Optogenetic manipulation of the neuronal activity enables one to analyze the neuronal network both in vivo and in vitro with precise spatio-temporal resolution. Channelrhodopsins (ChRs) are light-sensitive cation channels that depolarize the cell membrane, whereas halorhodopsins and archaerhodopsins are light-sensitive Cl− and H+ transporters, respectively, that hyperpolarize it when exogenously expressed. The cause-effect relationship between a neuron and its function in the brain is thus bi-directionally investigated with evidence of necessity and sufficiency. In this review we discuss the potential of optogenetics with a focus on three major requirements for its application: (i) selection of the light-sensitive proteins optimal for optogenetic investigation, (ii) targeted expression of these selected proteins in a specific group of neurons, and (iii) targeted irradiation with high spatiotemporal resolution. We also discuss recent progress in the application of optogenetics to studies of non-neural cells such as glial cells, cardiac and skeletal myocytes. In combination with stem cell technology, optogenetics may be key to successful research using embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) derived from human patients through optical regulation of differentiation-maturation, through optical manipulation of tissue transplants and, furthermore, through facilitating survival and integration of transplants.
Vertebrate and most invertebrate brains consist of huge number of neurons that are connected to each other to make a complex network. For example, in a human brain, there are 1010–12 neurons, each of which receives 102–3 synapses. The idea that this neuronal network generates brain function, the mind, via signal communication was first proposed over 100 years ago by a Spanish neuroanatomist, Santiago Ramón y Cajal (Cajal 1894). To this day our knowledge about the neuronal network, its organization and function, is still limited despite extensive research over the past 100 years. Furthermore, how this network is organized and becomes functional during the development of an animal remains to be elucidated.
The function of the neural network has been studied as a cause-effect relationship of stimulation and response (O'Connor et al. 2009). With the twentieth-century technological development in electronics, electrophysiology has long been one of the principal methods for the study of neurons and neural networks. That is, the neural network is electrically stimulated and the effects are electrically recorded. For example, Penfield and his colleagues electrically stimulated various regions of the human brain and electrically recorded the resulting muscle contractions in their series of experiments during the 1930s (Penfield & Rasmussen 1950). With these results they precisely mapped regions of the cerebral cortex involved in movement. Nowadays, electrical stimulation is applied using field or intracellular/patch electrodes. Although electrical field stimulation is simple, convenient and has high temporal resolution, the electrical field is generally non-uniform and many untargeted neurons are stimulated simultaneously. It is thus difficult to identify which neurons are stimulated. On the other hand, a single, identified neuron can be selectively stimulated with an intracellular or whole-cell patch electrode. However, the number of simultaneous stimulations is spatially limited as each electrode is independently manipulated.
Optical stimulation methods have received much attention recently with the technological development of modern optics. They have advantages over conventional electrical stimulation methods: finer spatiotemporal resolution and parallel stimulations at multiple sites (Callaway & Yuste 2002; Miesenböck 2004). These methods are also less harmful and more convenient than electrical stimulation methods. Another breakthrough combined optical stimulation with genetic engineering technologies, which is otherwise known as optogenetics. Light-sensing proteins of various living organisms are now available to be exogenously expressed in neurons and other target cells both in vivo and in vitro. Cellular functions such as the membrane potential, can thus be manipulated by light. In this review, we will focus on the basic principles of optogentic manipulation of neural and non-neural tissues. Optical probing methods that use protein sensors have not been discussed in this review as these investigations have been reviewed elsewhere (Miesenböck & Kevrekidis 2005; Palmer & Tsien 2006; Mank & Griesbeck 2008; Newman et al. 2011; Peterka et al. 2011; Knöpfel 2012).
