BIOLOGICAL OVERVIEW

Tuberous sclerosis complex (TSC) is an autosomal dominant disorder affecting 1 in 5800 individuals. TSC occurs in multiple organs, including the brain, eyes, skin, kidney, heart, lungs, and skeleton, and is characterized by the presence of benign tumor cells termed hamartomas. Hamartomas are a mass of disorganized but differentiated cells indigenous to the site. TSC hamartomas rarely progress to malignancy, but brain hamartomas frequently cause epilepsy, mental retardation, autism, or attention deficit-hyperactive disorder. One of the notable features of TSC hamartomas is the presence of giant cells in the tumors. Linkage studies in families with TSC have established two TSC loci, TSC1 (see Drosophila Tsc1) and TSC2, each accounting for approximately 50% of cases. The TSC1 gene encodes a novel protein, hamartin, that contains a single transmembrane domain and a large cytoplasmic tail with coiled-coil domains (van Slegtenhorst, 1997). The TSC2 gene, the homolog of Drosophila Gigas and the subject of this overview, encodes a novel protein, tuberin, that contains a region of homology to the GTPase-activating protein (GAP) for the small-molecular-weight GTPase Rap1. Clones of gigas mutant cells induced in imaginal discs differentiate normally to produce adult structures. However, the cells in these clones are enlarged (gigas means 'giant' in Latin) and repeat S phase without entering M phase. This result suggests that the TSC disorder may result from an underlying defect in cell cycle control (Ito, 1999 and references). Recent studies have focused on the role of the TSC genes as modifiers of the insulin pathway (Tapon, 2001, Potter, 2001 and Gao, 2001). In particular, it is unlikely that deficiencies in TSC gene function results in alterations in cell polyploidy (Ito, 1999), but instead the growth and size defects may arise through defective signaling in the insulin pathway (Tapon, 2001, Potter, 2001 and Gao, 2001).

Precise body and organ sizes in the adult animal are ensured by a range of signaling pathways. Rheb, a novel, highly conserved member of the Ras superfamily of G-proteins, promotes cell growth. Rheb is required in the whole organism for viability (growth) and for the growth of individual cells. These data demonstrate that Rheb is required for both cell growth (increase in mass) and cell cycle progression; one explanation for this dual role would be that Rheb promotes cell cycle progression by affecting cell growth. Consistent with this interpretation, flies with reduced Rheb activity are hypersensitive to rapamycin, an inhibitor of the growth regulator target of rapamycin (TOR), a kinase required for growth factor-dependent phosphorylation of ribosomal S6 kinase (S6K). In cultured cells, the effect of overexpressing Rheb was blocked by the addition of rapamycin. These results imply that Rheb is involved in TOR signaling (Patel, 2003). Additional studies show that Rheb functions downstream of the tumor suppressors Tsc1 (tuberous sclerosis 1)-Tsc2, with Tsc2/Gigas functioning as a GAP for Rheb (Saucedo, 2003; Zhang, 2003), and that a major effector of Rheb function in controlling growth is, in fact, ribosomal S6 kinase (Stocker, 2003). It is still not clear, however, how Rheb signals to TOR (Zhang, 2003).

Mutations in the TSC1 and TSC2 genes have been described in patients with tuberous sclerosis (Wilson, 1996 and van Slegtenhorst, 1997). Moreover, loss of heterozygosity at the TSC1 and TSC2 loci has been demonstrated in TSC patient lesions as well as in sporadic tumors of non-TSC patients (Green, 1994; Henske, 1995, and van Slegtenhorst, 1997). These results support a tumor suppressor function for both TSC1 and TSC2. The mechanism by which the loss of hamartin or tuberin produces tumors is unknown. The clinical features of TSC1 and TSC2 disease are indistinguishable (Povey, 1994), suggesting that the two TSC proteins participate in the same biochemical process. Since TSC mutations seem to affect cell growth, TSC proteins might be involved in the regulation of the cell cycle (Ito, 1999 and references).

