In the 1970s, a soil sample collected from the Vai Atare region of Rapa Nui (also known as Easter Island) in the Pacific Ocean was found to contain a bacterial strain of Streptomyces hygroscopicus producing a compound with potent antifungal activity (Ref. Reference Vezina, Kudelski and Sehgal1). The purified antibiotic, named rapamycin after the location where it was discovered, was characterised as a lipophilic macrolidic lactone – practically insoluble in water but soluble in most organic solvents. In addition to antifungal activity, rapamycin was found to have impressive immunosuppressive qualities (Refs Reference McAlpine2, Reference Sehgal3). The National Cancer Institute (NCI) identified rapamycin as a potent, mainly cytostatic agent against many types of human cancer (Refs Reference Douros and Suffness4, Reference Faivre, Kroemer and Raymond5). However, because of its low solubility in water and the fact that the patent for the drug had expired, the development of rapamycin as an antitumour agent was delayed considerably (Refs Reference Faivre, Kroemer and Raymond5, Reference Hidalgo6).

Intensive investigation of the mechanism of action of rapamycin led to the discovery of an evolutionarily conserved pathway that integrates signals arising from growth factors, nutrients and energy status. This pathway regulates metabolism, growth and proliferation in normal cells and plays a central role in cell proliferation, survival and drug resistance in many human cancers (Refs Reference Harris and Lawrence7, Reference Tee and Blenis8, Reference Bjornsti and Houghton9). The core of this pathway corresponds to the target of rapamycin (TOR), discovered originally through two mutations conferring resistance to the growth inhibitory effect of rapamycin in the budding yeast (Saccharomyces cerevisae) (Refs Reference Heitman, Movva and Hall10, Reference Kunz11, Reference Helliwell12, Reference Chen13). All eukaryotes studied to date contain a TOR gene. There are two yeast genes (TOR1 and TOR2); the mammalian homologue, named mammalian target of rapamycin (MTOR or FRAP1), is a single gene (Refs Reference Sabatini14, Reference Brown15, Reference Sabers16).

The core of the mTOR pathway: mammalian target of rapamycin complexes (mTORCs)

The MTOR gene maps to chromosome 1p36.2 in the human genome and encodes a 289 kDa protein that contains a kinase catalytic domain similar to that found in phosphoinositide 3-kinase (PI3K) (Refs Reference Brunn17, Reference Burnett18, Reference Cardenas and Heitman19). However, unlike PI3K, mTOR does not have lipid kinase activity and instead functions as a Ser/Thr protein kinase (Refs Reference Brunn17, 18). mTOR is the founding member of the PI3K-related kinase (PI3KK) family, which also includes DNA-dependent protein kinase, ATM (ataxia telangiectasia mutated), ATR (ATM- and Rad3-related) and SMG-1 (‘human homologue of suppressor with morphogenetic effect on genitalia 1’) (Ref. Reference Abraham20). In addition to the catalytic domain, the mTOR molecule contains 20 HEAT (Huntington elongation factor 1A–protein phosphatase 2A subunit–TOR) repeats that are crucial for protein–protein interactions, FAT (FRAP–ATM–TRAPP2) and FATC (FAT C-terminal) domains characteristic of the PI3KK family, a putative autoinhibitory repressor domain (RD) and an FKB12 rapamycin-binding domain (FRB). The FRB is the immediate target of rapamycin and rapamycin analogues (Fig. 1a) (Refs Reference Helliwell12, Reference Andrade and Bork21, Reference Bosotti, Isacchi and Sonnhammer22, Reference Peterson23, Reference Takahashi24, Reference Sekulic25, Reference Cafferkey26).