Fundamental molecular biology of optogenetics
The genes of two channelrhodopsins, channelrhodopsin-1 (ChR1) and channelrhodopsin-2 (ChR2), were first identified in the expressed sequence tag (EST) project of Chlamydomonas reinhardtii at the Kazusa DNA Research Institute, Japan (http://www.kazusa.or.jp/) by three independent groups (Nagel et al. 2002, 2003; Sineshchekov et al. 2002; Suzuki et al. 2003). The two papers by Nagel et al. (2002, 2003) were remarkable in that they identified both ChR1 and 2 as ion channels directly gated by light using Xenopus oocyte expression system. The existence of these proteins was first proposed following the results from electrical measurements taken from the intact alga (Harz & Hegemann 1991; Braun & Hegemann 1999). Sineshchekov et al. (2002) revealed that both ChR1 and ChR2 (Chlamydomonas sensory rhodopsins A and B [CSRA and B] in the original paper) mediate the light-dependent behavior of alga. Suzuki et al. (2003) showed that ChR1 apoproteins (archaeal-type Chlamydomonas opsin-1 [Acop-1] in the original paper) are localized in small regions of the plasma membrane covering the eyespot or stigma, where photoreceptors had been thought to be concentrated (Melkonian & Robenek 1980; Kateriya et al. 2004). ChR homologues have also been identified in other species (Zhang et al. 2011), including, Volvox carteri (VChR1 and VChR2), Chlamydomonas augustae (CaChR1), Chlamydomonas yellowstonensis (CyChR1), Chlamydomonas raudensis (CraChR2), Mesostigma viride (MChR1), Dunaliella salina (DChR), and the number of reported species homologues continues to increase (Ernst et al. 2008; Zhang et al. 2008; Kianianmomeni et al. 2009; Govorunova et al. 2011; Hou et al. 2012; Watanabe et al. 2012). Each ChR is a member of the microbial-type (archaeal-type, type I) rhodopsin family with a core structure of about 300 amino acids. The core structure consists of seven transmembrane helices (TM1-7) and a retinal that covalently binds to a consensus Lys residue at the middle of TM7 (Fig. 1A,B). Light absorption is followed by the photoisomerization of an all-trans retinal to a 13-cis configuration. This conformational change allows the channel structure to become permeable to cations, such as Na+, K+, Ca2+ and H+, (Fig. 1C, Bamann et al. 2008; Ernst et al. 2008; Stehfest & Hegemann 2010). This enables very rapid (within milliseconds) generation of a photocurrent across cell membranes expressing ChRs (Nagel et al. 2002, 2003; Boyden et al. 2005; Ishizuka et al. 2006). Despite extensive studies, researchers have yet to describe how the cations flow through the molecular structure (Sugiyama et al. 2009; Ruffert et al. 2011; Kato et al. 2012; Tanimoto et al. 2013).
When ChR2 is exogenously expressed in neurons, blue light irradiation evokes an inward current with membrane depolarization, which opens voltage-gated sodium channels and calcium channels, and generates an action potential (Boyden et al. 2005; Li et al. 2005; Ishizuka et al. 2006). Neuronal activity can also be negatively regulated by light if the neurons are engineered to express either a Cl−-transporting rhodopsin, such as, NpHR from Natronomonas pharaonis or an H+-transporting rhodopsin, such as, archaerhodopsin-3 (Arch) from Halorubrum sodomense and archaerhodopsin-T (ArchT) from Halorubrum strain TP009 (Han & Boyden 2007; Zhang et al. 2007a, 2011; Chow et al. 2010; Han et al. 2011). The cause-effect relationship between a neuron and its function in the brain (e.g., behavior) is bi-directionally investigated: the necessity through light-induced silencing of the neuron with hyperpolarizing rhodopsins and the sufficiency through light-induced activation with depolarizing rhodopsins.
To use these optogenetic molecules for neurobiological experiments, we have to meet at least three requirements: (i) selection of the light-sensitive proteins optimal for optogenetic investigation, (ii) targeted expression of the above proteins in a specific group of neurons, and (iii) targeted irradiation with high spatio-temporal resolution.
