in eukaryotes and is important for transcriptional and metabolic regulation in response to stress. In mammalian cells, AMPK is activated by stresses causing reduced energy availability and by hormones. AMPK has a major role in controlling glucose and lipid metabolism and serves to maintain the cellular energy balance. In the yeast Saccharomyces cerevisiae Snf1 protein kinase regulates gene expression and metabolism in response to stress, particularly carbon stress, and Snf1 is required for transcription of a large set of genes in response to glucose limitation and for adaptation to growth on nonpreferred carbon sources. The Carlson lab has shown that the Snf1 kinase cascade comprises three Snf1-activating kinases, Pak1, Tos3, and Elm1, which phosphorylate the activation-loop threonine. These three kinases have overlapping functions in vivo, and all three genes must be deleted to abolish Snf1 catalytic function. She spoke about the roles of these upstream kinases in regulating Snf1 activity and its subcellular localization.
In mammals, AMPK is activated by LKB1, an ortholog of the yeast upstream kinases. Carlson presented evidence that LKB1 activates Snf1 in yeast cells, indicating that the kinase cascade is functionally conserved between yeast and mammals. She also showed that another homolog of the yeast upstream kinases, Ca2+/calmodulin-dependent kinase kinase (CaMKK), also activates Snf1 in yeast. These findings indicate that CaMKK is a functional member of the Snf1/AMPK kinase family and support CaMKK as a likely candidate for an AMPK kinase in mammalian cells. Analysis of the function of the mammalian kinases in yeast provided insight into the regulation of Snf1; when activated by LKB1 or CaMKK Snf1 activity was significantly inhibited by glucose, suggesting that glucose signals are mediated by a factor other than the activating kinases. Protein phosphatase 1 was discussed as a possible candidate.
On the morning of May 13th, Thomas Neufeld (University of Minnesota, MI, USA) began the session on Organismal Growth and discussed work from his lab characterizing the role of TOR in regulating distinct membrane trafficking events in Drosophila. In response to nutrient withdrawal or TOR inactivation, a dynamic process of membrane reorganization known as autophagy is induced, in which bulk cytoplasm and organelles are engulfed in specialized vesicles and transported to the lysosome/vacuole for hydrolytic digestion. The degradation products of autophagy can be either oxidized for energy production or recycled as building blocks for macromolecular synthesis, and can provide an essential source of cellular nutrients under starvation conditions. Dr. Neufeld showed that insulin/PI3K and TOR signaling is necessary and sufficient to suppress autophagy during Drosophila development. Expression of factors such as PI3K or Rheb suppressed starvation-induced autopahgy, whereas negative regulators of TOR such as TSC1 and TSC2 induced autophagy in fed animals. Surprisingly, S6K was found to promote rather than suppress autophagy, and its downregulation by TOR may act to self-limit the autophagic response during chronic starvation. Autophagy was shown to play a significant, context-dependent role in mediating the growth effects of TOR. Under starvation conditions, cells incapable of autophagy were shown to have a growth advantage over their wild type neighbors, whereas in cells lacking TOR disruption of autophagy results in a further decrease in cell growth. Induction of autophagy independent of TOR activity resulted in a marked growth inhibition.
A possible role for TOR in regulating endocytosis was also discussed. This function was first suggested from the results of enhancer/suppressor screens in Drosophila, which identified the clathrin uncoating ATPase Hsc4 as a TOR-interacting gene. This role was confirmed in subsequent studies using endocytic tracers, which found that bulk endocytosis is significantly stimulated in cells with activated TOR signaling, and is suppressed in TOR mutants. Inactivation of TOR also resulted in increased endocytic degradation of nutrient transporters and growth factor receptors. Genetic disruption of endocytosis resulted in increased cell growth and PI3K activity. Together these results demonstrate that TOR regulates multiple processes involving membrane trafficking, which have a significant impact on cell growth.
Bruce Edgar (Fred Hutchinson Cancer Research Center, Seattle, WA, USA) continued the session on organismal growth by discussing targets of Myc and Rheb in Drosophila and the role of endocycles in growth control. The regulation of ribosome numbers is thought to be a critical mode of cellular growth control. Synthesis of ribosomal RNA (rRNA) is a limiting step in ribosome biogenesis and rates of rRNA synthesis are generally altered depending on a cell’s growth status. While studies in unicellular systems have addressed the mechanisms by which this occurs, few studies have applied a genetic approach to examine growth-dependent control of rRNA synthesis in metazoans. Dr. Edgar’s group has found that Drosophila Myc (dMyc) is a critical regulator of rRNA synthesis. Expression of dMyc is both necessary and sufficient to control rRNA synthesis and ribosome biogenesis during larval development. Stimulation of rRNA synthesis by dMyc appears to be mediated through a rapid, coordinated increase in the levels of the Pol I transcriptional machinery.
