special feature | COMMENTARY method of the year

Optogenetics

Optogenetics is a technology that allows targeted, fast control of precisely defined events in biological systems as complex as freely moving mammals. By delivering optical control at the speed (millisecond- scale) and with the precision (cell type–specific) required for biological processing, optogenetic approaches have opened new landscapes for the study of biology, both in health and disease.

Optogenetics1 is the combination of genetic ingful only in the context of other events to fragile mammalian , with photo- and optical methods to achieve gain or loss occurring in the rest of the tissue, the organ- currents that would also likely be too slow of function of well-defined events in specific ism and the environment as a whole. and weak to be useful. Moreover, scientists cells of living tissue. In the broadest sense2, did not believe that this approach would optogenetics encompasses a core technol- Single-component optogenetics lead to the long-sought single-component ogy—targetable control tools that respond In 1979 Francis Crick suggested that the strategy (a perception that had also slowed to light and deliver effector function—and major challenge facing was the development and application of green enabling technologies for (i) delivering light the need to control one type of cell in the fluorescent protein), as microbial into tissues under investigation, (ii) target- while leaving others unaltered. As require a chemical cofactor, all-trans , ing the control tools to cells of interest and electrodes cannot be used to precisely tar- to absorb photons. (iii) obtaining compatible readouts and per- get defined cells and drugs act much too An August 2005 report, however, forming analysis, such as targeted imaging slowly, Crick later speculated that light described that upon introduction of a or electrical recording of evoked activity. might have the properties to serve as a microbial gene without any other Certain elements have been known to control tool, but at the time neuroscientists parts, chemicals or components, neurons exist in earlier forms and in other contexts, knew of no clear strategy to make specific became precisely responsive to light4. though not conceptualized or developed as cells responsive to light. Several additional reports followed over a control technology, as far back3 as 1971, Yet 40 years ago microbial biologists knew the next year, and by 2010 channelrhodop-

Nature America, Inc. All rights reserved. All rights Inc. America, 1 Nature © 201 with their fundamental transition to the that some microorganisms produce visible sin, bacteriorhodopsin and emergence of optogenetics beginning in light–gated proteins that directly regulate the all had proved capable of turning neurons 2005 (Fig. 1) triggered by the demonstra- flow of ions across the plasma membrane. In on or off, rapidly and safely, in response to tion of single-component control tools in 1971, Stoeckenius and Oesterhelt discovered diverse colors of light (Fig. 2). Vertebrate neuroscience: microbial opsin genes that that bacteriorhodopsin acts as an ion pump tissues contain natural all-trans retinal— could safely confer to neurons both light- that can be rapidly activated by visible-light the cofactor essential for photonic control of detection capability and defined high-speed photons3. And this original theme from microbial opsins—and as a result research- effector function in a single readily targeta- 1971 of single-gene, single-component ers showed that optogenetic control was ble module4. control continued with the later identifica- feasible even in intact mammalian brain Although it arose from neuroscience, tion of other members of this family: halo­ tissue2 and freely moving mammals9,10. In optogenetics addresses a much broader in 1977 by Matsuno-Yagi and a fundamental shift from earlier approaches, unmet need in the study of biological sys- Mukohata5, and in 2002 microbial opsin genes therefore provided a tems: the need to control defined events by Hegemann, Nagel and their colleagues6. single-component strategy. in defined cell types at defined times in But it took more than 30 years for neu- Optogenetic tools have now changed intact systems. Such analyses are important roscientists to bring the two fields together the way neuroscience is conducted owing because cellular events are typically mean- because such an approach was thought to to a convergence over the past two years be very unlikely to work. Instead, scientists of the intrinsic tractability of the single- 7,8 Karl Deisseroth is at the Howard Hughes Medical considered multicomponent strategies component tools with rapid advances in the Institute, Department of Bioengineering and that involved no microbial opsin genes at associated enabling technologies. Obtaining Department of Psychiatry and Behavioral Sciences, all but rather cascades of different genes or precise causal control in intact systems as Stanford University, Stanford, California, USA. e-mail: [email protected] combinations of custom-synthesized chem- complex as behaving mammals is certainly PUBLISHED ONLINE 20 DECEMBER 2010; icals and genes. It was also likely that these important in neuroscience, just as in other DOI: 10.1038/NMETH.F.324 foreign membrane proteins would be toxic fields of biology, but historically, it has not