The FRT/FLP recombination system (Xu, 1993) was used to screen the left arm of the third chromosome for mutations affecting eye development. About 150,000 X-ray-mutagenized progeny were examined for abnormal morphology in eye clones and 20 complementation groups of lethal mutations were isolated. One of these groups (C1), comprised of eight alleles, contains mutations in gigas, a previously described mutant that exhibits a very similar large cell phenotype (Ferrús, 1976 and Canal, 1994). gigas was originally isolated as a mutant with larger bristles in clones (Ferrús, 1976), and gigas mutant photoreceptors in eye clones are two to three times larger and establish more synapses than normal neurons (Canal, 1994). Homozygous gigas animals are larval lethal and die by early third instar. C1 mutations produce enlarged cells in mutant clones in the eye and wing. All unit eyes (ommatidia) in mutant clones are two to three times larger in area than normal. Eye sections reveal that all the cells, including photoreceptor cells and nonneuronal accessory cells, are enlarged in the clone; however, the structure and organization of ommatidia are nearly normal. Photoreceptors are occasionally missing, especially at clone borders where there are both normal and enlarged cells in the same ommatidia. Although C1 mutant clones consist of larger cells, the developmental program of these cells seems to proceed normally. When mutant clones are generated in the wing, sensory bristles in the clones are larger but appear otherwise normal. In C1 clones on the wing blade, hair density is decreased. Since all wing blade cells have a single hair at the same position as each another, this result indicates that individual epidermal cells are larger. These results demonstrate that multiple cell types are affected by C1 (Ito, 1999).

Mutations have been characterized in both Tsc1 and Tsc2/gigas genes of Drosophila. Inactivating mutations in either gene cause an identical phenotype characterized by enhanced growth and increased cell size with no change in ploidy. Overall, mutant cells spend less time in G1. Coexpression of both Tsc1 and Tsc2 restricts tissue growth and reduces cell size and cell proliferation. This phenotype is modulated by manipulations in cyclin levels. In postmitotic mutant cells, levels of Cyclin E and Cyclin A are elevated. This correlates with a tendency for these cells to reenter the cell cycle inappropriately as is observed in the human lesions (Tapon, 2001).

The disease phenotype in humans and observations in Drosophila suggest that all the functions of TSC1 and TSC2 are mediated by a complex of the two proteins, and that neither protein has additional functions. Moreover, combined overexpression of both Tsc1 and Tsc2 readily elicits a phenotype in a variety of tissues, while overexpression of either protein alone is insufficient. This argues that most, if not all, of the endogenous Tsc1 and Tsc2 is sequestered in the complex and is unavailable to complex with exogenously supplied protein. Cells mutant for either Tsc1 or Tsc2 are larger than wild-type cells and show a marginal decrease in their division time. Thus, the rate of mass accumulation (growth) in Tsc1 and Tsc2 mutant cells must be greater than that of wild-type cells to allow them to be larger than their neighbors despite a similar division time. Hence, the primary alteration in the mutants may be an increase in the rate of cellular growth. Conversely, when both Tsc1 and Tsc2 are overexpressed, the growth rate of the tissue is reduced. The cells are smaller and take longer to divide than wild-type cells. The increase in duration of the cell cycle may represent an attempt to maintain a near normal cell size in a situation where the rate of growth is reduced. An alternate possibility is that Tsc1 and Tsc2 negatively regulate both cell growth and cell cycle progression independently (Tapon, 2001).

The increased growth rate of mutant tissue is apparent in both cycling and postmitotic cells. In cycling populations such as the wing disc and in the eye disc anterior to the morphogenetic furrow, mutant clones are larger than the corresponding wild-type twin spots. The difference in size between the clone and the twin-spot becomes more exaggerated after the cells have stopped dividing, resulting in a greatly increased representation of mutant tissue in the adult (Tapon, 2001).

A major pathway that regulates growth in Drosophila is the pathway that links signaling via the insulin receptor to phosphorylation of the ribosomal protein S6. This is thought to lead to increased ribosome biosynthesis and mass accumulation. Thus, loss-of-function mutations in the insulin receptor (inr), the insulin receptor substrate (chico), and in genes encoding the downstream signaling molecules PI3K, Akt (Dakt), and S6 kinase (dS6K), all reduce cell and organ size. Conversely, overexpression of PI3K or loss-of-function mutations in the insulin pathway antagonist PTEN (dPTEN) lead to increased cell and tissue growth. In addition, Ras1, possibly acting via dmyc, can also promote cell growth in response to extracellular growth factors. Increased activity of Cyclin D in a complex with cdk4 has also been shown to be a potent stimulus for growth in both dividing and postmitotic cells (Tapon, 2001 and references therein).