As in humans, signaling through the insulin/PI3K pathway, and the activity of the target-of-rapamycin (TOR) gene product, are critical regulators of cell growth in Drosophila. To determine how Rheb, the most proximal known activator of TOR, controls cell growth, the Edgar lab used RNAi to inhibit each of the ~20 components of the Insulin/TOR growth regulatory network in Drosophila S2 cells. Assays on the affected cells ruled out the regulation of glucose and amino acid import as a major mode of growth control by Rheb and TOR, and indicated that the control of protein synthesis is key. In vivo experiments using mutants and overexpression confirmed that Rheb/TOR activity is a critical regulator of ribosome biogenesis and general protein synthesis. An in vivo for modifiers of rheb identified 20 loci required for Rheb-driven cell growth, and one of these modifiers was identified as a ribosomal protein, RpL38, further supporting the notion that Rheb regulates translation. The experiments Dr. Edgar discussed indicate that the affect of Rheb/TOR activity on ribosome biogenesis cannot be attributed solely to the well-characterized TOR targets, S6K and 4E-BP, and therefore implicate other targets as important mediators of TOR activity. Accordingly, Dr. Edgar discussed new studies to characterize TIF-1A, a critical regulator of Pol I activity (and rRNA transcription) that is a transcriptional target of dMyc and also appears to be regulated by Rheb/TOR signaling at the activity level.
During larval development, Drosophila grow ~200 fold. Most of this growth occurs in cells that have exited the mitotic cycle and differentiated during embryogenesis. As they grow, these cells employ endocycles, becoming highly polyploid. Endocycle progression can be manipulated by altering parameters that affect cell growth, such as feeding or the activity of growth-regulatory genes (dMyc, CycD, InR, Rheb, TOR). Endocycles are therefore growth-coupled. Dr. Edgar’s group devised a model of the endocycle and tested this both experimentally and using computer-based simulations. The model posits that endocycles use an oscillator comprised of positive and negative feedback between the transcription factor, E2F1/DP, and the cell cycle regulatory kinase, Cyclin E/Cdk2, which triggers S-phase initiation. Positive feedback acts via the canonical pathway whereby CycE/Cdk2 phosphorylates the E2F1 co-repressor RBF, allowing E2F1 to stimulate cycE transcription. Negative feedback occurs because CycE/Cdk2 downregulates E2F1 protein levels, which plummet during S-phases. In simulations these relationships produced the pulses of CycE/Cdk2 activity that are required for the periodic licensing and then firing of DNA replication origins. The oscillator’s frequency was determined by rates of E2F1 and CycE synthesis and degradation, and one or both of these parameters was controlled, in turn, by rates of growth. As observed experimentally, enhanced growth caused this hypothetical oscillator to accelerate, yielding greater ploidy. When growth rates fell below a threshold, oscillations stopped, arresting the cycle. Dr. Edgar noted that further experimentation is required to validate this model, but that it nevertheless provides a framework for considering how cell growth and DNA replication cycles might be coupled. Finally, Dr. Edgar described an experiment in which the growth of polyploid salivary glands was increased by selectively accelerating their DNA endoreplication cycle using over-expressed dE2F1/DP. This experimental result demonstrates that, in this organ, growth and DNA endoreplication are mutually interdependent processes and that, developmentally, either process could be regulated to control the other.
Jacques Pouysségur (Institute of Signaling, Developmental Biology and Cancer Research, CNRS UMR, Nice, France) continued the session with a talk about hypoxia signaling and tumor growth. A critical problem facing growing cancer cells is the availability of nutrient supply, in particular oxygen and glucose. Rapidly expanding tumors become hypoxic and acidic and unless there are able to get a nutrient source, they will stop growing and die by necrosis. A remarkable signaling system devised for rapid adaptation to and survival in hypoxia has been conserved throughout evolution. The hypoxia inducible transcription factor HIF, is central to this adaptation. HIF-1 is a transcriptional complex capable of inducing many genes involved in angiogenesis, anaerobic glycolysis, pH regulation, migration, and apoptosis. In the presence of oxygen, the HIF-a subunits are targeted for destruction by proline hydroxylation, a specific modification that provides recognition for the E3 ubiquitin ligase complex containing the von Hippel-Lindau tumour suppressor protein (pVHL). Three mammalian HIF prolyl-hydroxylases (PHD1, 2, 3) were recently identified and shown, at least in vitro, to down regulate HIF-a subunits. These enzymes, together with the HIF asparaginyl hydroxylase (FIH), which ‘represses’ the activity of HIF-1a, belong to a large family of non-haem iron oxygenases that require O2 and 2-oxoglutarate for their function. In his presentation, J. Pouyssegur (JP) showed that specific ‘silencing’ of HIF prolyl-hydroxylase 2 (PHD2) with short interfering RNAs (siRNA) is sufficient to stabilize and activate HIF-1a, in normoxia, in all the human cells investigated. A remarkable synergy in the HIF-1a activation process was observed in normoxia by ‘co-silencing’ both, PHD2 and FIH. This action measured by a HIF-dependent reporter gene recapitulates a full hypoxic response.
