Optogenetics is a technique that involves the use of light to manipulate the activity of cells with high temporal and spatial precision, either in vitro or in vivo. By allowing individual cell types to be selectively targeted, and their activity switched on and off over a biologically relevant timescale of milliseconds, optogenetics provides a degree of specificity and control far greater than that which can be achieved using drugs or lesions. In 2010, Nature declared optogenetics to be their ‘method of the year’, while Science classed it as one of the breakthroughs of the last decade. So why has optogenetics generated so much excitement?

In 1999, Francis Crick—of DNA fame—suggested in a series of lectures at the University of California in San Diego that it should be possible to use light to precisely control the activity of individual neuronal subtypes. At the time it was not clear how this could be achieved, but the answer was to come only a few years later when researchers worked out how to mimic some of the processes at work in the eye.

The retina detects light with the aid of photoreceptors, which convert energy from photons into chemical signals that can modify neuronal activity. Photoreceptors contain a pigment called rhodopsin (Figure 1), which is made up of opsin—a 7-transmembrane helix protein—and a cofactor called retinal, which binds to a specific lysine residue in the 7th helix. Absorption of a photon causes retinal to isomerize, switching from 11-cis-retinal to all-trans-retinal. This induces a conformational change in the opsin (referred to as bleaching), which leads to activation of a G-protein coupled signaling cascade.

Figure 1. The photosensitive pigment molecule rhodopsin (bovine) with bound retinal (black). Image: S. Jähnichen (Wikimedia commons)

A number of researchers began to ask whether this system could be adapted to permit the control of neuronal activity using light. Gero Miesenböck, then at Yale, and co-workers identified three proteins that appeared to be essential for phototransduction in fruit flies: rhodopsin, arrestin-2 and the alpha subunit of the cognate heterotrimeric G protein. When they introduced the genes encoding these three proteins (which they referred to as “chARGe”) into cultured hippocampal neurons, the neurons fired action potentials upon exposure to white light. However, the time delay between light exposure and action potential firing was highly variable (ranging from a few hundred milliseconds to tens of seconds) and the three-gene system was relatively complex to manipulate [6].

In 2005, a breakthrough occurred when Karl Deisseroth and Ed Boyden at Stanford University found a way to make cells light-sensitive using only a single construct. It had been known since the 1970s that certain unicellular algae use light to control the flow of ions across their plasma membranes, but it was not until 2003 that researchers identified the protein responsible for this in Chlamydomonas reinhardtii (Figure 2.) [7]. Nagel et al demonstrated that the algae express a non-selective cation channel called channelrhodopsin-2 (ChR2), which is directly gated by blue light. Deisseroth and Boyden’s solution was to introduce ChR2 into mammalian hippocampal neurons in vitro and to show that shining blue light onto these cells caused them to fire 1–2 milliseconds later. Switching the light off caused the cells to stop firing, providing a simple and effective means of controlling neuronal activity [8, 9].

Figure 2. Channelrhodopsin-2 (ChR2) was first identified in the unicellular algae Chlamydomonas reinhardtii (pictured). Image: Dartmouth Electron Microscope Facility, Dartmouth College (Wikimedia commons)

The crystal structure of channelrhodopsin was solved early in 2012 [10]. Like vertebrate rhodopsin, ChR2 consists of the light-sensitive protein opsin and the bound cofactor retinal. But unlike rhodopsin, ChR2 is not coupled to a G-protein mediated signaling cascade; instead, ChR2 incorporates an ion channel directly, enabling it to respond much more rapidly to light.

When neurons are at rest, the inside of the cell is negatively charged relative to the outside. This generates a potential difference across the membrane of roughly -70 mV (the resting potential). When ChR2 in the neuronal membrane is illuminated with blue light, all-trans-retinal isomerizes to 13-cis-retinal, and this conformational change opens the ChR2 cation channel. Positively charged ions flow into the cell, which becomes depolarized (the potential difference across the membrane becomes less negative). If the depolarization reaches a critical threshold, a chain of events is initiated that leads ultimately to the cell firing an action potential (spiking). Switching the blue light off causes 13-cis-retinal to quickly revert to all-trans-retinal, closing the channel and repolarizing the cell.