Although ChR2 has been widely used to photostimulate neurons, recent technological developments will help us exploit the full potential of light-gated ion channels.
The peak absorption of ChR2 is at 460–470 nm, the wavelength preferentially absorbed by live tissue (Yaroslavsky et al. 2002; Aravanis et al. 2007). Although some native and altered ChRs absorb green (Yizhar et al. 2011a; Mattis et al. 2012), the more red-shifted ChRs absorbing red or even near infra-red light would be more desirable for relatively deep tissue penetration and/or even irradiation. Parallel irradiation may also be facilitated by using ChRs with various wavelength sensitivities in combination with multi-colored optics. Experimentally, ChR-dependent photostimulation may be used in combination with fluorescent probes of calcium, membrane potential or other cellular functions. For example, the excitation spectrum of Fluo-3, one of the most popular fluorescent Ca2+ indicators, overlaps with the absorption spectrum of ChR2. Therefore, it would be difficult to measure Fluo-3 fluorescence during ChR2-dependent photostimulation. The irradiation used for the measurement of Fluo-3 fluorescence would evoke the ChR2 photocurrent, which should inevitably change the intracellular Ca2+ concentration. On the other hand, the wavelengths 340 and 380 nm, which are optimal for the use of fura-2, would minimally evoke the ChR2 photocurrent. Recently, various genetically encoded fluorescent reporters/sensors have become available to probe living cells at the functional level (Zhao et al. 2011b; Knöpfel 2012). ChRs with various absorption spectra would facilitate these studies.
Photocurrents may be enhanced by facilitating protein folding and membrane expression. Mis-folded or mis-directed molecules are most likely toxic to the cell through endoplasmic reticulum (ER) stress (Ron & Walter 2007; Kim et al. 2008). This becomes apparent when the protein synthesis is accelerated using powerful promoters such as those derived from viruses.
During blue light irradiation, the ChR2 photocurrent peaks almost instantaneously, but desensitizes rapidly to a steady state within several tens of milliseconds (Fig. 2A). It takes several tens of seconds for full recovery from desensitization. The prominent desensitization of the ChR2 photocurrent limits its application for repetitive stimulation at high frequency. Both the peak and steady-state photocurrents have their ceilings, even with enhanced light (Fig. 2B) since desensitization and its rate are enhanced with an increase in light power density (Fig. 2C).
The turning on/off rate of the photocurrent should be adjusted to the neuron and the stimulation frequency. In the central nervous system (CNS), some neurons tend to fire at high frequency. The relatively slow kinetics of ChR2 (τOFF = 10–20 ms) is inadequate to drive these neurons, particularly at high frequency. Prolonged depolarization often induces multiple spikes even with irradiation of short duration. ChR variants with small τOFF (Lin et al. 2009; Wang et al. 2009a; Gunaydin et al. 2010; Wen et al. 2010; Berndt et al. 2011) would solve these problems. On the other hand, some variants of ChRs, such as ChR2(C128S), ChR2(D156A) and ChR2(C128S/D156A), have very slow deactivation kinetics and the photocurrent can be terminated with different colors of light (Berndt et al. 2009; Bamann et al. 2010; Yizhar et al. 2011b; Fig. 1C). These step-function opsins (SFOs) and stable SFOs (SSFOs) are over two-orders more sensitive to dim light and are suitable for stimulating relatively deep neurons.
ChRs with variable ion selectivity would enable broader applications. Although the membrane potential is shifted to the negative direction in a light-dependent manner by the use of a Cl−-transporting or an H+-transporting protein, such as NpHR and Arch, its efficiency of this model is limited because only one ion is transported across the membrane with the absorption of a single photon. However, if ChRs were designed to be selectively permeable to Cl− or K+, these ChRs would allow the bulk of ions to flow with the absorption of a single photon. Therefore, their hyperpolarizing effects would be expected to be much larger than those of Cl−/H+ transporters. ChRs that are not permeable to Ca2+ are suitable to examine the effects of depolarization per se. On the other hand, those selective for Ca2+ (Kleinlogel et al. 2011a; Prigge et al. 2012) could be useful to manipulate the intracellular Ca2+.