The enhanced growth observed in the Tsc1 or Tsc2 mutants most resembles the results of inactivating PTEN or increasing Ras1 or dmyc activity. In each of these situations, there is a reduction in the length of the G1 phase. In contrast, increased growth driven by Cyclin D/cdk4 does not alter the distribution of cells in different phases of the cell cycle. The effects of the combined overexpression of Tsc1 and Tsc2 displays genetic interactions with multiple pathways. The phenotype is influenced by alterations in the levels of dS6K, PTEN, Ras1, dmyc, cyclin D, and cdk4. Thus, Tsc1 and Tsc2 may function downstream of the point of convergence of these pathways. Alternatively, Tsc1 and Tsc2 may primarily antagonize one of these pathways, but this effect could be overcome by increasing the activity of one of the others (Tapon, 2001).

Imaginal discs containing large mutant clones of either Tsc1 or Tsc2 are significantly larger than wild-type imaginal discs but are patterned relatively normally. Mutations in several cell size regulators such as E2F or Rbf have been shown to affect the size of individual cells but do not alter the final size of the organ. The mechanisms that define organ size are poorly understood. The complex of Tsc1 and Tsc2 restricts the growth of organs in vivo. Screens for genes that interact with Tsc1 and Tsc2 are likely to identify additional regulators of organ size (Tapon, 2001).

Tsc1 and Tsc2 may modulate the cell cycle via changes in cyclin levels. In Tsc1 and Tsc2 mutant clones, the levels of both Cyclin E and Cyclin A are elevated. Cell growth driven by dmyc or Target of rapamycin (dTor) elevates Cyclin E levels. It has been postulated that Cyclin E may function as a 'growth sensor' in a manner analogous to CLN3 in yeast and that the translation of Cyclin E is more efficient in cells that have an increased rate of growth. The increased levels of Cyclin E may be responsible for the shortening of G1 in Tsc1 and Tsc2 mutants. It is unclear why Cyclin A and Cyclin B are also elevated in mutant cells. Cyclin A is normally expressed at high levels in G2. In Tsc1 or Tsc2 mutants, Cyclin A levels are elevated in the post-mitotic cells of the eye disc that are clearly not arrested in G2. Thus it seems likely that the increased growth in mutant cells may also lead to increased levels of mitotic cyclins. Alternatively Tsc1 and Tsc2 may function in a pathway that negatively regulates cyclin levels (Tapon, 2001).

While the increased levels of cyclins are likely to be a response to the increased growth rate of mutant cells, the possibility that they are in some way responsible for the increased growth rate cannot be excluded. In Drosophila, the Cyclin D/cdk4 complex serves to promote growth. In such a scenario, the loss of Tsc1 or Tsc2 gene function may lead to elevated levels of cyclins leading to increased growth and proliferation. Surprisingly, increased expression of Cyclin E, which is thought to primarily promote S-phase entry and not growth, is also able to suppress the phenotype induced by overexpression of Tsc1 and Tsc2. This might reflect the existence of feedback loops where Cyclin E might downregulate the levels or activity of the Tsc1/Tsc2 complex. Alternatively, in some circumstances, Cyclin E might assume some of the functions of the growth promoting Cyclin D. Indeed, in mammalian cells, cyclin E has been shown to fully compensate for the loss of cyclin D1 (Tapon, 2001).

Tsc mutant cells fail to maintain a developmentally induced G1 arrest posterior to the second mitotic wave in third instar eye imaginal discs. The establishment of this G1 arrest requires a downregulation of Cyclin E and Cyclin A expression. However, the transient cell cycle arrest in the morphogenetic furrow occurs normally in Tsc1 and Tsc2, suggesting that it is the maintenance of G1 arrest that is perturbed rather than its initial establishment. Postmitotic cells continue to grow abnormally in Tsc1 and Tsc2 mutants and express elevated levels of Cyclin E and Cyclin A. A likely model is that inappropriate and continued growth in postmitotic cells leads to an accumulation of Cyclin E and the mitotic cyclins. This would eventually force cells to overcome a developmentally regulated cell cycle arrest and to reenter the cell cycle. Indeed, many of the lesions in patients with TSC occur in organs that consist predominantly of postmitotic cells such as the heart and brain. A successful therapeutic strategy in tuberous sclerosis is likely to be one that can curtail the inappropriate cell growth (Tapon, 2001).