Since their first proof-of-concept experiment, Deisseroth and Boyden have supplied optogenetic constructs, including new ones such as Chronos and Chrimson [11], to labs around the world, and use of the technique continues to grow. For example, TA Spix et al injected AAV2-synFLEX-Chronos-GFP and AAV2-syn-Chronos-GFP into mouse brains to study deep brain stimulation [12]. Zhang X et al injected AAV2/9-EF1α-DIO-ChR2-mCherry into the central nucleus of the amygdala and the paraventricular nucleus in mice to study the neuronal control of humoral immune responses in spleen [13]. Siciliano CA et al injected AAV5-DIO-ChR2-eYFP into mouse medial prefrontal cortex to study compulsive alcohol drinking [14]. Nakashima A et al generated olfactory sensory neuron-specific ChR2 mice to investigate olfactory map formation [15]. Marshel JH et al discovered a red-shifted channelrhodopsin, ChRmine, through functional metagenomic screening of Marine Microbial Eukaryote Transcriptome Sequencing Project from species Tiarina fusus strain LIS for their study of cortical layer-specific dynamics during perception [16]. ChRmine has been used to activate deep brain neural circuits without intracranial surgery [17].

Reversible optogenetic control of subcellular protein localization using Arabidopsis red light-inducible phytochrome (PHYB-PIF) system and precise light illumination has been described in zebrafish embryos [18] and has been developed into a protocol named optoSOS, which was used to study the dynamics of Ros-Erk signal transmission in normal and cancerous cells [19].

One of the strengths of optogenetics is its flexibility: the same technique can be used to up- or downregulate cellular activity in vitro or in vivo, in a multitude of brain regions (and, increasingly, other tissues) in a number of species. The same basic steps are required for all of these applications:

1. Design the construct. The gene encoding a light-sensitive opsin (e.g., ChR2, but a number of others are now available, each with distinct properties) must be placed under the control of a promoter that will direct its expression at the desired time and place.
2. Introduce the construct into cells. For in vitro experiments, this can be achieved through transfection (e.g., by using electroporation or calcium phosphate to make ‘holes’ in the cell membrane through which the construct can pass; or by placing the construct inside liposomes, which then fuse with the membrane and release their contents into the cell). For in vivo work, the construct is generally packaged into a virus, which is then injected into the target region (viral transduction). Alternatively, transgenic mice can be generated (or, in some cases, purchased) that express the opsin under the control of a specific promoter.
3. Select a light source. For in vivo experiments, this is typically an LED or a laser coupled to a fiber optic cable to allow delivery of light to a precise area.
4. Measure the effects of manipulating cellular activity. The method chosen will depend on the specific goals of the experiment, but could include techniques such as electrophysiological recordings, calcium imaging, opto fMRI, molecular biology or behavioral testing, to name but a few.

ChR2 is still a mainstay of the optogenetics toolkit, however there are also a number of additional variants available, which have been engineered to possess specific properties. These include the ChR1/VChR1 family of channelrhodopsins, which are more potent versions of ChR2 that are activated by red-shifted—rather than blue—light and CatCh, a variant with an accelerated response time and higher light sensitivity due to an increase in calcium permeability [20, 21]. Venkatesh HS et al transduced ChR2–YFP lentiviral vector (pLV-ef1-ChR2(H134R)-eYFP WPRE) into SU-DIPG-VI and SU-DIPG-XIII-FL cells to study optogenetic depolarization of glioma [22]. The existence of channelrhodopsins sensitive to other wavelengths makes it possible to perform combinatorial experiments, in which the activity of several populations of cells (each expressing a different opsin) can be manipulated simultaneously. Sensitivity to red-shifted light is particularly valuable because longer wavelengths tend to scatter less and can therefore penetrate deeper into tissue.