Figure 1

Mammalian target of rapamycin protein structure and complexes. (a) The human mammalian target of rapamycin (mTOR) gene encodes a protein of 2549 amino acids. Point mutation at the catalytic domain changing Asp2338 to Ala eliminates mTOR kinase activity. The FAT and FATC domains, flanking the catalytic domain are believed to be necessary for the appropriate folding or organisation of the catalytic domain and perhaps interact. The FRB region, just upstream of the catalytic domain, is characteristic of mTOR and the only region for which definite structural information exists. Mutation of Ser2035 to Arg in this region abolishes the binding of the FKBP12–rapamycin complex. The 20 HEAT repeats at the N-terminus, each ~39 amino acids long, contain several conserved hydrophobic residues. It is believed that each repeat forms a pair of antiparallel α-helices that stack in parallel with other repeats forming a large surface for protein–protein interactions. Between the catalytic and the FATC domain there is a poorly understood regulatory region containing two phosphorylation sites: Thr2446 and Ser2448. Deletion of this region results in activation of mTOR. Recently it was shown that both sites are phosphorylated by the downstream target of mTOR, S6K1, and not by AKT as it was initially believed. However, the functional significance of this regulatory loop between mTOR and S6K1 remains elusive. (b) There are at least two different mTOR protein complexes: mTORC1 and mTORC2. Most of the known functions of mTOR, and particularly the regulatory effect on protein translation and autophagy, are attributable to mTORC1 and can be inhibited by rapamycin. The two other proteins of the complex, GβL and Raptor, have no catalytic activity. The human gene for Raptor encodes a protein 1335 amino acids long, which contains a unique N-terminal region, three HEAT motifs and seven WD40 repeats. GβL protein, which is 260 amino acids long, consists almost entirely of seven WD40 repeats, is similar in sequence to the β subunits of heterotrimeric G proteins and interacts with the kinase domain of mTOR. Collectively, mTOR, GβL and Raptor have 23 HEAT and 14 WT40 motifs, increasing the potential of protein–protein interaction. No definite model exists for mTORC1 function; however, it seems that Raptor acts as adaptor that presents substrates to mTOR, and that Raptor and GβL both modulate mTOR activity. Many of the signals that regulate mTORC1 activity concentrate on the TSC1–TSC2 complex. For example, activated AKT or ERK pathways result in phosphorylation of TSC2 and inactivation of TSC1–TSC2. TSC1–TSC2, when active, inactivates the small protein Rheb, promoting the hydrolysis of GTP bound to Rheb. The protein that catalyses the opposite step remains elusive. Rheb seems to regulate mTOR indirectly, through FKBP38, a member of the FK506-binding protein (FKBP) family. FKBP38 binds to mTOR and inhibits its activity. Rheb interacts directly with FKBP38 and prevents its association with mTOR in a GTP-dependent manner. The second complex – mTORC2 – is known to influence cytoskeleton organisation and cell polarity and contains the evolutionary conserved protein Rictor (instead of Raptor), SIN1 and protor-1. SIN1 exists in at least three different isoforms defining three different mTORC2 complexes with different target specificities. However, all of these isoforms contribute to AKT activation, creating a regulatory loop for the activation of mTORC1. Insulin regulates the activity of two of these mTORC2 variations. However, the exact mechanism of mTOR2 activation and whether it is upstream of mTORC1 remains unclear. mTORC2 function is not generally inhibited by rapamycin. Abbreviations: FAT, FRAP–ATM–TRAPP2; FATC, FAT C-terminal; HEAT, Huntington elongation factor 1A–protein phosphatase 2A subunit–TOR; FKBP12, FK507-binding protein 12; FRB, FKBP-rapamycin binding; mTORC1/2, mammalian target of rapamycin complex 1/2.

mTOR functions as part of a larger multiprotein signalling complex. Recent studies show that mTOR participates in at least two different signalling complexes – mTORC1 and mTORC2 – which have different functions (Fig. 1b) (Ref. Reference Loewith27). mTORC1, which is sensitive to rapamycin, is currently believed to have three components: mTOR, Raptor (regulatory-associated protein of mTOR; an adaptor protein approximately half the size of mTOR) and GβL (G protein β-subunit-like protein, also known as mLST8; a 36 kDa protein believed to be important for sensing upstream signals) (Refs Reference Kim28, Reference Hara29). Both Raptor and GβL are evolutionary conserved. However, the exact relationship between these components and their functions are not completely understood. It is believed that upstream signals associated with growth factors, energy status, nutrient availability and various conditions of stress, such as hypoxia, reactive oxygen species, DNA damage and hyperosmotic environment regulate the strength of association between mTOR and Raptor and, in this way, the recruitment of various downstream substrates of mTOR (Refs Reference Kim30, Reference Sarbassov31). In turn, activation of mTORC1 promotes ribosome biogenesis and the rates of translation and protein synthesis. Activation of mTORC1 also inhibits autophagy, influencing metabolic mass, cell proliferation and cell survival (Ref. Reference Sarbassov, Ali and Sabatini32). All these functions are inhibited by rapamycin or rapamycin analogues (Ref. Reference Kim30).