Regulation of TORC1 in response to amino acid starvation via lysosomal recruitment of TSC2

Small GTPases act as molecular switches that alternate between GTP-bound and GDP-bound states, thereby regulating a vast array of cellular parameters, including mitochondrial activity, cell growth, cell metabolism, and cell morphology. Small GTPases are activated or inactivated by guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs), respectively, which regulate the GDP/GTP load of the GTPase. Hence, understanding how GEFs and GAPs are regulated is an important aspect of understanding GTPase function. Compared to the regulation of GEFs, relatively little is known about how the activity of GAPs is regulated. GAPs acting on members of the Rho and Arf GTPase superfamilies are activated via membrane recruitment, causing rearrangements in the structure of the GAPs upon membrane binding. Regulation of GAPs acting on members of the Ras superfamily of GTPases is less well understood. One such GAP is composed of the TSC1/TSC2/TBC1D7 trimeric tumor suppressor complex, which acts on the small GTPase Rheb. Although it is known that activity of this complex is regulated by phosphorylation on multiple sites, it is not yet clear how these phosphorylations affect TSC1/2 activity at the molecular level (Demetriades, 2014).

The kinase TOR complex 1 (TORC1) is a potent anabolic regulator of cellular growth and metabolism that is often hyperactivated in human cancers. To be active, TORC1 needs to bind a molecule of Rheb in the active, GTP-bound state. By inactivating Rheb, the TSC1/2 complex is therefore a critical upstream inhibitor of TORC1. The TSC1/2 complex acts as a central point of integration of almost all known inputs regulating TORC1, including cellular stresses such as low oxygen or low ATP, and various growth-promoting signals, such as PI3K, Ras, TNF, and Wnt signaling. The importance of the tuberous sclerosis complex (TSC) on TORC1 signaling and growth is highlighted by the fact that TSC2-inactivating mutations have been found in various human growth-related diseases. One other important input regulating TORC1 activity is the availability of amino acids. Whether TSC2 is also involved in regulating TORC1 in response to amino acids, however, is unclear because various studies have come to differing conclusions (Demetriades, 2014).

Unlike all the other inputs that regulate TORC1 via TSC1/2 and Rheb, amino acids regulate TORC1 via a separate set of small GTPases, the Rag GTPases. The Rag GTPases form heterodimeric complexes consisting of RagA or RagB bound to RagC or RagD. These complexes are stably anchored to lysosomal membranes via the LAMTOR/Ragulator complex. In the presence of amino acids, the Rag dimers are in an 'active' conformation with RagA or RagB bound to GTP and RagC or RagD bound to GDP. The active Rag dimers recruit TORC1 to the lysosomal surface, where it binds Rheb to form an active holoenzyme. In the absence of amino acids, the Rag GAP complex termed GATOR1 causes the Rag dimers to switch into an inactive conformation containing GDP-bound RagA/B, thereby releasing TORC1 from the lysosomal surface. This causes TORC1 to become inactive, presumably because it no longer binds active Rheb on the lysosomal surface. Hence, TORC1 activation can currently be viewed as consisting of two aspects-the activation of Rheb in response to a plethora of regulatory inputs and the localization of TORC1 to lysosomal membranes in response to amino acids, which allows it to meet Rheb (Demetriades, 2014).

This study uncover subcellular localization as a mechanism regulating activity of the TSC1/2 GAP complex. Upon amino acid removal, TSC1/2 is recruited to lysosomes via binding to the Rag proteins, thereby bringing TSC1/2 in close proximity to its target, Rheb. This suggests that relocalization of GAPs to the vicinity of their substrates is one mechanism for their regulation. Unexpectedly, it was found that regulation of Rheb by TSC1/2 upon amino acid starvation is required for TORC1 to be released from lysosomal membranes. This suggests a 'dual anchoring' mechanism of TORC1 at the lysosome, perhaps with the Rag proteins playing a crucial role in recruiting TORC1 to the lysosomal membrane and Rheb helping to retain it there. Xells lacking TSC2 were found to be impaired in their response to amino acid starvation, failing to efficiently turn off TORC1. As a result, cells lacking TSC2 are very sensitive to amino acid starvation and die under conditions that control cells can cope with. In sum, these data indicate that the TSC1/2 complex is responsive to amino acid starvation and participates in amino acid signaling to TORC1. Hence, the TSC1/2 complex appears to play a role in regulating TORC1 in response to all regulatory inputs known to date (Demetriades, 2014).