While the rapid activation and deactivation kinetics of ChR2 are advantageous in some experiments, other studies require stable neuronal activation over longer periods. To achieve this using ChR2 would require lengthy illumination of tissue, which can lead to damage due to overheating. To overcome this problem, a class of opsins known as step-function opsins (SFOs) has been developed. A single pulse of blue light switches the SFO channel to an open state, and this is maintained for up to a minute once the light has been switched off (alternatively, a single pulse of yellow light can close the channel). Stabilized step-function opsins (SSFOs) extend this principle even further, allowing stable activation of neurons several millimeters below the brain surface using only a brief, low-intensity pulse of light. The channel deactivation time constant of SSFOs is roughly 29 minutes [23]. By contrast, other variants of ChR2 such as the ChETA family [24] and ChIEF [25, 26] allow neurons to be triggered to fire at higher frequencies than is possible with ChR2.

To inhibit neuronal firing, a light-sensitive inwards chloride channel called halorhodopsin (NpHR) — which is derived from the halobacterium Natronomonas pharaonis — can be used. When illuminated with yellow light, NpHR conducts chloride ions into the cell, which thus becomes hyperpolarized. Most experiments today use an ‘enhanced’ halorhodopsin known as eNpHR3.0, which has been mutagenized to generate larger photocurrents and more effective membrane hyperpolarization than NpHR. For example, Siciliano CA et al injected antegradely-transporting AAV5-CaMKIIα-eNpHR3.0-eYFP into mouse medial prefrontal cortex to achieve photoinhibition at dorsal periaqueductal gray [14]. An alternative option is to use the light-activated outwards proton pump, archaerhodopsin (Arch); activation of this leads to an outwards flux of protons, again hyperpolarizing the cell [27, 28]. Mutations of channelrhodopsins led to the generation of another inhibitory light-sensitive chloride channel, iChloC [29], and iChloC has been used by N Kataoka et al [26].

A red-shifted cruxhalorhodopsin, Jaws, derived from Haloarcula (Halobacterium) salinarum (strain Shark) has also been developed to enable optical activation though more penetrating red light [30]. For example, Kim J et al injected AAV8-hSyn-Jaws-KGC-GFP-ER2 from UNC Viral Core in Long Evans rats for neural silencing [31].

However, it is not only ion channels that can be controlled with light: intracellular signaling cascades can be activated or inhibited using OptoXRs. These are opsin-receptor chimeras in which the intracellular loops of the rhodopsin protein have been replaced with intracellular loops from other G-protein coupled receptors. This enables G-protein mediated signaling cascades to be switched on and off within cells, eliciting patterns of cellular activity that are potentially more physiologically relevant than those that result from the light-induced opening and closing of ion channels.

The opsin of choice must then be placed under the control of a promoter, which will drive its expression in cells. The simplest approach is to place the opsin downstream of a strong ubiquitous promoter, such as elongation factor 1α (ELF-1α), synapsin, cytomegalovirus (CMV) or CAG. This will give robust opsin expression in almost any cell type in which the construct is present. Alternatively, cell-type specific promoters such as α-calcium/calmodulin-dependent kinase II (αCamKII), which is expressed in forebrain pyramidal neurons, can be used to target opsin expression more precisely. However, expression from these promoters can be relatively weak compared to that from ubiquitous promoters.

Figure 3. An example of a gene cassette that can be used to achieve conditional expression of an opsin. ChR2 and a fluorescent marker (mCherry) are placed downstream of a floxed stop cassette (red hexagon). Excision of the stop cassette by Cre recombinase leads to expression of opsin/marker. Image: adapted from Schmitt et al., 2012 [1].

The choice of a promoter can also be influenced by the mechanism that will be used to introduce the construct into the tissue. For in vitro experiments, the construct is inserted into a bacterial plasmid, and the plasmid is then introduced into cells via transfection. This involves making temporary ‘holes’ in the cell membrane, e.g., through the addition of calcium phosphate, or by exposing the cells to short pulses of an electric field (electroporation). Although technically challenging, electroporation can also be performed in utero: introducing the construct into a single embryonic cell at a specific stage of development ensures that the construct will be expressed in the entire lineage descended from that cell. Plasmid-based approaches such as these permit the use of relatively large promoters to drive opsin expression.