Some optimal optogenetic molecules could be obtained by genome mining (Zhang et al. 2011). The absorption/action spectra were red-shifted in one of the ChRs from Volvox carteri (VChR1) and in one from Mesostigma viride (MChR1) (λmax = 520–540 nm). Variants of ChRs with various properties were also generated by targeted mutagenesis, by transmembrane helix shuffling, or combinations thereof (Table 1). Among them, those including TM1-2 of ChR1, generally showed reduced desensitization and enhanced photocurrents with improved folding/membrane expression (Lin et al. 2009; Wang et al. 2009a; Lin 2010; Mattis et al. 2012; Prigge et al. 2012). On the other hand, the photocurrent retardation of ChR1 was overcome by exchanging TM6 with its counterpart in ChR2 (Wen et al. 2010). Thus designed channelrhodopsin-green receiver (ChRGR) showed relatively large photocurrents with red-shifted spectral sensitivity (identical to ChR1), small desensitization and rapid on/off kinetics. With these advantages, the use of ChRGR would enable one to inject a current into a neuron with a time course that could be predicted by the intensity of light (opto-current clamp). Structural data, such as that provided by X-ray crystallography, facilitates the design of ChR variants with the desired spectral sensitivity, ion permeability and so on (Kato et al. 2012).
Various methods are now available for the targeted expression of exogenous molecules and have been extensively reviewed (Yizhar et al. 2011a). For example, virus vectors derived from various serotypes of adeno-associated virus (AAV), human HIV virus (lentivirus) and Sindbis virus have been made in many laboratories and are distributed worldwide. Alternatively, various electroporation methods have been devised. For example, retinal ON bipolar cells were specifically targeted by introducing ChR2 gene connected to the mGluR6 promoter sequence through electroporation (Lagali et al. 2008). The cortical layer-specific expression of ChR2 was induced by timed in utero electroporation of mouse (Petreanu et al. 2007; Hull et al. 2009). In these experiments, the GABAergic interneurons were not usually transfected because they migrate tangentially from the medial ganglionic eminence (MGE; Nadarajah & Parnavelas 2002). ChRs can also be efficiently expressed in the spinal cord motorneurons of the chick embryo (Li et al. 2005; Kastanenka & Landmesser 2010; Sharp & Fromherz 2011) using in ovo electroporation technology (Odani et al. 2008). A small number of physiologically-identified neurons could also be targeted for the gene expression with a combination of imaging using single-cell electroporation methods (Kitamura et al. 2008; Uesaka et al. 2008; Steinmeyer & Yanik 2012).
Another typical strategy of targeted gene expression is the generation of transgenic animal models for experiments (Tables 2,3). Conditional expression systems, such as the Cre/loxP recombinase system, the tTA-tetO (Tet-On/Off) system and the Gal4/UAS system, are particularly promising. For example, in the case of the Cre/loxP mouse system, a driver mouse retains the cre-recombinase gene from the P2 bacteriophage regulated by a promoter that is predetermined to act in a particular cell type. When it is mated with mice of another reporter line retaining the gene of interest in loxP-flanked (“floxed”)-stop or floxed-inverse (FLEX) cassettes, the productive molecules are expressed in a particular group of cells in the pups harboring both the cre-recombinase gene and floxed gene of interest (Witten et al. 2011; Madisen et al. 2012). The tTA-tetO system has the advantage in that specific gene expression can be temporally regulated by the application of a chemical substance, doxycycline (Tanaka et al. 2012). Using this system, ChR2 was selectively expressed in recently activated neurons in the hippocampus (Liu et al. 2012). Nowadays, various driver mice have been generated and distributed from researchers and bio-resource facilities (Madisen et al. 2010; Yizhar et al. 2011a) and the number of reporter mice for optogenetics is increasing (Table 3).
