including directly targeting raptor, a compo- nent of rapamycin-sensitive mTORC1 (ref. 5). Both AMPK and mTOR sense and control the energy status of a cell (ATP:AMP ratio) and regulate key aspects of cell growth and, as part of this, glucose metabolism. In a quiescent cell, most energy (in the form of ATP) is generated in the mitochondria through oxidative phosphorylation, includ- ing the oxidation of fatty acids and amino acids — catabolic metabolism. On activation, T cells massively increase their glucose uptake and shift to producing ATP by glycolysis (anabolic metabolism) instead of cataboli- cally (Fig. 1). mTOR is activated by signalling molecules, growth factors and antigen-induced T-cell-receptor signalling, and its activity enables a cell to increase glycolysis and ATP accumulation, which opposes AMPK activa- tion6. Although some of the processes involved in the switch from catabolic to anabolic metab- olism are fairly well understood, the reversal from an anabolic to a catabolic state is not as well characterized. One could speculate that rapamycin and metformin facilitate the switch from a glucose-dependent anabolic state (effec- tor T cell) to a catabolic state of metabolism (memory T cell) by blocking mTORC1 activ- ity (Fig. 1). But how a change in the metabolic signature of a T cell could enhance memory T-cell numbers and function is unknown. How can rapamycin, a drug known for its immunosuppressive effects, enhance the func- tion and formation of T-cell memory? The answer may lie in dosage and, more importantly, timing. Whereas treatment with a low dose of rapamycin during the first 8 days after T-cell activation enhanced the numbers and function of memory T cells, a higher dose, closer to ther- apeutic levels, hampered the T-cell response, as would be expected of an immunosuppres- sant3. Interestingly, both papers2,3 clearly show that the higher dose of rapamycin enhanced memory T-cell function and recall ability if administered after day 8. At this point, the vigorous cell proliferation that is characteristic of the effector stage has ceased and cells begin to enter the more quiescent memory state. Recent data4 suggest that mTOR can form dif- ferent complexes (aside from mTORC1 and mTORC2) depending on the phase of the cell cycle, and little is known about their inter action with rapa mycin. In addition, as the metabolic signature of a cell changes along with its acti- vation state, rapamycin might differentially affect a cell depending on its cell cycle and metabolic state. A long-standing paradigm in immunology proposes that, after the peak of the prolifera- tive response, the programmed cell death of effector T cells is caused by a lack of growth and survival factors — conditions that could also affect cell metabolism. However, recent experiments7 indicate that, in a physiological setting, effector T-cell viability and conversion to memory T cells are not regulated by compe- tition for growth and survival factors. Thus, it is more likely that the metabolic switch is either programmed early after T-cell activation or occurs as a secondary effect after a quiescent stage has been entered. Both Pearce et al.2 and Araki et al.3 establish a crucial role for mTOR-mediated metabolic changes in enhancing T-cell memory. Does changing the metabolism of T cells through manipulation of mTOR hold promise for improving future vaccination strategies? mTOR is involved in regulating a plethora of functions in many cell types, and rapamycin administration is associated with many side effects. Thus, a more targeted approach will be required to harness their memory-enhancing ability. Identifying the downstream signalling pathways that lead to enhanced T-cell memory on inhibition of mTOR complexes will be a first step in that direction. ■ Martin Prlic and Michael J. Bevan are in the Department of Immunology, University of Washington, Seattle, Washington 98195-7370, USA. e-mails:
[email protected]; mbevan@ u.washington.edu 1. Williams, M. A. & Bevan, M. J. Annu. Rev. Immunol. 25, 171–192 (2007). 2. Pearce, E. L. et al. Nature 460, 103–107 (2009). 3. Araki, K. et al. Nature 460, 108–112 (2009). 4. Thomson, A. W., Turnquist, H. R. & Raimondi, G. Nature Rev. Immunol. 9, 324–337 (2009). 5. Gwinn, D. M. et al. Mol. Cell 30, 214–226 (2008). 6. Jones, R. G. & Thompson, C. B. Immunity 27, 173–178 (2007). 7. Prlic, M. & Bevan, M. J. Proc. Natl Acad. Sci. USA 105, 16689–16694 (2008). NANOOPTICS Photons pushed together Michel Orrit Photons don’t interact well with each other, which is a real headache for researchers developing all-optical transistors for computing applications. But a single molecule can mediate photon–photon affairs. Transistors in computers generally modify electrical signals (source currents) using other electrical signals (gate voltages). The nonlinear interactions in and between networks of tran- sistors thus underpin all electronic computing. Similar nonlinear networks can be found in the brain1, where neuron firing is triggered by the outputs of other neurons, or in microfluidic chips2, in which fluid pressures govern fluid flow. But on page 76 of this issue, Hwang et al.3 describe how the transmission of a light beam can be modulated by another light beam: the basic process of an all-optical transistor. Optical transistors are a long-sought goal because they could form the basis of optical computers that use photons instead of electrons as signal carriers. Not only are photonic signals faster than their electronic counterparts, but they are also easier to transmit directly over large distances. And because photons carry no electrical charge, photonic signals do not per- turb each other (unlike electronic signals). Fur- thermore, photons offer advantages for future quantum-computing applications, because the coherence of their quantum states can be maintained at practically useful temperatures; Figure 1 | The metabolic state of T-cell memory. Naive T cells that have not been exposed to antigen are quiescent, but undergo metabolic conversion (catabolic metabolism to anabolic metabolism) on stimulation with an antigen such as a pathogen. This switch allows effector T cells to use mTORC1- dependent glycolytic energy production (anabolic metabolism) to sustain rapid proliferation and biosynthetic needs. At the end of the effector stage, T cells either die by programmed cell death or enter the quiescent memory stage and switch back to catabolic metabolism. Pearce et al.2 and Araki et al.3 show that rapamycin and metformin can enhance memory T-cell formation by inhibiting the protein complex mTORC1, thus leading to changes in cell metabolism. Naive T cells Memory T cellsEffector T cells Anabolic metabolismCatabolic metabolism Catabolic metabolism Proliferation Cytokines Antigen Switch to memory T cells mTORC1 AMPK mTORC1 Rapamycin Metformin 42 NATURE|Vol 460|2 July 2009NEWS & VIEWS 39-47 News and Views MH IF.indd 4239-47 News and Views MH IF.indd 42 26/6/09 17:26:1226/6/09 17:26:12 © 2009 Macmillan Publishers Limited. All rights reserved the same feat in electrons is possible only at temperatures below 1 kelvin (ref. 4). Previous attempts at all-optical data process- ing have been thwarted by the weakness of photon–photon interactions. Such inter- actions depend on the strengths of the electric fields of the photons, but these fields are far weaker than those of the atoms and molecules of the medium through which they travel. The interactions between photons, which have to be mediated by atoms (or molecules), are therefore usually negligible. Hwang et al.3 have overcome this problem by using a sin- gle absorber — an atom, a quantum dot or a molecule — to mediate photon–photon interactions. The basic principle is as follows. A light beam (the gate beam) is aimed at a suitable dye molecule, so that the molecule absorbs every photon and enters an excited state. The excited molecule can then alter the propagation of pho- tons from another incident beam (the source beam). In effect, the molecule ‘slows down’ gate photons so that they can interact with source photons in a nonlinear way. The single mol- ecule is analogous to a crossing between a rail- way line (the gate beam) and a road (the source beam). For short, fast trains, the road traffic is hardly perturbed. However, if trains are long and slow, every train blocks a large number of cars, heavily perturbing road traffic. Unfortunately, the probability of photon absorption by a single molecule is usually tiny, so that merely detecting the nonlinear interaction of optical waves at a molecule is difficult. But theoretical studies show that, for tightly focused light beams that cause ‘life- time-limited’ molecular transitions (which create excited states that relax to the ground state only by emitting photons, rather than by other mechanisms), the probability grows and can even reach 100% (ref. 5). So, for example, a molecule that reflects source photons in its ground state might