• Electroporation or viral vectors. The majority of in vivo experiments use viral transduction to introduce the promoter-opsin construct. This limits the size of the promoter because the entire construct must be packaged into the virus. The most commonly used viruses are adeno-associated viruses (AAV) and lentiviruses. Approximately 5 kilobase pairs (kbp) worth of genetic material can be packaged into an AAV, whereas lentiviruses can hold roughly 8 kbp. However, AAVs have the advantage that they can replicate their DNA in the cytoplasm of the host cell. For example, Hirokawa J et al injected adeno-associated virus 2/9 serotype carrying EF1a-DIO-ChR2-EYFP32 or hSyn-DIO-{mCAR}off{ChR2}on and CAV2-Cre 4.1E12 into rat striatum or orbitofrontal cortex [32]. Zhang J et al injected into AAV-DIO-ChR2 or AAV-DIO-GFP into a specific region to study sour sensing [33]. Szőnyi A et al injected AAV2/5-EF1α-DIO-hChR2(H134R)-eYFP from Penn Vector Core into mouse brain regions to study the role of brainstem nucleus incertus GABAergic cells in contextual memory formation [34]. By contrast, lentiviruses must integrate their genetic material (in the form of RNA) into the host genome, and this can sometimes result in the deleterious disruption of host genes.
  The virus is introduced into the target tissue by stereotaxic injection, whereupon it will infect many cells. However, only those that express the selected promoter will manufacture the opsin protein.
• Using the Cre recombinase system
  One of the drawbacks of using cell-type specific promoters is that they can give rise to relatively low levels of gene expression. An alternative means of achieving robust cell type-specific gene expression is to combine the use of a strong ubiquitous promoter with that of a Cre recombinase driver mouse. This will limit activation of the promoter to the desired cell types. A full discussion of the Cre recombinase system is beyond the scope of this review. However, in brief, Cre recombinase is an enzyme that catalyzes site-specific recombination between two DNA recognition sequences known as loxP sites. Any DNA that is present between two loxP sites of the same orientation (‘floxed’ DNA) will be excised.
  In optogenetics, an opsin gene is often placed downstream of a strong ubiquitous promoter (such as ELF-1α or CMV), but separated from it by a stop cassette. When transcribed and translated, the stop cassette introduces a stop codon into the polypeptide chain, preventing expression of the opsin. However, the cassette is flanked by two loxP sites. This means that when the construct is injected into the brain of a Cre driver mouse—in which Cre recombinase is expressed in specific cell types or brain regions—excision of the stop cassette and expression of the opsin will occur, but only in those areas that express Cre.
  However, the size limitations imposed by viral vectors mean that only relatively small stop cassettes can be used, and this can lead to ‘leakiness’ with the expression of the opsin gene in Cre-negative cells. To overcome this problem, many researchers now use double floxed inverted open reading frame (DIO) viral vectors, which give selective Cre-mediated opsin expression without requiring the use of stop cassettes. In brief, an inverted version of the opsin gene is flanked by two incompatible loxP variants. In the presence of Cre recombinase, inversion of the opsin gene occurs, and the opsin is expressed.
• Ready-made transgenic mice
  In recent years, it has become possible to purchase mice that show stable opsin expression under the control of promoters such as ChAT (specific to cholinergic neurons) or VGAT (GABAergic interneurons) throughout the brain. Alternatively, researchers can purchase a line of mice expressing a floxed opsin gene, together with an additional Cre driver line (increasing numbers of which are also available for purchase). Crossing these two lines will generate offspring that show stable opsin expression in Cre-expressing tissues.

Opsins can be targeted not only to specific cell types, but also to specific sites within cells (i.e., to the axon or axon terminal of neurons rather than the cell body). This can be achieved by fusing Cre recombinase to proteins that are transported along axons, such as wheat germ agglutin (WGA) or tetanus toxin fragment C (TTC). An alternative strategy is to focus the light source on axons rather than cell bodies. This makes it possible to modulate the activity of neurons based on their projections rather than their pattern of gene expression, and can be useful in species that are less amenable to genetic manipulation, such as rats and primates.