26 | VOL.8 NO.1 | JANUARY 2011 | nature methods COMMENTARY | special feature

Going forward, applications of these and Opsin engineering and genomic expansion; generalizable opsin targeting other tools to tissues and systems through- strategies; brain-disease and neuroscience out biology will become a major theme. 75 applications; stem cells, heart cells, 70 muscle cells and human neurons studied Such tools have already been applied to 65 60 nonneuronal systems, including glial, mus- 55 Bacteriorhodopsin described cle, cardiac and embryonic stem cells, and 50 as a single-component Fiberoptic interface (May 2007) and 45 light-activated regulator of single-component control of freely moving the temporal precision provided by light 40 transmembrane ion flow mammals (October 2007) described may continue to be crucial in this regard by 35 Microbial opsins first brought 30 to neurons (August 2005) providing a defined event in a defined cell 25 Halorhodopsin described population at specific times relative to envi- 20 Channelrhodopsin described 15 ronmental events. 10 5 Different speeds of operation are required Number of PubMed entries 0 for different cells and tissues; for example, cardiac optogenetics may not always 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 require the millisecond-scale precision of Year tools used in fast-spiking central nervous system neurons because the heart in some Figure 1 | Graphical illustration of ‘optogenetics’ emerging in the scientific literature. Demonstration of single-component optogenetic control of neurons with microbial opsins4 was followed by ways operates on slower timescales, though corresponding optogenetic terminology2 in October 2006, and corresponding optogenetic control of faster experimental methods may unlock freely moving mammals using microbial opsins and the fiberoptic neural interface9,10. Also marked are preciously unanticipated regimes of fast identifications of bacteriorhodopsin3, halorhodopsin5 and channelrhodopsin6, all of which were much tissue computation and adaptation. And later (2005–2010) shown to function as fast, single-component optogenetic tools in neurons. Numbers even the microbial opsins can act by bio- indicate only publications searchable by ‘optogenetics’ or derivatives thereof on 1 December 2010. chemical rules as well; the trace Ca2+ flux of channelrhodopsin has been used to drive been possible to deliver causal, temporally described optical control of distinct G pro- activation of nonexcitable glial cells19 in the precise gain or loss of function in one type tein–coupled receptor (GPCR) biochemi- central nervous system and could be used to of brain cell or in a defined projection from cal pathways in freely moving mammals drive this ubiquitous second messenger in one brain region to another. using vertebrate rhodopsin-GPCR chi- a rich array of Ca2+-modulated cells rang- Now, in addition to conferring temporal meras (optoXRs), which recruit pathways ing from insulin-secreting pancreatic beta precision and applicability to freely moving that are governed by distinct heterotri- cells to T lymphocytes, in intact endocrine- 14 mammals, the single-component character meric G-protein pathways (Gs and Gq) . exocrine and immunologic tissues. In stem of the microbial opsin system has enabled Subsequently optical control over small cell biology and engineering, biochemical or generalizable targeting. Latest-generation GTPases (with resulting changes in cellular electrical drive can now be delivered inde- Cre recombinase–dependent opsin-express- shape and motility) was achieved in cul- pendently and selectively to niche cells, stem ing viruses have dovetailed with the exten- tured cells using novel strategies from sev- cells or stem cell progeny, even with intact sive and growing resource of mouse lines eral different laboratories15,16. The GTPases tissue and animals.