absorb gate photons and enter an excited state that is transparent to source photons. It has therefore been pro- posed6 that a single gate photon could switch a whole beam of source photons from being reflected to being transmitted (Fig. 1). Hwang et al.3 provide the first experimental proof of this concept. In their experiment, Hwang and colleagues excited a dye molecule to a vibrational sublevel of its electronic excited state using a continu- ous laser beam (the gate beam). The authors needed to be able to distinguish between the gate and source beams, and the easiest way to do this is to use different frequencies for each beam. But different frequencies wouldn’t ordi- narily be able to excite the absorber molecule to the same electronic state. The authors there- fore chose a system in which both the gate and source beams can access the same electronic state, albeit at different vibrational sublevels. This is rather inefficient for practical appli- cations, because the probability of photon absorption generating vibrational excited states is small (about 1 in 1,000), but for the purposes of Hwang and colleagues’ experiment it was a convenient ploy. The authors probed photon absorption of the molecule’s main electronic state using a laser source beam that produced light at the same narrow range of frequencies as those absorbed by the molecule. They found that the molecule allowed more source photons to pass through after it had been irradiated by the gate beam. Again, the design of the experi- ment is not perfect, because the desired effect is reduced by relaxation of the excited state to vibrational sublevels of the ground state. Nonetheless, the authors observed that the absorption of just one gate photon led to an approximately 10% modulation of the source beam. Because the source beam is faint, this corresponds to each gate photon interacting with one source photon. This demonstration3 of photon–photon interactions in an all-optical transistor is a mile- stone in the field, but real applications are still far away. Too many gate photons are required to create an effect, and the modulation of the source beam is too weak to act as a signal for another transistor in a network. Furthermore, because both source and gate beams excite the absorber molecule to the same electronic state, one gate photon cannot efficiently influence the behaviour of more than one source photon. This is a problem, because in real-world appli- cations, the source beam will be much stronger (consisting of more photons) than the gate beam, and so the proportion of source photons perturbed by gate photons will be smaller than in Hwang and colleagues’ experiment (which uses a weak source beam). So how could this device be developed into a practical all-optical transistor? First, it would be better to use a metastable state rather than an excited state6, because a single gate photon would then enable the molecule to transmit many source photons for a long time. For example, the lifetime of a ‘triplet’ metastable state is a thousand to a million times longer than that of the excited state used in Hwang and colleagues’ experiment. Second, the ideal absorber for all-optical transistors would involve transitions to just two excited levels, whereas the light absorption of all molecules in condensed matter corresponds to many tran- sitions, each associated with different vibra- tional sublevels of electronic states. An atom would be much closer to the ideal three-level system, but atoms cannot easily be immo- bilized in optical circuits. A possible alter- native would be to enhance the coupling of light to two molecular transitions only — one for the source and one for the gate — by concentrating the optical field so that the frequency matches only one vibrational sub- level. This can be done using devices known as optical cavities7, or by using a ‘lightning-rod effect’ close to a metallic nanoantenna8. Hwang and colleagues’ work3 is not only an experimental tour de force, but it also provides a clear road map towards practical, single- molecule switching systems, and identifies the major obstacles to be overcome. The authors use an organic molecule at low temperatures as the absorber for their study, because this offers practical advantages for demonstra- tion purposes. But the underlying principle of their transistor applies to many other materials9, ranging from cold atoms and trapped ions to quantum dots. So the future is certainly bright for photon