The data presented in this study suggest a model whereby, in the presence of amino acids, mTORC1 accumulates on lysosomes due to a dual anchoring activity composed primarily of the Rag proteins but supported by Rheb. While amino acids are present, binding between the Rag proteins and TSC1/2 is low, causing the TSC1/2 complex to remain cytoplasmic. Upon amino acid removal, the Rag proteins cause mTORC1 to be released from lysosomes via two independent activities, both of which result from the Rag proteins shifting to an inactive conformation. First, the Rag proteins reduce their binding for mTORC1, thereby releasing one of the two activities tethering mTORC1 at the lysosome. Second, the Rag proteins actively recruit TSC2 to the lysosome. This allows TSC2 to act on Rheb, thereby releasing the second tethering activity keeping mTORC1 on the lysosome. In the absence of TSC2, this second activity is unaffected, causing mTORC1 to remain lysosomally localized (Demetriades, 2014).

The LAMTOR complex (composed of the p18, p14, and MP1 proteins) has been shown to serve as a docking point for mTORC1 and MEK/ERK complexes, regulating their recruitment to late endosomes/lysosomes and their activation status. These data demonstrate that integrity of the LAMTOR complex is critical for proper TSC2 subcellular localization upon amino acid withdrawal, therefore highlighting the importance of this scaffold complex for endomembrane-mediated activation/inactivation of signaling pathways (Demetriades, 2014).

The data presented in this study show that the TSC1/2 complex is part of the molecular machinery required for mTORC1 to respond properly to the absence of amino acids. The TSC1/2 complex responds to amino acid starvation by changing its subcellular localization and TSC2 is required for mTORC1 to be fully released from lysosomes and fully inactivated upon amino acid removal. That said, however, in cells lacking TSC2, there is nonetheless a clear initial drop in TORC1 activity upon amino acid removal. The remaining activity is then sustained indefinitely. Hence, mTORC1 appears to consist of two pools or two degrees of activation, one of which requires TSC2 to become inactive upon amino acid withdrawal and one of which responds independently of TSC2. This might explain why previous studies arrived at differing interpretations of their data because there is some response of TORC1 to amino acid removal in TSC2 null cells; however, the response is severely blunted compared to controls. Further work will hopefully shed light on these two pools of activity. Although the impairment in mTORC1 response to amino acids in TSC2 null cells is partial, it is nonetheless of critical physiological relevance because TSC2 null MEFs die upon amino acid removal in sharp contrast to control MEFs (Demetriades, 2014).

Various insights can be derived from these data. (1) Regulation of mTORC1 activation could previously be rationalized as consisting of two independent, parallel steps: first, regulation of Rheb via TSC1/2 in response to a plethora of signals including stresses and growth factor signaling, and second, regulation of mTORC1 subcellular localization to lysosomal membranes in response to amino acids. Only when mTORC1 is properly localized to meet active Rheb would an active holoenzyme form. The data presented in this study blur the distinction between these two steps because Rheb also affects mTORC1 localization, and amino acids also signal through TSC1/2. Instead, the two sets of regulatory inputs into mTORC1 appear to be more integrated. (2) Amino acid removal is 'dominant' over growth factor signaling, causing mTORC1 to shut off despite the presence of growth factors. This was previously explained by the fact that, in the absence of amino acids, mTORC1 could not localize near active Rheb to form an active complex. The dual tethering model is also consistent with this notion but for a slightly modified reason, which is that amino acid starvation acts to sever both the Rag and Rheb lysosomal tethering activities. (3) Seen from the perspective of the Rag proteins, they swap binding partners depending on the state of amino acid signaling, binding preferentially to mTORC1 in the presence of amino acids, and binding preferentially to the TSC1/2 complex in the absence of amino acids. Consequently, mTORC1 and TSC1/2 also swap subcellular localizations (Demetriades, 2014).