mTORC2, which is generally rapamycin insensitive, contains at least four components: mTOR, GβL, Rictor (rapamycin-independent companion of mTOR; a ~200 kDa adaptor protein) and mSIN1 (mammalian stress-activated protein kinase interacting protein 1) (Refs Reference Sarbassov33, Reference Jacinto34, Reference Jacinto35, Reference Frias36). Very recently a fifth component of undetermined significance, named protor-1, was discovered (reviewed in Ref. Reference Guertin and Sabatini161). mTORC2 is regulated by unknown upstream signals, and the exact mechanisms of mTORC2 activation are poorly understood. However, it is believed that various growth factors contribute to mTORC2 activation. Compared with mTORC1, the function of mTORC2 is less well studied, but it is known that mTORC2 activity regulates cytoskeleton organisation and promotes activation of AKT (also known as protein kinase B) (Refs Reference Sarbassov31, Reference Sarbassov33, Reference Jacinto34). Since AKT is an important upstream regulator for the activation of mTORC1, an autoregulatory loop between the two complexes seems to exist (Refs Reference Sarbassov31, Reference Sarbassov, Ali and Sabatini32).

Downstream effectors of mTOR

Although the regulation of protein production has been studied mainly at the genetic and transcriptional levels, it is becoming clear that regulation of translation is equally important (Ref. Reference Ruggero and Pandolfi37). mTORC1 acts mostly at this level, regulating the rate of mRNA translation and ribosome biosynthesis. Two critical mTORC1 downstream effectors of translation are the eukaryotic initiation factor 4E (eIF4E) binding protein 1 (4E-BP1) and p70 ribosomal protein S6 kinase (S6K1) (Fig. 2) (Refs Reference Beretta38, Reference Price39, Reference Chung40, Reference Hay and Sonenberg41).

Figure 2

Downstream mediators of mammalian target of rapamycin. 4E-BP1, S6K1 and eEF2 kinase are among the crucial downstream mediators of mammalian target of rapamycin (mTOR) that orchestrate the response of the translational machinery to various stress signals and growth factors. In the inactive state, 4E-BP1 blocks eIF4E from binding the cap structure of mRNA, and the translation elongation factor eEF2 is phosphorylated and inactivated by the corresponding eEF2 kinase. In the active state, activated mTORC1 phosphorylates and neutralises 4E-BP1, so that eIF4E can bind the cap structure of mRNA. In addition, S6K1, activated by mTORC1, phosphorylates and activates the eIF4B RNA helicase. Activated S6K1 phosphorylates and inhibits eEF2 kinase so that the underphosphorylated and active eEF2 can contribute to the elongation step of translation. Finally, activation of mTORC1 enhances ribosome biogenesis through many intermediate steps generally promoting the rate of translation. Abbreviations: 4E-BP1, 4E-binding protein-1; eEF2, eukaryotic elongation factor 2; eIF-4E, eukaryotic initiator factor 4E; mTORC, mammalian target of rapamycin complex; S6K1, p70 ribosomal protein S6 kinase.

The rate of translation of most eukaryotic mRNAs is dependent on the binding of initiator factors (eIF) to the cap (m⁷GpppN – where N is any nucleotide) structure at the 5′ untranslated mRNA terminus (Ref. Reference Gingras, Raught and Sonenberg42). The eIF4E 5′CAP-binding subunit of the eIF4F initiator complex is recruited and neutralised by 4E-BP1. Direct or indirect phosphorylation of 4E-BP1 by activated mTORC1 disrupts the association between 4E-BP1 and eIF4E, facilitating the 5′CAP-binding function of eIF4E and thereby promoting translation (Fig. 2) (Refs Reference Gingras, Raught and Sonenberg42, Reference Gingras, Raught and Sonenberg43, Reference Wang and Proud44).