Table 2. Transgenic animal lines for optogenetic manipulations
The Gal4/UAS system is popular for experiments using Drosophila and zebrafish. In these animals, very large numbers of various Gal4-expressing lines have been produced by chance using enhancer/gene trapping strategies, and then screened for useful phenotypes (Scott 2009). Similar variations of gene-expression patterns could be generated by a transgenic system of mammals using a Thy-1.2 genomic expression cassette that is dependent on differences in chromosomal integration sites and/or in the number of inserted copies (Arenkiel et al. 2007; Tomita et al. 2009; Ji et al. 2012). Much progress would be expected with the generation of reporter lines harboring ChR variants optimized for the various experimental designs. The Brainbow reporter system, in which the genes of interest are connected with loxP and its variants in tandem, may be used to express combinations of optogenetic proteins, which could vary between neurons (Livet et al. 2007).
Optimizing optical systems
The selection of light sources and light delivery methods has been extensively reviewed (Carter & de Lecea 2011; Yizhar et al. 2011a). Photocurrents generated by the activation of ChRs are dependent on light power density (LPD) directed on target cells (Ishizuka et al. 2006). Although both the peak and the steady-state photocurrents are almost always positively related to the LPD, those of many ChRs and their variants reach near the maximum with LPD as high as 10 mW/mm2 (Mattis et al. 2012). However, SFOs such as ChR2 (C128S) and ChR2 (D156A) were found to be more sensitive to light in the order of 102 or greater (Berndt et al. 2009; Bamann et al. 2010). This is because the population light sensitivity of the protein is dependent on the OFF kinetics as well as dependent on the protein's intrinsic light sensitivity (Sugiyama et al. 2009; Mattis et al. 2012). The SSFO mutant of ChR2, ChR2 (C128S/D156A) acts as a photon integrator because the peak photocurrent amplitude is dependent on total photon exposure during irradiation (Yizhar et al. 2011b). On the other hand, the H+ or Cl− transporters in this model were less sensitive to light, and completely effective hyperpolarization can only be induced with a LPD of >10 mW/mm2 (Han & Boyden 2007; Zhang et al. 2007a; Chow et al. 2010; Mattis et al. 2012). Using whole-cell patch clamping, the change in a neuronal membrane potential can be determined by providing a constant or modulated light. For example, the native response of cells (spontaneous oscillations in the membrane potential, spontaneous impulses, etc.) during given depolarization or hyperpolarization can be obtained with constant irradiation. Using various protocols, modulated light, such as square pulses, ramp pulses and sine waves, have been applied to manipulate the cell (Opto-current clamp, Yawo 2012). Opto-current-clamp experiments using Zap-function waves (swept-frequency oscillation) are important for elucidating the resonance frequency of a neuron (Gutfreund et al. 1995; Tohidi & Nadim 2009), and with this frequency, the firing pattern of a local network can be determined (Wen et al. 2010).
Previously, most irradiating devices have delivered monochromatic light to a single spot. However, it has become increasingly necessary to develop multicolored and spatiotemporally patterned irradiating devices. For example, one needs to deliver different wavelengths to activate one-by-one depolarizing opsins (e.g, ChR2) and hyperpolarizing opsins (e.g, NpHR; Han & Boyden 2007; Zhang et al. 2009; Chow et al. 2010; Gradinaru et al. 2010; Zorzos et al. 2010). Alternative irradiation of blue and yellow light is necessary to exploit the full potential of SFOs and SSFOs (Berndt et al. 2009; Yizhar et al. 2011b). If a neuron were designed to express fusion proteins of ChR and NpHR (Kleinlogel et al. 2011b), it could be manipulated in either the positive or negative direction, simply by switching the color of the light between blue and yellow.