It was noticed that, knockdown of Rag proteins did not result in as strong a reduction in TORC1 activity in the presence of amino acids as has been previously reported. The simplest explanation is technical-that the Rag knockdowns are not strong enough to fully abrogate Rag recruitment of mTORC1 in the presence of amino acids but are sufficient to impair Rag recruitment of TSC2 in the absence of amino acids. In that case, optimizing the Rag knockdowns might lead to even stronger effects than the ones presented in this study. Two alternate biological explanations, however, might be worth investigating in the future. The first is that the data suggest a dual anchoring mechanism of mTORC1 at the lysosomal membrane-one by the Rag proteins and one by Rheb. It is possible that the relative contribution of lysosomal tethering of mTORC1 by the Rag proteins and by Rheb might depend on their relative levels of expression and activation in the system being studied. This balance will likely depend on the cell line and on cell culture conditions. A second possible explanation could be one of biological kinetics, influenced by treatment strategy. The outcome might be quantitatively different if one looks at acute amino acid removal from cells adapted to complete medium (which is done in this study) or if one looks at amino acid add-back to cells that have equilibrated their signaling to the absence of amino acids. Indeed, it was seen that, the Rag proteins were knocked down in HEK293FT cells, no dramatic reduction is seen in mTORC1 activity in the presence of amino acids. However, if amino acids are removed for 1 hr and then amino acids were re-added for 30 min, the same degree of Rag knockdown causes an obvious reduction in mTORC1 activity. Likewise, in Drosophila S2 cells, RagC knockdown only had a mild effect on TORC1 activity in untreated cells but severely blunted the ability of cells to respond to amino acid add-back. This difference between amino acid removal and amino acid add-back raises the interesting possibility that the Rag proteins are key in recruiting mTORC1 to the lysosome, a process that happens upon amino acid readdition, and that both the Rag proteins and Rheb work together to keep mTORC1 on the lysosome once it is there. Indeed, in agreement with this model, mTOR is able to be recruited to the lysosome upon amino acid readdition in TSC2 null MEFs in which Rheb is knocked down, indicating that, although Rheb tethers mTOR to the lysosome upon amino acid removal, it is not required for de novo recruitment of mTOR to the lysosome upon amino acid add-back (Demetriades, 2014).

Previous reports have shown that hyperactive mTORC1 signaling or dysregulated translation can lead to a metabolic mismatch in supply and demand, leading to cellular or organismal death. It is hypothesized that, if TSC2 is required for mTORC1 activity to respond to amino acid starvation, then TSC2 might also be necessary for cells to respond physiologically to this stress. Indeed, TSC2 knockout MEFs die upon removal of amino acids, whereas control cells do not. The fact that cells with elevated mTORC1 activity due to impaired nutrient sensing die when deprived of amino acids raises the interesting hypothesis that limiting nutrient supply to tumors of certain genotypes might have a beneficial effect on their treatment. Consistent with this effect being due to elevated mTORC1 activity, the death of TSC2 knockout MEFs is rescued by rapamycin treatment. This leads to the unexpected finding that rapamycin can actually promote cell survival under nutrient deprivation conditions, which might have therapeutic implications in mTOR-related malignancies (Demetriades, 2014).

This study identified the subcellular localization of TSC1/2 as one mechanism regulating this GAP holoenzyme. In an accompanying manuscript, It has been shown that TSC2 subcellular localization is also regulated by insulin signaling. They show that, in the absence of fetal bovine serum (FBS), TSC2 is lysosomally localized. This study shows that, in the absence of amino acids, TSC2 is lysosomally localized, even in the presence of growth factor signaling (FBS). Hence, the presence of both amino acids and growth factor signaling are required to keep TSC2 in the cytoplasm, and as long as one of the two is missing, TSC2 becomes lysosomally localized. Combined, these findings raise the possibility that TSC2 subcellular localization is a general mechanism for regulating this complex. It would be interesting to study whether the other inputs known to regulate TSC2 also affect its subcellular localization (Demetriades, 2014).

CycD/Cdk4 and discontinuities in Dpp signaling activate TORC1 in the Drosophila wing disc

The molecular mechanisms regulating animal tissue size during development are unclear. This question has been extensively studied in the Drosophila wing disc. Although cell growth is regulated by the kinase TORC1, no readout exists to visualize TORC1 activity in situ in Drosophila. Both the cell cycle and the morphogen Dpp are linked to tissue growth, but whether they regulate TORC1 activity is not known. This study developed an anti-phospho-dRpS6 antibody that detects TORC1 activity in situ. Unexpectedly, it was found that TORC1 activity in the wing disc is patchy. This is caused by elevated TORC1 activity at the cell cycle G1/S transition due to CycD/Cdk4 phosphorylating TSC1/2.TORC1 is also activated independently of CycD/Cdk4 when cells with different levels of Dpp signaling or Brinker protein are juxtaposed. This study has thereby characterize the spatial distribution of TORC1 activity in a developing organ (Romero-Pozuelo, 2017).

During animal development, tissues increase tremendously in mass, yet stop growing at very stereotyped sizes in a robust manner. For instance, the Drosophila wing is specified as a cluster of circa 50 cells, which increases in mass ~500-fold before terminating growth. Once growth has ceased, the left and right wings of an individual fly are virtually identical in size, to within 1%, illustrating the robustness of this process. How animal tissue size is regulated is a fundamental open question in developmental biology (Romero-Pozuelo, 2017).