The HIF-1a subunit possesses two transactivating domains N- and C-TAD, the C-TAD being specifically inhibited via FIH. Because FIH has much higher affinity for O2 than PHD, Pouysségur hypothesized that the N- and C-TAD could induce their own set of specific genes, dependent on the tissue hypoxic gradient. He thinks this is an important issue in development allowing adequate balance of expression of survival- versus apoptotic-genes depending of the severity of hypoxia. Experiments validating this working hypothesis were presented.
Finally in the context of the mTOR pathway, specifically inhibited in hypoxic cells, Pouysségur showed that the hypoxia-induced “pro-apototic gene” BNIP3 could play a key role in the initiation of autophagy and cell death associated with acidic microenvironment.
Mario Pende (Faculté de Médecine Necker, Université Paris, Paris, France) completed the organismal growth session by presenting studies using the mouse as a model organism, to define how the control of cell size and number contribute to the final body size. Gene targeting experiments identified a number of genes affecting mouse body size. These include the insulin/insulin-like growth factor receptors, the insulin receptor substrate IRS, and the serine/threonine kinases mTOR, Akt1 and S6K1. mTOR coordinates the increase in cell number (cell proliferation) and cell size (cell growth) depending on nutrient availability, a conserved function throughout the evolution from yeast to mammals. By inactivating the mTOR substrates S6K1 and S6K2 and by using the mTOR inhibitor rapamycin, the Mario Pende group is addressing how growth and proliferation are controlled by this pathway. He showed that muscle cell size and cell number are regulated by separate branches of the mTOR pathway and that S6K1 is selectively required for size control. S6K1 deficient muscle fibers fail to adapt their size depending on nutrient levels. He then showed that the S6K1 deletion causes resistance to the hypertrophic action of IGF1 and hypersensitivity to the metabolic action of insulin. He proposed that S6K1 may serve as a nutrient effector promoting growth and inhibiting further nutrient uptake. This led to the idea of combining S6K1 deletion with the inactivation of Akt2, a major insulin effector in mouse metabolism. The Akt2/S6K1 deficient mice are glucose intolerant and show a predisposition to type 2 diabetes. Since tumours having deregulated activity of the Akt pathway are often extremely sensitive to rapamycin, the Mario Pende group is also addressing the role of S6K1 in tumorigenesis. In a mouse model of insulinoma triggered by oncogenic Akt, the S6K1 deletion is sufficient to block pancreatic beta tumorigenesis.
On the morning of May 14th, Nahum Sonenberg (McGill University, Montreal, Canada) began the last day of the meeting with a presentation on translational regulators. Translation initiation factors and their regulators are phosphorylated by multiple signalling pathways. Major phosphorylation targets are the translational repressors, 4E-BPs, which bind to the cap binding protein eIF4E. 4E-BPs are phosphorylated by mTOR (mammalian target of rapamycin) and possibly other kinases. To study the physiological function of the 4E-BPs he generated ‘knock out’ mice and flies. Mice which are deficient in 4E-BP2 exhibit changes in synaptic plasticity and memory and learning (Banko et al., J. Neurosci. 25, 9581, 2005). Mice which are deficient in both 4E-BP1 and 4E-BP2 become insulin resistant and glucose intolerant when fed a high fat diet. They are also more prone to high fat diet-induced obesity (in preparation). Flies contain only one 4E-BP species (d4E-BP), whose transcription is activated by dFOXO in response to oxidative stress. Flies which are deficient for d4E-BP are hypersensitive to oxidative and nutritional stress and exhibit a shortened life span, similar to dFOXO-null flies. Strikingly, oxidative stress resistance can be restored to dFOXO-null flies by ectopically expressing d4E-BP (Tettweiler et al., Genes and Development 19, 1840, 2005). These results demonstrate that the oxidative stress phenotype of dFOXO is mediated by d4E-BP. The different biological effects of 4E-BPs are consistent with their important function as effectors of mTOR.