S6K1, which is phosphorylated and activated by mTORC1 and PDK1, was previously thought to play a critical role in the translation of mRNAs with the TOP (terminal oligopyrimidine) sequence at the 5′ untranslated region, among which are many that encode components of the translational machinery (Ref. Reference Meyuhas45). However, more recent studies have shown that S6K1 (and the similar S6K2), as well as the phosphorylation status of its well-known target, ribosomal protein S6, are dispensable for 5′TOP mRNA translation, although mTORC1 activation remains crucial (Refs Reference Pende46, Reference Tang47, Reference Stolovich48). Recent studies have established two more connections between mTORC1-mediated S6K1 activation and translation (Ref. Reference Peterson and Sabatini49). First, activated S6K1 phosphorylates the RNA helicase eIF4B, promoting the association of eIF4B with the eIF3 scaffold component of the initiator complex, and the association of eIF3 with the eIF4G scaffold component of the eIF4F initiator complex (Refs Reference Holz50, Reference Shahbazian51). Second, activated S6K1 phosphorylates and inactivates the eukaryotic elongation factor 2 (eEF2) kinase, resulting in underphosphorylated and functional eEF2. The latter is important for the promotion of the elongation steps of translation (Fig. 2) (Refs Reference Browne and Proud52, Reference Wang53).

Regulation of ribosomal biogenesis is a central mechanism for the control of global levels of translation and cell growth in eukaryotes, and is tightly regulated by mTORC1 (Ref. Reference Mayer and Grummt54). Activation of mTORC1 promotes the biosynthesis of ribosomal RNAs and proteins by activating transcription, mediated by RNA polymerase (Pol) I, Pol II and Pol III in the nucleus and accelerating the translation of 5′TOP mRNA into ribosomal proteins in the cytoplasm (Ref. Reference Mayer and Grummt54). Although the exact mechanisms for the regulation 5′TOP mRNA in the cytoplasm remain elusive, it seems likely that enhancement of transcription in the nucleus by Pol I is mediated partly by S6K1, which activates and increases the availability of TIF-IA (transcription initiator factor IA) and UBF (upstream binding factor), essential cofactors of Pol I (Refs Reference Mayer55, Reference Hannan56).

Autophagy, more specifically macroautophagy, the process by which cellular components including organelles are catabolised and recycled, is an important survival mechanism that is tightly regulated by mTORC1 (Ref. Reference Shintani and Klionsky57). The exact mechanisms by which activated mTORC1 inhibits autophagy are poorly understood, although it seems probable that negative regulation of a homologue of ATG1 kinase that mediates early activation steps of the autophagic process is involved (Refs Reference Kamada58, Reference Mordier59).

Downstream targets of mTORC2 that mediate cytoskeleton reorganisation seem to include the small GTPases Rho and Rac, and the protein kinase PKCα (Refs Reference Sarbassov33, Reference Jacinto34). However, knowledge about the pathway downstream of mTORC2 is limited.

Upstream regulators of mTOR

Pivotal to the understanding of upstream regulation of mTOR activity was the discovery of the importance of TSC1 and TSC2 (tuberous sclerosis complex 1 and 2, also known as amartin and tuberin, respectively) (Refs Reference Tapon60, Reference Gao61). Lack of TSC1 or TSC2 expression results in overactivation of mTOR (Refs Reference Goncharova62, Reference Kwiatkowski63). The heterodimer of TSC1/TSC2 is a GTPase-activating protein (GAP) for the small GTP-binding protein Rheb (RAS homologue enriched in brain), which catalyses the hydrolysis of Rheb-bound GTP and promotes the conversion of Rheb-GTP to Rheb-GDP (Refs Reference Saucedo64, Reference Stocker65, Reference Zhang66, Reference Tee67). In turn, Rheb-GTP activates mTOR kinase (Ref. Reference Long68). Signals regulating mTORC1 activity including those associated with growth factors, energy status and various stressful stimuli (e.g. hypoxia) target the TSC1/TSC2 heterodimer (Fig. 1B) (Refs Reference Sarbassov, Ali and Sabatini32, Reference Hay and Sonenberg41).