In general, a single neuron receives many convergent inputs from various neurons, each of which fires with a different pattern, and outputs its activity pattern divergently as action potentials initiated usually at its initial segment of the axon. Therefore, to analyze the input-output relationship of a neuron, a network or even a brain, systematic experiments using parallel and spatiotemporally patterned stimulations are necessary. Optogenetics is considered to be the most suitable method to achieve this (Miesenböck 2004; Ishizuka et al. 2006). Previously, spatially patterned photostimulation methods using a laser scanning microscope have been used for the functional mapping of the cortex (Wang et al. 2007a; Hira et al. 2009). Laser irradiation was also spatiotemporally controlled using an acousto-optic device (Shoham et al. 2005; Wang et al. 2011). Otherwise, a specimen could be two-dimensionally targeted on a scanning stage under a single collimated laser beam (Ayling et al. 2009). In these studies, each targeted area was sequentially, but not simultaneously, illuminated with other areas. Recently, two-dimensional array light-emitting diodes (LEDs) have been designed for patterned photostimulation (Grossman et al. 2010). At present, the irradiance is uneven in any given field on the specimen because inter-LED space is present. Nevertheless, this method could become one of the most ideal tools given that LEDs have high temporal resolution, high power and multiple colors. Spatial light modulators based on a digital micro-mirror device (DMD) or liquid crystal display (LCD) have also been proposed as other ideal tools for patterned photostimulation (Stirman et al. 2012), since they are currently used for projecting multi-colored patterned images. Using DMD, 380 and 505 nm LED lights were alternately applied on a region of interest (ROI) to switch on and off the light-sensitive ionotropic glutamate receptor (LiGluR; Wang et al. 2007b). The ChR2-expressing neurons were spatially differentiated from other neurons in the circuit using DMD array in Caenorhabditis elegans (Guo et al. 2009) and in zebrafish (Zhu et al. 2012). A DMD-based commercial projector was used in the patterned activation of retinal ganglion cells that express ChR2 (Farah et al. 2007) and mouse olfactory bulb in which ChR2 is expressed in the glomeruli (Dhawale et al. 2010). By using DMD, these studies could use patterned photostimulation either temporally or spatially, but not both. Recently, the image projector was applied to deliver multicolored light with a pre-programmed spatiotemporal pattern (Stirman et al. 2012; Sakai et al. 2013) (Fig. 3A,B). With this projector-managed optical system (PMOS), the depolarizing rhodopsins (such as, ChR2) and the hyperpolarizing rhodopsins (such as, ArchT and Mac) may be differentially activated (Fig. 3C,D). These devices would facilitate in vitro studies of neuronal networks and their dysfunctions, and studies using small animal models such as nematode worms, flies and zebrafish. They would also enable one to deliver multicolored and patterned light in the brain of animals in vivo in combination with a microendoscopy technique (Hayashi et al. 2012; Osanai et al. 2013).
In the future, researchers may be able to devise light-emitting nanoparticles that could be controlled and charged by magnetic fields or near-infrared light to generate specific wavelengths of light (Barandeh et al. 2012; Yue et al. 2012). If successful, it would become unnecessary to consider the problem of inserting many optic fibers into the brain.
Application to non-neural tissues
Although it was originally applied in neural tissues to manipulate the neuronal activity, optogenetics is widely applicable for other types of cells and biological systems (Table 4).