Nature America, Inc. All rights reserved. All rights Inc. America, 1 Nature © 201 selectively expressing Cre recombinase in were activated by recruitment to the plasma In general, optogenetic tools can now be defined cell types; optogenetic control can membrane via optically modulated protein- selected that are appropriate for the target now be delivered to defined cells in freely protein interactions, a method that may tissue of choice, with regard to electrical or moving mice with substantial versatility11. ultimately become generally suitable for biochemical effector function, speed and Likewise, a strategy of illuminating axons controlling additional classes of biochemi- other properties. far from the opsin-transduced cell body cal signal transduction (particularly if the (capitalizing on trafficking of light-sensitive method can be shown to operate in single- Future directions: expanding the toolkit molecules down the axon itself) enables component fashion in intact mammalian Initially researchers noted that the fidelity of projection-based control of cells12,13 with- tissues, as seems likely in cases in which optogenetic control via opsin gene expres- out requiring any genetic knowledge at all, relevant flavin or biliverdin chromophores sion was not optimal, with noise in the which is important for versatile optogenetic are present). In this issue, Lim and col- system characterized by occasional extra control in less genetically tractable species leagues17 discuss strategies for development action potentials or missing action poten- such as rats and primates. These and other of biochemical optogenetic control. Finally, tials4. Subsequent optimization of microbial general-purpose targeting strategies heavily microbial adenylyl cyclases with low activity opsins addressed this problem for optoge- depend on the single-component property. in the dark have been described that repre- netics, returning conceptually to the 1971 sent a marked advance over earlier microbi- discovery of bacteriorhodopsin and building Future directions: beyond neuroscience al cyclases from the optogenetic perspective, upon mutagenesis of the bacteriorhodopsin Molecular engineering has also enabled use a flavin chromophore and, as optoXRs, gene in many laboratories over decades. As optogenetic control of well-defined bio- are suitable for single-component optical a result, modified opsins for optogenet- chemical events. Early in 2009, capitalizing control in neurons18. Together, these experi- ics now include fast and slow mutants that on the retinoid content of mammalian brain ments opened the door to optogenetics in variously enable high-fidelity control over tissue and the low activity of retinal-based essentially every cell and tissue whether high-frequency trains20, signaling modules in the dark, researchers electrically excitable or not. bistable changes in excitability21 and orders-

nature methods | VOL.8 NO.1 | JANUARY 2011 | 27 special feature | COMMENTARY

of-magnitude increased light sensitivity21. Electrical stimulation Optogenetic excitation Optogenetic inhibition Additionally, stronger opsin expression and other improved properties have resulted from creating chimeras between different opsins in combination with mutagenesis22. In this issue, Hegemann and Möglich23 discuss the interface between bacteriorho- dopsin mutagenesis and the optimization of