manipulations at molecular scales. ■ Michel Orrit is in the Huygens Laboratory, Leiden Figure 1 | Design for an all-optical transistor. A single molecule, represented here as a rotating mirror, can in principle behave as an all-optical transistor — it can modulate the transmission of a beam of light (the source beam, blue) in response to another beam of light (the gate beam, red). The waist-shaped surface represents a beam of light, focused on the molecule. The diagrams under each of the transistors represent the electronic energy levels of the molecule. a, If the molecule is in its ground state (g) and the source photons are equivalent in energy to the electronic energy transition from g to an excited state (e), then the source photons are resonantly scattered (totally reflected) as electrons oscillate between the e and g states. b, A gate photon of appropriate energy (different from that of the source photons) excites the molecule to a long-lived excited state (s). c, The excited molecule no longer absorbs source photons, which are instead perfectly transmitted. Hwang et al.3 report the first all-optical transistor that works on similar principles. a b c e s g e s g e s g Electron Energy 43 NATURE|Vol 460|2 July 2009 NEWS & VIEWS 39-47 News and Views MH IF.indd 4339-47 News and Views MH IF.indd 43 26/6/09 17:26:1426/6/09 17:26:14 © 2009 Macmillan Publishers Limited. All rights reserved Institute of Physics, Leiden University, 2300 RA Leiden, the Netherlands. e-mail:
[email protected] 1. Livet, J. et al. Nature 450, 56–62 (2007). 2. Psaltis, D., Quake, S. R. & Yang, C. Nature 442, 381–386 (2006). 3. Hwang, J. et al. Nature 460, 76–80 (2009). follicles showed that extra TERT enhances the expression of genes targeted by β-catenin7. How- ever, exactly how TERT affects gene activity was unclear. TERT provides the reverse transcriptase enzyme activity of telomerase — it synthesizes DNA at the ends of chromosomes, transcribing from an RNA template provided by the telom- erase RNA component, TERC. Interestingly, the effects of TERT on hair-follicle growth are independent of its reverse transcriptase activity and of TERC, suggesting that, in this context, TERT has an atypical function5,7. Park et al.1 provide a molecular basis for these unexpected observations. They purified TERT protein complexes from mammalian cells to identify any novel components, and were surprised to discover that these com- plexes contained BRG1. BRG1 is a subunit of a complex of proteins that alters the confor- mation of chromatin to facilitate transcrip- tion. β-catenin is known to bind directly to BRG1, resulting in enhanced expression of β-catenin target genes8. Thus, the existence of TERT–BRG1 complexes provided a possible molecular link between TERT and β-catenin. Subsequent experiments showed that TERT interacts directly with BRG1, and that com- plexes of TERT, β-catenin and a TCF protein bind to β-catenin target genes in cells from mouse small intestine. The authors1 found that, in several cell types, TERT is required for expression of Wnt-reg- ulated genes. In mouse embryonic stem (ES) cells, deletion of TERT reduces expression of Wnt target genes. This inhibition is overcome by the addition of enzymatically inactive TERT, indicating that, similarly to its effects on hair growth, the effect of TERT on Wnt target genes in ES cells is independent of telomerase’s reverse transcriptase activity. Strikingly, depletion of TERT in embryos of the frog Xenopus laevis produces devel- opmental defects similar to those seen in mouse embryos that lack β-catenin9. Excess Wnt signalling in X. laevis embryos causes duplication of the embryo’s anterior–posterior axis, resulting in the development of two-headed tadpoles10. Park and colleagues1 discovered that TERT overexpression in X. laevis embryos synergizes with β-catenin to pro- mote expression of Wnt reporter genes and axis duplication. Taken together, these findings provide convincing evidence for TERT as a key component of β-catenin tran- scriptional complexes in various contexts. Telomerase activity is particularly important in stem cells and other progenitor cells to maintain their extensive proliferative capacity and to prevent cellular senescence — a form of cell-cycle arrest that can be triggered by shortened telomeres11,12. CELL BIOLOGY The not-so-odd couple Sarah E. Millar Actively dividing cells do so at a risk — with each division, chromosome ends tend to shorten. Pairing proteins that promote