Table 4. Optogenetic manipulation of non-neural tissues
Glial cells in the brain tissue may be potential targets of optogenetic manipulation. For example, astrocytes, which had been considered merely supportive cells for neurons, were recently revealed to regulate synaptic transmission and plasticity and are involved in the regulation of brain function (Volterra & Meldolesi 2005; Perea et al. 2009; Henneberger et al. 2010; Allaman et al. 2011; Panatier et al. 2011; Min & Nevian 2012; Navarrete et al. 2012; Schmitt et al. 2012; Wang et al. 2012), although some controversy remains (Lovatt et al. 2012). Optogenetic stimulation induced the elevation of cytosolic Ca2+ in ChR2(H134R)-expressing astrocytes in the brainstem, and triggered respiratory activity in rats in vivo via an adenosine triphosphate (ATP)-dependent mechanism (Gourine et al. 2010). The elevation of cytosolic Ca2+, which triggers the release of glutamate via anion channels, was also optogenetically induced in astrocytes expressing either LiGluR, ChR2(H134R) or CatCh (Li et al. 2012). Photostimulation triggered the release of glutamate from Burgmann glia, which expresses ChR2(C128S), in the cerebellum (Sasaki et al. 2012). This was followed by AMPA receptor activation in Purkinje cells and induction of long-term depression of parallel fiber-to-Purkinje cell synapses via metabotropic glutamate receptor activation.
In the developing and adult brain, microglia are suggested to reshape synapses and thus to be involved in the regulation of normal brain function, such as learning and memory as well as pathological reactions (Nimmerjahn et al. 2005; Wake et al. 2009, 2011; Tremblay et al. 2010; Paolicelli et al. 2011; Ekdahl 2012; Schafer et al. 2012). Microglia dysfunction is possibly related to neurodevelopmental disorders such as obsessive-compulsive disorder and Rett syndrome (Chen et al. 2010; Derecki et al. 2012). They also give certain signals to astrocytes through ATP release (Pascual et al. 2012). Recently, Tanaka et al. (2012) generated transgenic mice that express ChRs in astrocytes and microglia under the regulation of the tTA-tetO system. This transgenic system may facilitate our understanding of how these glial cells function in the brain.
Other excitable tissues such as cardiac, skeletal and smooth muscles are potential targets of optogenetics because their functions are dependent on changes in the membrane potential. Photosensitive cardiomyocytes were differentiated from ChR2(H134R)-expressing embryonic stem cells (ESCs) by optical modification of their pacemaking activities (Bruegmann et al. 2010; Abilez et al. 2011). Additionally, transgenic mice and zebrafish were generated to study arrhythmias and to spatially map the cardiac pace-making region (Arrenberg et al. 2010; Bruegmann et al. 2010). In vitro as well as in vivo studies of cardiac dysfunctions will be facilitated by the use of optogenetics because heart muscles can be directly stimulated without contact and with high spatiotemporal resolution (Knollmann 2010). A potential clinical application of optical pacing was suggested by modifying the pacing of cardiomyocytes through optogenetic stimulation of HEK293 cells, which form syncytia with cardiomyocytes (Jia et al. 2011). Skeletal muscle myotubes that were developed from ChR2-expressing C2C12 myoblasts, an immortal cell line of murine skeletal myoblasts originally derived from satellite cells (Yaffe & Saxel 1977), were demonstrated to be contractile with optical stimulation, showing twitch-like contractions at low frequency and tetanus-like contraction at high frequency (Asano et al. 2012). In this study, a line of C2C12 myoblasts was established to express ChR2 and fused with unmodified C2C12 to form multi-nucleated myotubes (Fig. 4A). The maturation of these photosensitive myotubes was facilitated by rhythmic stimulation using either an electrical field or blue light, and they become contractile muscle fibers. To overcome muscle weakness such as muscular dystrophy and amyotrophic lateral sclerosis (ALS), human muscle tissue could also be substituted with optogenetically facilitated myogenic development of myoblasts derived from induced pluripotent stem cells (iPSCs) or mesenchymal stem cells (MSCs) derived from the recipient (Dezawa 2008). There are also potential bioengineering applications such as, wireless drive of muscle-powered actuators or microdevices (Feinberg et al. 2007), as skeletal muscle cells are effective force transducers that can generate contractile energy efficiently through biochemical reactions.