optogenetic tools. Corral Marina Going forward, molecular tools will rap- idly prove important for optogenetics in Figure 2 | Principle of optogenetics in neuroscience. Targeted excitation (as with a blue light–activated several other ways as well, beyond targeted channelrhodopsin) or inhibition (as with a light–activated halorhodopsin), conferring cellular mutagenesis of known genes. Certainly specificity and even projection specificity not feasible with electrodes while maintaining high temporal subcellular targeting of optogenetic tools (action-potential scale) precision. is of great interest, and a new frontier con- sists of delivering optical control (whether never fully possible with simultaneous elec- as brain slices or surface brain structures2,12. biochemical or electrical) to well-defined trical stimulation at the same site owing to Although in some ways the orthogonality subcellular domains or intracellular (for electrical artifacts that can now be avoided of electrical and optical methods suggests example, membranous) compartments. with optical stimulation. As simultaneous a special value for the mixed-modality, Screens for optical tools that modulate readout measures for optogenetically con- optrode-style approach, readouts may also protein-protein interactions may open trolled systems become more rich and be achieved with genetically encoded opti- the door to optogenetic control of kinases complex, the concept of ‘reverse engineer- cal measures of activity such as genetically and transcription factors. And molecular ing’ of biology will be taken further. This encoded Ca2+ indicators and voltage sen- engineering will deliver optogenetic tools will allow us to infer computational roles of sors. Certainly all-optical interrogation of with altered chromophore dependence (for biological tissues, based on how they trans- neural circuits has already been carried out example, enabling new uses of endogenous form the information we provide and how using Ca2+ dyes or voltage-sensitive dyes chromophores such as biliverdin or flavin) these transformations are altered in com- for output that are spectrally compatible as well as altered effector function. plex disease states (in much the same way with microbial opsins for input26. In recent Moroever, rapidly accelerating molecular that reverse engineering is carried out on years, researchers have made great strides in genomics efforts will continue to expand computer chips to determine the underly- achieving specific genetically encoded read- the optogenetics toolkit—which now ranges ing processing). outs, which Peron and Svoboda27 disscuss across and beyond the visible spectrum—a Novel devices and systems are required in this issue, along with complementary process that began with the discovery of a to advance this vision. Beginning in 2007, light-input strategies for superficial neural red-shifted channelrhodopsin24 for com- soon after fiberoptic and laser-diode tools9 structures. binatorial control in 2008. Although most enabled optogenetic control even deep in microbial opsin genes do not express well in the of freely moving mammals10, Conclusion 25,26 Nature America, Inc. All rights reserved. All rights Inc. America, 1 Nature © 201 mammalian neurons , it has been found closely related hybrids of fiberoptics and As discussed here, efforts to expand the that the major underlying problem is one of electrodes13 (‘optrodes’) allowed high-speed capabilities of microbial-opsin optogenet- membrane trafficking25. This cell-biological simultaneous readouts that kept pace with ics since 2005 have spanned genomic tool concept led to the identification of mem- the high-speed inputs of optogenetics19. discovery, molecular engineering, opsin brane trafficking motifs that, when added A major area of future work will be targeting and optical-device develop- to specific locations on microbial opsin expansion of the capabilities of these mixed- ment. The importance of optogenetics as genes, confer robust expression and opto- modality devices with regard to (i) output a research tool continues to grow rapidly, genetic control to opsins that are otherwise channel number and type, (ii) smooth and it is now used in more than 800 labo- problematic to express or express poorly, temporal and spatial integration with ratories around the world. In this context, including the original microbial opsin gene, increasingly complex optical input chan- it is intriguing to note that a membrane bacteriorhodopsin25. These molecular prin- nel number and type, and (iii) closed-loop trafficking–enhanced microbial opsin (in ciples will provide a wealth of diverse light control. (Whereas many studies have been this case, a halorhodopsin25,26, eNpHR2.0) sensitivity and effector function properties, published describing light-triggered behav- has recently been delivered to living unlocking the potential of thousands of ior or light-triggered physiology, only a few human neural tissue (the ex vivo retina) microbial opsin genes that occur through- have emerged on behavior-triggered light with potent optogenetic functionality28. out the major kingdoms of life. or physiology-triggered light, and system- Yet the most fundamental impact of atically closing the control loop will enable optogenetics, even on human health, does Future directions: reverse engineering real-time bidirectional communication not arise from direct introduction of opsins As another distinct advantage, fast and between input and output streams.) into human tissue but rather from use as a specific optical control opens a new land- Cell-type specificity is still not readily research tool to obtain insights into complex scape for systems physiology by permitting enabled for electrical recording, with the tissue function, as has already been the case simultaneous input-output interrogation exception of targeted microelectrode- or for Parkinson’s disease19. Owing to techno- of excitable tissue. Electrical recording was patch clamp–accessible preparations such logical limitations in probing intact neural