Yves Barde (Biozentrum, University of Basel, Basel, Switzerland) gave a presentation on the control of neuronal growth by extrinsic factors. To a large extent, the adult central nervous system (CNS) of higher vertebrates is a post-mitotic organ. Yet it can undergo measurable volume changes, and not all of them result from the widespread death of cells and of their processes, as is the case in Alzheimer’s disease for example. In particular, in one of the most common human pathological conditions, namely depression, structures such as the hippocampus undergo regressive changes that are best explained by shrinkage of neurons and of their processes. The most common treatments of this condition target the monoamine system with drugs that block the re-uptake of dopamine and of serotonin. While these treatments are often effective, they typically take a few weeks to develop beneficial effects. In rodent models, recent work has revealed that these drugs increase the levels of one the most studied CNS growth factor, brain-derived neurotrophic factor (BDNF), both in the hippocampus and in the cortex. Furthermore, intraventricular injection of BDNF causes anti-depressant-like effects in rodents. Conversely, reducing the action of BDNF using animals that either lack one copy of the bdnf gene or express a truncated form of its tyrosine kinase receptor TrkB blocks the action of antidepressants. How BDNF mediates its growth effects on neurons has been difficult to study so far at the biochemical level. As BDNF is also increasingly recognised to play a key role in memory, perhaps by mechanisms requiring increased translation of specific mRNAs localised in dendrites, there is an urgent need to better understand the biochemistry of cell growth in neurons. However, cultured tumor cells such as neuroblastoma cells are not a satisfactory model of CNS neurons and the isolation of neurons from the embryonic brain remains problematic. Only relatively few cells can be obtained from well-defined brain areas such as the hippocampus, while the dissociated cortex typically leads to the generation of ill-defined cultures. Recently, it has become apparent that these difficulties can be circumvented by the use of mouse embryonic stem cells. In addition to their availability in unlimited quantities, these cells can readily be genetically manipulated. When rapidly diving stem cells are treated with retinoic acid, they can all be converted to a defined type of progenitor cells, namely Pax6-positive radial glial cells. These cells then go on to differentiate into pyramidal neurons, much like they do in vivo in the developing cerebral cortex and in the hippocampus. They not only express high affinity receptors for BDNF, but they also respond biochemically to the addition of this growth factor. In particular, pathways such as the MAP kinase and the TOR-dependent S6 kinase pathway can readily be activated. The use of this system should open the way for an unbiased analysis of the biochemical events triggered by the addition of growth factors and triggering growth, including the formation of new dendritic branches and of spines. Conversely, regressive events such as the elimination of neuronal processes can be studied with engineered ES cells by regulating the levels of expression of the neurotrophin receptor p75, a cell surface protein that is typically up-regulated in neurons in a number of pathological conditions.
Pierre Leopold (Université de Nice, Nice, France) gave a presentation on the humoral control of growth in Drosophila. The control of growth at the level of a whole organism involves a series of intricate humoral regulations allowing the coordination of growth programs in all tissues. One key feature of this global control is to integrate extrinsic influences, like nutrition, as well as intrinsic mechanisms which are linked to the program of development. Pierre Léopold and co-workers presented a series of experiments addressing these issues in the genetic model of the fruit fly Drosophila. These experiments suggest that nutrition is coupled with growth control through the function of a general sensor operating in the fat body of the animal. This specific organ combines the functions of both the liver and the fat tissue of vertebrates. Leopold presented evidence that upon specific amino acid starvation, the sensor mechanism in this tissue operates through a downregulation of the TOR branch of the Insulin/IGF signalling pathway (IIS), and induces a remote inhibition of organismal growth via local repression of PI3-kinase signaling in peripheral tissues. The nature of this remote inhibition of IIS is not well understood, but several mechanisms could be involved, like a control of insulin/IGF secretion in the brain cells that produce it or a modification of its biological function through an association with binding proteins similar to the IGF-BPs and ALS. These findings overall suggest that TOR could be a central controller of organismal growth through its participation as a general nutrition sensor, and that insulin/IGF signaling would respond to the function of this general sensor in peripheral tissues.
Matthias Peter (University of Zurich, Zurich, Switzerland) gave the closing presentation of the meeting by discussing the regulation of the cell cycle and protein translation by growth signals. To identify factors required for exit from the cell cycle upon starvation, a collection of yeast mutants was screened for starvation sensitivity. 216 genes were identified including the PI 3-kinase VPS34 required for transport of proteins to the vacuole. To identify a factor that may need to be transported to the vacuole to obtain cell cycle exit upon nutrient deprivation, a collection of GFP fusion strains was screened. This screen identified the Ras exchange factor CDC25 as a protein that is targeted to the vacuole and degraded upon nutrient deprivation. The original starvation sensitivity screen also identified the eIF2alpha kinase GCN2, suggesting that the regulation of protein synthesis is a key aspect of the starvation response. The role of URI1 in the regulation of translation was also discussed.
All presentations at the meeting were followed by extensive discussion and at the completion of the meeting one hour was dedicated to discussing outstanding questions in the field.
The meeting was closed on May 15th and participants departed the same day.