Precise control of cellular activity can only be achieved if there is precise control of the light that activates the opsin. For in vitro experiments, a constant light source might be combined with an ultra-fast shutter, or a switch used to control the activity of an LED. Miniaturized LEDs are sometimes used in vivo, but a common alternative is an optrode (an LED or a laser light source coupled to an optical fiber). LEDs have the advantage of being cheaper and easier to control than lasers, but they emit light in all directions and this can limit the amount of light that enters the optrode.

Simply bathing large populations of neurons in light will not mimic the precise firing patterns of the brain, in which neuronal activation is under fine temporal control. Moreover, the brain is a 3D object. Ed Boyden’s group has therefore developed a device consisting of microfabricated waveguides, which can be used to deliver multiple individual light sources into the brain in a 3D pattern [35]. Anthanide-doped upconversion nanoparticles can also be used to convert tissue-penetrating near-infrared light to the activating blue emission [36].

Optogenetics has traditionally been combined with standard electrophysiological read-outs (field and whole-cell recordings), or with indirect measures of neuronal activity such as calcium imaging [37]. However, in 2010, Lee and co-workers integrated optogenetics with functional magnetic resonance imaging to give ‘ofMRI’ [38]. They used the technique to demonstrate that increased activity in local excitatory neurons caused (as opposed to simply correlated with) an increase in the BOLD signal, and showed that ofMRI could be used to visualize the effects of precise optogenetic manipulations on global brain activity. Yang W et al used volumetric two-photon calcium imaging to measure neural activity in mouse neocortex in vivo with cellular resolution along with neural stimulation with two-photon optogenetics [39].

One of the key strengths of optogenetics is that it can be used to study the effects of temporarily and reversibly altering neuronal activity on an animal’s behavior (e.g., [40]. This in stark contrast to lesions, which are irreversible, and pharmacological manipulations, in which hours or days may be required for drugs to wash out of the brain.

Two key factors to control for are the expression of the opsin and activation of the light source. Cells/animals that express the opsin should be compared with those that do not, to control for the possibility that insertion of the opsin gene may have disrupted other genes in the host genome. It is also important to compare cells/animals in the presence and absence of light. This is because illumination can lead to heating, which can cause tissue damage. Owen SF et al reported 0.2-2 °C temperature increase through commonly used illumination protocols [41]. Blue light was found to induce marked heating as well as changes in the fMRI signal [42] and to alter gene expression in mouse cortical cultures [43], likely due to phototoxic interactions with the culture media [44].

The nematode worm, Caenorhabditis elegans (Figure 4) was the first multicellular animal to have its genome fully sequenced and is a popular choice in optogenetics experiments. Although C. elegans does not produce all-trans-retinal, this can be added to the bacterial lawn that the worms feed on. Worms that are raised on bacterial lawns without the addition of all-trans-retinal can then serve as valuable experimental controls.

Figure 4. The anatomy of C. elegans. Image: Blaxter (2011) [2].

The worm is otherwise well-suited to optogenetics by virtue of its simple nervous system (consisting of just 320 cells) and transparent body wall, which renders those cells readily accessible to light. In 2005, Nagel et al produced genetically modified worms that expressed ChR2 in mechanosensory neurons, or in motor neurons within their body wall. Shining blue light on the worms activated these neurons and caused the worms to reverse or to contract their muscles respectively [45]. Two years later, ChR2 and halorhodopsin (NpHR) were used to produce bidirectional control of C. elegans body wall muscle contraction [46].

Subsequent advances have made it possible to selectively target individual cells. This has been accomplished via the use of combinatorial genetics to limit opsin expression to cells that express overlapping promoters [1], and via precise targeting of the light source. Two studies from 2011 achieved the latter in freely moving worms, one with the aid of an LCD projector [47] and the other using a digital micromirror device (DMD) that consisted of hundreds of thousands of microscopic independently adjustable mirrors [48]. Both systems used specially designed algorithms to track and predict the position of a target cell within the body of a moving worm, so that the light source could be aimed accordingly. The effects on behavior of manipulating the activity of individual neurons could then be studied.