In the above studies, it was noteworthy that undifferentiated cells, such as ESCs and myoblasts, could have genes of light-sensing proteins to become differentiated photosensitive cells and organs. A significantly broader application of optogenetics could be facilitated if used in combination with the ESC/iPSC technology (Dolmetsch & Geschwind 2011; Dottori et al. 2011; Shi et al. 2012a). Human cells, particularly cells derived from patients, could be optogenetically studied to reveal the mechanisms of dysfunction and to evaluate the effectiveness of treatment (Zhang et al. 2010; Egawa et al. 2012; Israel et al. 2012; Shi et al. 2012b). This level of research would be further facilitated if researchers had open-access to human cell resources (Wray et al. 2012). ESCs/iPSCs could be made photosensitive by genetic modification with ChRs and enabled to differentiate with rhythmic photostimulation, as Ca2+ influx through voltage-dependent Ca2+ channels is facilitated by depolarization (Fig. 4B). Alternatively, Ca2+ redistribution could be directly mediated by photo-activated ChRs (Nagel et al. 2003; Berthold et al. 2008; Caldwell et al. 2008; Ernst et al. 2008; Kleinlogel et al. 2011a; Prigge et al. 2012). Photosensitive differentiated cells could be transplanted into tissue to study the function of cells or tissue in vivo or to exogenously regulate function using light (Weick et al. 2010; Stroh et al. 2011). For example, iPSC-derived photosensitive dopaminergic neurons could be optically regulated to release dopamine to reduce the symptoms of Parkinson's disease (Wernig et al. 2008; Gibson et al. 2012). Finally, the optical drive of photosensitive neurons could facilitate their survival and integration into the host's neural network (Kastanenka & Landmesser 2010; Wyatt et al. 2012) as these processes are known to be dependent on the activity in either developing (Hubel et al. 1977; Harris 1981; Katz 1999; Hensch 2004; Sanes et al. 2012) or adult animals (Koike et al. 1989; van Praag et al. 1999, 2005; Deisseroth et al. 2004; Komitova et al. 2006; Tashiro et al. 2007; Waddell & Shors 2008).
Optogenetic manipulation has broader applications than simply replacing electrical stimulation. Up- or downregulating light-sensitive proteins could be expressed in a subset of neurons under the regulation of the cell type-specific promoters and enhancers. The function of these neurons may then be determined with evidence of necessity and sufficiency using optogenetic strategies even in a living brain. It is possible some groups of neurons may express light-sensitive proteins in unintentional transgenic products, such as those of enhancer or gene trapping methods. Neurons projecting to a specific region or those connected to specific types of target cells have been retrogradely labeled and optogenetically studied (Lima et al. 2009; Ohara et al. 2009; Kato et al. 2011; Masamizu et al. 2011; Kiritani et al. 2012). Optogenetic manipulation is not limited to the membrane potential. For example, several intracellular messenger molecules and proteins have been optically manipulated; Ca2+ (Caldwell et al. 2008; Kleinlogel et al. 2011a), cyclic AMP (Iseki et al. 2002; Nagahama et al. 2007; Schröder-Lang et al. 2007; Ryu et al. 2010; Stierl et al. 2011; Weissenberger et al. 2011), cyclic GMP (Ryu et al. 2010) and Rho-GTPase (Levskaya et al. 2009). Animal rhodopsins have been modified to drive the G-protein-coupled signaling pathway by light (Kim et al. 2005; Li et al. 2005; Airan et al. 2009; Oh et al. 2010; Ye et al. 2011). It is also possible to manipulate gene expression directly (Takahashi et al. 2007) or indirectly by light (Ye et al. 2011). Thus, with the development of sophisticated optic systems, cellular functions may be elegantly and precisely manipulated.
This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas “Mesoscopic Neurocircuitry” (No. 23115501) and Grant-in-Aid for challenging Exploratory Research (No. 23659105) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan and the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO). We are grateful to Drs W. Shoji and M. Watanabe for reviewing the manuscript and to B. Bell for language assistance.