Two of the most important mTORC1-regulating pathways are the AKT and the Ras–ERK (extracellular-signal-regulated kinase)/MAPK (mitogen-activated protein kinase) pathways emanating from cell membrane growth-factor-activated tyrosine kinase or seven-transmembrane receptors (Refs Reference Cantley69, Reference Johnson and Lapadat70, Reference Roux and Blenis71). A critical step for AKT activation is phosphorylation at the D3 position of the inositol ring of the plasma membrane lipid phosphatidylinositol 4,5-bisphosphate [PtdIns(Reference Douros and Suffness4,Reference Faivre, Kroemer and Raymond5)P ₂] by PI3K. PI3K is activated by interaction with activated receptor tyrosine kinases (RTKs) and/or associated adaptors, such as those of the IRS family. Phosphorylation of PtdIns(Reference Douros and Suffness4,Reference Faivre, Kroemer and Raymond5)P ₂ produces PtdIns(Reference Sehgal3,Reference Douros and Suffness4,Reference Faivre, Kroemer and Raymond5)P ₃, which offers plasma membrane docking sites for proteins that contain the pleckstrin homology (PH) domain, such as AKT and PDK1 (phosphoinositide-dependent kinase). In this way, PtdIns(Reference Sehgal3,Reference Douros and Suffness4,Reference Faivre, Kroemer and Raymond5)P ₃-associated AKT is phosphorylated in the catalytic domain by the proximal PtdIns(Reference Sehgal3,Reference Douros and Suffness4,Reference Faivre, Kroemer and Raymond5)P ₃-associated PDK1 and becomes activated (Ref. Reference Cantley69). However, full activation of AKT requires a second phosphorylation at the conserved C-terminal hydrophobic motif by a second ‘PDK2’ kinase. A recent study has shown that this ‘PDK2’ is, in fact, mTOR kinase nucleated at mTORC2 (Ref. Reference Sarbassov31). Activated AKT, in turn, phosphorylates TSC2 and inactivates the TSC1/TSC2 heterodimer, neutralising the inhibitory activity of TSC1/TSC2 on mTORC1 (Refs Reference Gao61, Reference Inoki72, Reference Potter, Pedraza and Xu73).

In the Ras–ERK pathway, activated RTKs, through adaptors, recruit guanine-nucleotide exchange factors, such as SOS (son of sevenless), which stimulate the exchange of GDP for GTP on the GTPase Ras, allowing the activation of downstream effectors such as the MAPKKK Raf. It has been shown that downstream effectors of the Ras–ERK pathway – ERK (a MAPK) and the MAPK-activated protein kinase (MP) p90 ribosomal S6 kinase – can phosphorylate TSC2 and inactivate the TSC1/TSC2 heterodimer (Refs Reference Ma74, Reference Ballif75, Reference Roux76).

The AMP-activated protein kinase (AMPK) has a central place in the energy-status-associated mTOR signalling pathway (Ref. Reference Wullschleger, Loewith and Hall77). AMPK is a Ser/Thr kinase that senses the AMP:ATP ratio and is activated by conditions that impinge on the energy balance of the cell, thereby increasing the relative AMP concentration. Fully activated AMPK, for which the additional step of its phosphorylation by the Ser/Thr kinase LBK1 is necessary, is able to phosphorylate TSC2, inactivating TSC1/TSC2 (Refs Reference Inoki, Zhu and Guan78, Reference Corradetti79, Reference Shaw80). AMPK regulation of mTORC1 is also noteworthy because it has been shown to be used by DNA-damage-response signals mediated by activated p53 (Ref. Reference Feng81).

When there is a scarcity of nutrients (particularly amino acids leucine and arginine), the activity of the mTORC1 is low; nutrient abundance boosts mTORC1 activity (Ref. Reference Wullschleger, Loewith and Hall77). Where the exact nutrient input signal is localised within the mTOR regulatory network is controversial. There are contradictory data placing the input above the level of TSC1/TSC2, or lower, at Rheb, or even at the level of mTORC1. Recently, it was shown that hVps34, a class III PI3K (totally unrelated to the PI3K mentioned in the AKT pathway), can be activated by amino acids and inhibited by AMPK (Refs Reference Byfield, Murray and Backer82, Reference Nobukuni83). However, the exact mechanism by which hVps34 regulates mTORC1 activity, and whether this occurs in parallel with the TSC–Rheb–mTORC1 pathway, remains uncertain.