cell division with a chromosome-end repair factor is a smart way to solve this problem. Embryonic development and homeostasis of adult tissues are regulated by a relatively small number of signalling pathways with astound- ingly diverse functions. These include con- trolling the rate of cell division, regulating the differentiation of cells into organs with com- plex structures, and activating adult stem cells. The functional complexity of signalling path- ways is achieved in part by the inter action of proteins in specific cell types with core compo- nents of signalling pathways, which modulates pathway activity and confers cell-type-specific functions. A study by Park et al.1 on page 66 of this issue identifies one such protein that functions in tissue-progenitor cells to increase the transcription of genes activated by the Wnt–β-catenin signalling pathway. Unexpectedly, this protein turns out to be an essential component of tel- omerase, a protein–RNA complex that has an apparently unrelated role in protecting the ends of chromo- somes (telomeres) from shortening during DNA replication2. Park et al. propose an intriguing functional connection. The Wnt–β-catenin signalling path- way stimulates proliferation of embry- onic progenitor cells and adult stem cells in self-renewing tissues such as the intestine, the haematopoietic sys- tem and hair follicles3. Wnt proteins bind to membrane-bound Frizzled receptors and LRP co-receptors, and this binding prevents degradation of cytoplasmic β-catenin. β-catenin translocates to the nucleus, where it activates target genes by binding to LEF/TCF transcription factors3. The first inklings of a link between β-catenin and telomerase came from studies of adult stem cells in the hair follicle. Throughout adult life, hair follicles undergo cycles of growth and regression that are dependent on stem cells located in a region of the follicle known as the bulge. Expression of stable, active β-catenin protein in skin epithelial cells causes proliferation of bulge stem cells and initiation of a new phase of hair growth4. Previous work5 had created mice in which extra copies of the gene encoding TERT, the protein component of telomerase, can be switched on in adult life in skin epithelial cells. Surprisingly, this study5 revealed that extra TERT mimics the prolif- erative and hair-growth-promoting effects of β-catenin. Another group6, working inde- pendently, found that continuous expression of TERT in skin epithelial cells enhances stem- cell proliferation in response to hair plucking or topical treatment with a tumour-promoting chemical. Subsequent experiments in hair TCF TERT C-catenin TERC TERT Telomere extension TERT Wnt target gene Prevention of senescence Progenitor-cell proliferation Embryonic development Tissue renewal Telomerase BRG1 a b Transcription Figure 1 | The TERT–β-catenin connection. a, The telomerase complex functions in progenitor cells to repair chromosome ends, known as telomeres, during cell division. TERT provides reverse transcriptase activity to the complex, and uses TERC, the RNA component of telomerase, as a template. b, Park et al.1 find that TERT also increases the transcriptional activity of β-catenin/TCF complexes through interaction with BRG1, a factor that binds the Wnt signalling molecule β-catenin and alters the conformation of chromatin. These two separate functions of TERT may simultaneously prevent cellular senescence and increase proliferation of progenitor cells, permitting embryonic development and renewal of adult tissues. 4. Koppens, F. H. L. et al. Nature 442, 766–771 (2006). 5. Zumofen, G., Mojarad, N. M., Sandoghdar, V. & Agio, M. Phys. Rev. Lett. 101, 180404 (2008). 6. Chang, D. E., SØrensen, A. S., Demler, E. A. & Lukin, M. D. Nature Phys. 3, 807–812 (2007). 7. Duan, L.-M. & Kimble, H. J. Phys. Rev. Lett. 92, 127902 (2004). 8. Merlein, J. et al. Nature Photon. 2, 230–233 (2008). 9. Lounis, B. & Orrit, M. Rep. Prog. Phys. 68, 1129–1179 (2005). 44 NATURE|Vol 460|2 July 2009NEWS & VIEWS 39-47 News and Views MH IF.indd 4439-47 News and Views MH IF.indd 44 26/6/09 17:26:1526/6/09 17:26:15 © 2009 Macmillan Publishers Limited. All rights reserved /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.00000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 2400 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck true /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox false /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description