As mitotically growing tissues develop, two independent cellular processes occur in a coordinated manner: proliferation and cell growth. By itself, proliferation -- the division of cells -- does not lead to mass accumulation. This was nicely shown in the Drosophila wing where overexpression of E2F speeds up the cell cycle, but leads to a normally sized tissue containing more, smaller cells. For a tissue to grow, cells need to accumulate biomass. The mechanisms interconnecting cell proliferation and cell growth are not completely understood. In organisms from yeast to humans, growth is in large part regulated by the target of rapamycin complex 1 (TORC1) kinase. TORC1 promotes biomass accumulation by promoting anabolic metabolic pathways such as protein, lipid, and nucleotide biosynthesis, while repressing catabolic processes such as autophagy. Hence, to understand tissue growth it would be of interest to study the spatial distribution of TORC1 activity in a developing tissue. This line of investigation has been hampered, however, by the lack of readouts for TORC1 activity that can be used in situ (Romero-Pozuelo, 2017).

One signaling pathway that strongly affects tissue size is the Dpp pathway. Dpp is expressed and secreted by a stripe of cells in the medial region of the wing imaginal disc, and forms an extracellular morphogen gradient that both helps to pattern the wing and affects its size. In the absence of Dpp signaling during development, only small rudimentary wings are formed. In contrast, overexpression of Dpp leads to strong tissue overgrowth, in particular along the axis of the morphogen gradient. Several models have been proposed for how Dpp signaling regulates wing size. The exact mechanism by which Dpp regulates tissue size, however, is an unresolved issue. Dpp signaling acts to repress expression of a transcription factor called Brinker. Brinker appears to mediate most of the size effects of Dpp signaling. When Brinker is genetically removed, Dpp signaling becomes dispensable for wing growth. Given that Dpp signaling promotes tissue growth, an open question is whether Dpp signaling promotes TORC1 activity (Romero-Pozuelo, 2017).

Thia study examined whether Dpp signaling promotes TORC1 activity in the Drosophila wing disc. To this end, a phospho-RpS6 (pS6) antibody was developed that allows TORC1 activity to be assayed in situ in tissue. This reagent reveals unexpectedly that TORC1 activity in the growing wing disc is neither uniform nor graded, but is instead patchy. This patchiness is mediated via CycD/Cdk4 and the tuberous sclerosis 1 (TSC1)-TSC2 complex in response to cell cycle stage. Using this pS6 antibody, this study found that TORC1 activity is also induced by discontinuities in Dpp signaling or discontinuities in Brinker levels. It is proposed that these discontinuous conditions may be analogous to regenerative conditions that happen in the wing disc in response to tissue damage. In sum, this work reveals the pattern of TORC1 activity in the context of a developing organ (Romero-Pozuelo, 2017).

TORC1 activity in the wing disc is modulated by the cell cycle, with cells in early S phase showing the highest TORC1 activity. Interestingly, an accompanying paper finds similar results in the Drosophila eye disc (Kim, 2017). This might reflect a metabolic requirement by early S-phase cells for large amounts of nucleotide biosynthesis, an anabolic process promoted by TORC1. Indeed, in various contexts S6K and TORC1 activity were found to be required for the transition from G1 to S. Connections between mechanistic TOR (mTOR) and the cell cycle have previously been found in cultured cells. In human fibroblasts, mTOR shuttles in and out of the nucleus in a cell cycle-dependent manner, peaking in the nucleus shortly before S phase. The relevance of this subcellular relocalization to what is observe in this study, however, is unclear. In fibroblasts, S6K1 activity was found to be highest during early G1, whereas in HeLa cells it was found to be highest during M phase. In sum, it is unclear to what extent cells in culture recapitulate endogenous development, or whether the influence of the cell cycle on TORC1 activity is very context dependent. The TSC1/2 complex has been reported to be phosphorylated by cell cycle-dependent kinases.TSC1 is phosphorylated on Thr417 by Cdk1 during the G2/M transition. This inhibitory phosphorylation would lead to elevated TORC1 activity during G2/M, which does not fit with what was observe here, and thus might be relevant in a different developmental context. Instead, this study found that TSC2 can be phosphorylated by the CycD/Cdk4 complex on Ser1046, and possibly other sites as well, and that this leads to activation of TORC1. This fits with several observations in the literature. Firstly, in U2OS cells the TSC complex was also found to bind cyclin D, leading to its phosphorylation at unknown sites. In U2OS cells, this causes destabilization of the Tsc1 and Tsc2 proteins, which was not observed in this study. Secondly, Tsc1/2 and CycD/Cdk4 were previously found to interact genetically in Drosophila: The reduced tissue growth caused by Tsc1 + Tsc2 overexpression was found to be fully suppressed by expression of CycD + Cdk4. This fits well with the current data suggesting that CycD/Cdk4 directly inhibits the TSC complex via phosphorylation. Thirdly, Cyclin D and Cdk4 were previously reported in Drosophila to promote cell and tissue growth, fitting with activation of the TORC1 complex by CycD/Cdk4. It is worth noting that some patchy TORC1 activity is still seen in CycD- or Cdk4-null discs and in discs with the single phospho-site mutations in TSC2. Hence it is possible that Cdk4 may not be the only factor regulating TORC1 activity in response to the cell cycle, and that Cdk4 might phosphorylate TSC2 on additional sites (Romero-Pozuelo, 2017).