In addition, the links between many stress signals, including those associated with hyperosmotic conditions, overproduction of reactive oxygen species, mechanical stressors and the mTOR regulatory network is generally poorly understood (Ref. Reference Reiling and Sabatini84). Interestingly, the hypoxia input that culminates in mTORC1 downregulation was shown recently to be above the TSC1/TSC2 level and to depend on a hypoxia-induced 34 kDa protein, Redd1/RTP801 (Refs Reference Brugarolas85, Reference Reiling and Hafen86). However, the exact mechanism by which Redd1 leads to TSC1/TSC2 inhibition remains unknown.

mTOR is a major integrator of oncogenic pathways

The importance of the mTOR pathway for human cancer may be better appreciated after realising that almost every member of the mTOR regulatory network behaves as an oncogene or tumour suppressor gene. Although mutations of the MTOR gene itself have not been described in cancers to date, genetic alterations or deregulated expression of RTKs, RAS, ERK, PI3K, PTEN, AKT, TSC1/TSC2, 4EBP-1, eIF4E and S6K1 have been identified in many types of cancer (Refs Reference Faivre, Kroemer and Raymond5, Reference Shaw and Cantley87, Reference Cully88). In addition, germline genetic alterations of PTEN, TSC1/TSC2, NF1 and LBK1 are the cause of various tumour-prone syndromes, each one characterised by a specific spectrum of neoplasms (Refs Reference Faivre, Kroemer and Raymond5, Reference Shaw and Cantley87, Reference Cully88).

The mechanisms by which overactivation of the mTOR pathway promotes the oncogenic phenotype are under intense investigation and are not fully elucidated. However, it seems that both deregulation of autophagy and protein translation by mTOR play major roles. On the one hand, suppression of autophagy (also a consequence of mTOR activation) by inactivation of beclin – an important downstream autophagy regulator – is oncogenic in animal models (Refs Reference Liang89, Reference Yue90). On the other hand, the translation of many proteins that promote cell-cycle progression, including cyclins and other oncogenes (e.g. Myc), are promoted by mTOR activation (Ref. Reference Averous and Proud91). Accordingly, the mTOR downstream regulator of translation eIF4E has been shown to be oncogenic in many studies including animal models of lymphoma (Refs Reference Ruggero92, Reference Wendel93).

Clinical aspects of mTOR signalling in lymphomas

The therapy of patients with haematological malignancies, including lymphomas, is in flux because efforts are focusing on the development of biologically targeted therapies, as opposed to the traditional, and mostly empirical, chemotherapeutical approaches. Here we summarise the findings of recent studies regarding the activation status, regulation and downstream targets of mTOR signalling in lymphomas (Table 1), as well as clinical efforts to exploit mTOR inhibition as part of clinical trials of lymphoma patients.

Table 1

mTOR signalling in lymphomas

Abbreviations: AKT, akt murine thymoma viral oncogene homologue; BCL-XL, long isoform of the BCL-2-related gene; BCR, B-cell receptor; BCR–ABL, fusion product of breakpoint cluster region gene on chromosome 22 and Abelson kinase; C/EBPβ; CCAAT enhancer binding protein β; cIAP1, cellular inhibitor of apoptosis 1; 4E-BP1, 4E-binding protein 1; eIF-4E, eukaryotic initiator factor 4E; ERK, extracellular-signal-regulated kinase; FLIP, FLICE-inhibitory protein; IGF, insulin-like growth factor; IL, interleukin; MCL-1, myeloid cell leukaemia 1; NF, nuclear factor; PEL, pleural effusion lymphoma; S6K1, p70 ribosomal protein S6 kinase; SKP2, S-phase kinase-associated protein 2; Syk, spleen tyrosine kinase; VEGF, vascular endothelial growth factor; XIAP, X-linked inhibitor of apoptosis.