28 | VOL.8 NO.1 | JANUARY 2011 | nature methods COMMENTARY | special feature

circuits with cellular precision, our current divisions, but the broad mesoscale of cellular 9. Aravanis, A.M. et al. J. Neural Eng. 4, S143–S156 understanding of brain disorders does not connections in intact neuronal circuitry was (2007). 10. Adamantidis, A.R., Zhang, F., Aravanis, A.M., do full justice to the brain as a high-speed largely inaccessible until Ramón y Cajal and Deisseroth, K. & de Lecea, L. Nature 450, 420– cellular circuit. Rather than conceptualizing his students and colleagues used the Golgi 424 (2007). the brain as a mix of , ide- technology to systematically map local cir- 11. Tsai, H.C. et al. Science 324, 1080–1084 (2009). 12. Petreanu, L., Huber, D., Sobczyk, A. & Svoboda, ally we would be able to move toward a cir- cuit relationships with cellular precision yet K. Nat. Neurosci. 10, 663–668 (2007). cuit-engineering approach, in which devas- still within the intact system. Inferences from 13. Gradinaru, V. et al. J. Neurosci. 27, 14231–14238 tating symptoms of disease are understood even these descriptive anatomical meth- (2007). 14. Airan, R.D., Thompson, K.R., Fenno, L.E., to causally result from specific spatiotempo- ods still reverberate through neuroscience. Bernstein, H. & Deisseroth, K. Nature 458, ral patterns of aberrant circuit activity relat- Optogenetics has targeted the analogous 1025–1029 (2009). ing to specific neuronal populations. But need for causal control of defined small-scale 15. Levskaya, A., Weiner, O.D., Lim, W.A. & Voigt, C.A. Nature 461, 997–1001 (2009). technology has been lacking for the requisite events occurring in specified cellular popu- 16. Wu, Y.I. et al. Nature 461, 104–108 (2009). high-speed, targeted, causal control of intact lations while these populations still remain 17. Toettcher, J.E., Voigt, C.A., Weiner, O.D. & neural circuit function, and this challenge embedded and functioning within larger Lim, W.A. Nat. Methods 8, 35–38 (2011). extends to basic neuroscience and other intact-tissue systems, at appropriate spatial 18. Stierl, M. et al. J. Biol. Chem. published online 28 October 2010 (doi:10.1074/jbc. biological systems as well. Although opto- and temporal resolution and under normal M110.185496). meets this challenge, much work or pathological conditions. 19. Gradinaru, V., Mogri, M., Thompson, K.R., remains, including advancing the genomic Henderson, J.M. & Deisseroth, K. Science 324, COMPETING financial INTERESTS 354–359 (2009). expansion of optogenetic tools, refining the The author declares no competing financial interests. 20. Gunaydin, L.A. et al. Nat. Neurosci. 13, 387–392 molecular engineering for optimized func- (2010). tionality and developing light and genetic 1. Deisseroth, K. Sci. Am. 303, 48–55 (2010). 21. Berndt, A., Yizhar, O., Gunaydin, L.A., 2. Deisseroth, K. et al. J. Neurosci. 26, 10380–10386 Hegemann, P. & Deisseroth, K. Nat. Neurosci. 12, targeting strategies for various biological (2006). 229–234 (2009). systems and animal models. 3. Oesterhelt, D. & Stoeckenius, W. Nat. New Biol. 22. Lin, J.Y., Lin, M.Z., Steinbach, P. & Tsien, R.Y. The challenges faced today in the study of 233, 149–152 (1971). Biophys. J. 96, 1803–1814 (2009). 4. Boyden, E.S., Zhang, F., Bamberg, E., Nagel, G. 23. Hegemann, P. & Möglich, A. Nat. Methods 8, diverse intact biological systems conceptually & Deisseroth, K. Nat. Neurosci. 8, 1263–1268 39–42 (2011). parallel the challenge faced by (2005). 24. Zhang, F. et al. Nat. Neurosci. 11, 631–633 more than a hundred years ago, with the com- 5. Matsuno-Yagi, A. & Mukohata, Y. Biochem. (2008). mon theme being the need to link informa- Biophys. Res. Commun. 78, 237–243 (1977). 25. Gradinaru, V. et al. Cell 141, 154–165 (2010). 6. Nagel, G. et al. Science 296, 2395–2398 (2002). 26. Zhang, F. et al. Nature 446, 633–639 (2007). tion across spatial scales. At that time in his- 7. Zemelman, B.V., Lee, G.A., Ng, M. & Miesenbock, 27. Peron S. & Svoboda, K. Nat. Methods 8, 30–34 tory, microscopy had defined small cellular G. 33, 15–22 (2002). (2011). elements of the brain and large architectonic 8. Banghart, M., Borges, K., Isacoff, E., Trauner, D. & 28. Busskamp, V. et al. Science 329, 413–417 Kramer, R.H. Nat. Neurosci. 7, 1381–1386 (2004). (2010). Nature America, Inc. All rights reserved. All rights Inc. America, 1 Nature © 201

nature methods | VOL.8 NO.1 | JANUARY 2011 | 29