What are the roles of CycD/Cdk4 in cell cycle progression and cell growth? Whereas mammals have three cyclin D genes, CycD1-3, and two CycD binding kinases, Cdk4 and Cdk6, Drosophila has a single CycD, a single Cdk4, and no Cdk6. Hence Drosophila provides an opportunity to elucidate the function of the CycD/Cdk4 complex without difficulties arising from redundancy. Indeed, results in Drosophila clearly show that CycD/Cdk4 promotes cell growth and not cell cycle progression. Both CycD- and Cdk4-null animals are viable, and fluorescence-activated cell sorting (FACS) analysis of null cells revealed that they have a normal cell cycle profile, indicating that they are dispensable for normal cell cycle progression. Instead, Cdk4- and CycD-null animals are 10%-20% smaller than controls, indicating that they promote cell growth. The finding that CycD/Cdk4 activates TORC1 during the G1/S transition can provide one mechanism by which the CycD/Cdk4 complex promotes growth. Hence, from these data it is proposed that in Drosophila the CycD/Cdk4 complex is not part of the core machinery required for cell cycling, but is rather an effector 'side branch' activated at G1/S to promote cell growth. Data from the mouse suggest something similar. CycD1, CycD2, and CycD3 knockout mice are all viable. One could imagine this to be due to redundancy between these three genes, but actually CycD1, CycD2, CycD3 triple-knockout mice survive to mid-gestation, and the triple-knockout mouse embryonic fibroblasts proliferate relatively normally. The mid-gestation lethality of the triple knockouts appears to be due to specific effects in hematopoietic and myocardial cells. Hence, cyclins D1-D3 are also dispensable for cell cycle progression in mice. Interestingly, CycD1 knockout mice and CycD1, CycD2 double-knockout mice are viable but have reduced body size, reminiscent of the size phenotype observed in CycD knockout flies. In sum, despite CycD/Cdk4 being claimed in most reviews on the cell cycle as playing an important role in G1/S progression, it appears that this complex may function rather to promote cell growth in a cell cycle-dependent manner (Romero-Pozuelo, 2017).

Does Dpp control growth in the wing? When discontinuities in Dpp activity or in Brinker levels were genetically induce, activation was observed of TORC1 at the site of discontinuity. Hence, Dpp signaling per se does not appear to activate TORC1; rather, the comparison between high Dpp signaling and low Dpp signaling cells does. In an unperturbed disc, no pattern of pS6 staining was observed that correlates with the Dpp activity gradient, which is highest medially and drops toward the anterior and posterior extremities. This might be due to the fact that in an unperturbed disc the Dpp and Brinker gradients are smooth and do not have such discontinuities. A similar effect of Dpp was previously observed on cell prolife ration, except that in this case the effect of the Dpp discontinuity was very transient, lasting only a few hours after clone induction, whereas the effect seen on growth is sustained. Dpp signaling is, nonetheless, required for growth, because in the absence of Dpp, small vestigial wings are formed. Hence one interpretation might be that low levels of Dpp signaling are continuously required for growth, but that Dpp signaling becomes instructive for tissue growth only when discontinuities in the gradient arise, perhaps as a result of tissue damage or cell delamination, to initiate a regenerative response (Romero-Pozuelo, 2017).

One additional interesting non-autonomous phenomenon observed is that sometimes when a region of the wing disc has high pS6 levels, the rest of the disc loses its typically patchy pS6 pattern and becomes pS6 negative. This phenomenon is not understood, and future work will be necessary to understand it molecularly (Romero-Pozuelo, 2017).

GENE STRUCTURE

date revised: 4 April 2022

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.