Pharmacological inhibitors of mTOR

In addition to the prototype mTOR inhibitor rapamycin (sirolimus), three more mTOR inhibitors are available for clinical trials: CCI-779 (temsirolimus), RAD001 (everolimus) and AP23573 (Refs Reference Faivre, Kroemer and Raymond5, Reference Dancey150). All inhibitors are derivatives of rapamycin with which they share the same basic chemical structure; however, each has a different chemical group at the position of the C40 hydroxyl group of rapamycin. CCI-779 is essentially a prodrug because it is metabolised to rapamycin inside the organism, whereas RAD001 and AP23573 are metabolised differently (Fig. 3). To inhibit mTOR, all must initially form an intracellular complex with the immunophilin FKB12. This complex, in turn, binds the FRB domain of mTOR, inhibiting mTOR function. The inhibition of mTOR by rapamycin and its derivatives is regarded as quite specific, and rapamycin treatment of mice embryos induces the same developmental defects as knockout of the mTOR gene (Refs Reference Murakami151, Reference Gangloff152). Until recently, it was thought that rapamycin and its derivatives inhibit only the mTOR protein nucleated in mTORC1, which is mostly responsible for regulation of protein translation and autophagy. However, it has now been shown that prolonged treatment with rapamycin, in some types of cells, also can inhibit mTORC2, which is mainly responsible for activation of AKT (Ref. Reference Sarbassov153). This finding can be explained by the assumption that within the cell there exists an equilibrium between the pool of mTOR associated with Raptor (mTORC1) and that of mTOR associated with Rictor (mTORC2). Inhibition of mTORC2 by prolonged treatment with rapamycin is attributed to the exhaustion of mTOR available for binding to Rictor, because of the direct inhibition of mTOR bound to Raptor (Ref. Reference Sarbassov153).

Figure 3

Pharmacological inhibitors of mammalian target of rapamycin. The basic chemical structure of rapamycin is shared among all rapamycin derivatives (left). The shaded area corresponds to the position of different chemical groups in the derivatives shown below. This area is distinct from the rapamycin sites for binding to immunophilin FKBP12 and mammalian target of rapamycin (mTOR), so that the mechanism of mTOR inhibition is common for rapamycin and its analogues. Rapamycin, as well its analogues, associates initially with FKBP12, in order to inhibit mTOR (right). The combined surface of the rapamycin–FKBP12 complex binds, in turn, the FRB domain of mTOR in mTORC1 and inhibits its activity. It seems that FKBP12 presents rapamycin to mTOR in a conformation-favourable state for interaction with the FRB domain. Immunophilins, such as FKBP12, belong to a protein family with at least 12 members in humans, and function as accessory helper enzymes or chaperones, contributing to the stability of multiprotein complexes. A common feature of immunophilins is they can catalyse the interconversion of cis and trans amide bonds adjacent to proline residues in peptide substrates, the so-called PPIase (prolyl isomerase) activity altering the spatial rearrangement of protein domains. This activity is inhibited by binding of each immunophilin with the corresponding drug. However, it has been shown that inhibition of PPIase activity of FKBP12 by rapamycin binding is not essential for mTOR inhibition. Abbreviations: FKBP12, FK507 binding protein 12; FRB, FKBP-rapamycin binding; mTORC, mammalian target of rapamycin complex.

Unlike rapamycin, its derivatives do not induce significant immunosuppression when administered intermittently in the appropriate clinical setting (Refs Reference Faivre, Kroemer and Raymond5, Reference Dancey150). Rapamycin is produced in oral formulation for immunosuppression after kidney transplantation. CCI-779 (Wyeth Pharmaceuticals, Madison, NJ, USA) can be administered orally or intravenously. RAD001 (Novartis, Basel, Switzerland) is available in oral formulation and is in the advanced stage of development for use as an immunosuppressive in solid-organ transplantation. AP23573 (Ariad Pharmacheuticals, Cambridge, MA, USA) can be given orally or intravenously. Side effects common to all mTOR inhibitors include hypertension, oedema, hyperlipidaemia and gastrointestinal symptoms. Also, dose-limiting toxicities of rapamycin derivatives include reversible mucositis, asthaenia and thromphocytopaenia (Refs Reference Faivre, Kroemer and Raymond5, Reference Dancey150).

Clinical trials of mTOR inhibitors in lymphomas

Although rapamycin has been studied meticulously as an immunosuppressive drug, clinical experience using it as an antineoplastic agent is limited, especially for haematological malignancies (Ref. Reference Witzig and Kaufmann154). However, all rapamycin derivatives are currently used in Phase I/II clinical trials as antineoplastic agents and some Phase III clinical trials are in progress (Refs Reference Faivre, Kroemer and Raymond5, Reference Witzig and Kaufmann154, Reference Costa155).

One of the more promising results published to date regarding using mTOR inhibitors in lymphoma patients comes from a Phase II trial using CCI-779 in the treatment of MCL patients (Ref. Reference Witzig156). In this study, 34 patients with MCL, either refractory to previous therapy or relapsed, received CCI-779 as a single agent. Twelve patients showed partial response and one patient had a complete response, with an overall response rate of 38%. No deaths from the treatment were reported. The most frequent toxicity, which was also a cause for dose reduction, was thrombocytopaenia. Interestingly, reduction of the dose (from 250 mg to 25 mg intravenously once a week) seemed to decrease toxicities without affecting the response rate. Currently an international randomised Phase III trial comparing CCI-779 with control chemotherapy in MCL patients is ongoing.

In a Phase I/II trial of RAD001, which included four patients with MCL and six with CLL/SLL, three patients showed a 27–34% reduction of lymphadenopathy in radiological studies, and the drug was considered safe and worthy of further investigation (Ref. Reference Yee157). In another trial using AP23573 to treat patients with lymphoid malignancies one of nine CLL/SLL patients showed a partial response (Ref. Reference Costa155).

Additional clinical trials using rapamycin derivatives to treat Hodgkin lymphoma, PCM and other lymphomas, as monotherapy or in combination with other agents such as rituximab, are in progress, but the results are not yet published.

Research in progress and outstanding questions

Since preliminary clinical studies of mTOR inhibitors in cancer patients have shown promising results, further investigation will continue to accumulate considerable empirical knowledge about the spectrum of neoplastic diseases that will be successful targets for mTOR inhibitors. This will lead to better drug design and dosing, as well as the development of combinations of mTOR inhibitors with classical chemotherapeutical agents or new cell-signalling inhibitors that will be beneficial in various clinical settings.

However, many questions regarding the mTOR-signalling pathway that have clinical implications remain to be answered. What is the molecular signature that determines mTOR inhibition sensitivity in each clinical setting? Preliminary in vitro studies in PCM cells have shown that PTEN mutation and AKT overactivation are the most critical parameters of rapamycin sensitivity (Ref. Reference Shi141). Does this hold true for other lymphoid malignancies and, if not, what are the critical pathways that contribute to mTOR activation in each case? Also, why does mTOR inhibition induce only cell-cycle arrest in some lymphomas, whereas in others mTOR inhibition also induces apoptotic cell death? Regarding the resistance of some cancer cells to rapamycin, it seems that the internal wiring of the mTOR network may play an important role. For example, in vitro studies in PCM cells showed in a subset of cases that rapamycin treatment induced overactivation of AKT. This was due to decreased IRS serine phosphorylation as a result of cancellation of the inhibitory feedback between the downstream mTOR effector S6K1 and the IRS–PI3K link upstream of AKT (Ref. Reference Shi158). In which other neoplasms is this feedback connection important for rapamycin resistance? Will simultaneous inhibition at the level of AKT and mTOR by two independent inhibitors be able to overcome rapamycin resistance, as suggested in some studies (Refs Reference Shi158, Reference Sun159)? Some recent in vitro results suggest that prolonged rapamycin treatment – in addition to directly inhibiting mTORC1 – also has an indirect inhibitory effect on mTORC2 and therefore AKT activation through mTOR depletion (Refs Reference Sarbassov153, Reference Zeng160). Why is this true in some types of neoplasm and not in others, and is this a critical parameter of rapamycin resistance? The answers to these questions and many others of importance to our understanding of biology and patient care are likely to be forthcoming.