C. elegans has often been used in research into synaptic proteins and mechanisms of synaptic transmission, and optogenetics has continued this tradition. ChR2-mediated stimulation was used to study synaptic transmission at the neuromuscular junction [49], while the light-activated adenylyl cyclase PACα has been used to manipulate cAMP production at the same synapse [50].

Some of the first experiments to use light to control the electrical activity of genetically modified neurons were performed in the fruit fly (Drosophila melanogaster) by Gero Miesenböck and co-workers (see ‘A brief history of optogenetics’). This group was also among the first to demonstrate the remote control of behavior using optogenetics. In 2005, Lima and Miesenböck used light to elicit escape behaviors—such as jumping, wing beating and flight—in transgenic flies that possessed a subset of neurons that had been genetically modified to render them photosensitive [51].

By 2007, ChR2 was being used in Drosophila (although, like C. elegans, Drosophila do not produce their own retinal and must be supplied with it in food). Hwang et al used optogenetics to identify the neuronal basis of the pain response in flies [52]. They generated flies that expressed ChR2 in individual neuronal classes, and showed that illuminating classes I, II or III triggered widespread muscle contraction. By contrast, illuminating class IV neurons induced defensive pain-related behavior, suggesting that these neurons are Drosophila nociceptors (pain receptors).

Optogenetics has been used to condition appetitive and aversive responses to odors in Drosophila. Fruit flies are attracted to sugary substances such as marzipan and bananas, and the activation of olfactory neurons that are sensitive to these particular odors triggers an approach response. By contrast, flies find other odors repulsive, and the activation of olfactory neurons sensitive to these odors will trigger an escape response [53].

Bellman and co-workers placed individual fruit flies in a petri dish in which two quadrants were illuminated with blue light, and two were not (Fig. 5a). Fruit flies normally find blue light aversive (Fig. 5b). However, when the researchers used the blue light to activate ChR2-expressing neurons sensitive to desirable odors, the flies suddenly found the blue light far less aversive than they had previously (even though there was no odor present) (Fig. 5c). Conversely, when the researchers illuminated neurons sensitive to aversive odors, the flies behaved as though as they had detected an unpleasant smell and moved away from its apparent source (the light) [3, 54].

Figure 5. Optogenetics can be used to control the behavior of fruit flies. (a) Individual flies are placed in a petri dish in which two quadrants are illuminated with blue light. (b) Fruit flies normally find blue light aversive. (c) When blue light triggers the activation of neurons sensitive to desirable odors, fruit flies lose some of their aversion to blue light. Image from: Störtkuhl and Fiala, 2011 [3] (NB. From a Frontiers journal but published under a CCBY 3.0 license)

Using a similar principle, Miesenböck and co-workers used optogenetics to make fruit flies learn from a ‘mistake’ that they only thought they had made. Miesenböck proposed that the fly’s decision-making circuitry consists of an ‘actor’, which directs actions, and a ‘critic’ that monitors the consequences of those actions and provides feedback to the actor. Manipulating the activity of the critic should therefore make it possible to control the fly’s behavior. To test this idea, Miesenböck and co-workers placed individual flies in a horizontal tube with a different odor concentrated at each end. The flies walked back and forth inside the tube, sampling the two odors. Whenever they reached the midpoint, they had to make a decision: turn back and remain in the familiar odor, or continue onwards and try a new one.

The flies had been genetically manipulated to express an opsin in a random subset of dopaminergic neurons throughout the brain. Whenever the flies chose one of the odors, the researchers activated the dopaminergic neurons. If those neurons comprised (or contained) the ‘critic’, this should make the fly want to avoid whichever odor it found itself in during the illumination. If those neurons did not include the critic, the fly’s behavior should remain unchanged.

The researchers tested large numbers of flies, each expressing light-sensitive proteins in a different combination of dopaminergic cells. In the majority of cases, the light had no effect on the insects’ behavior. However, in a minority of cases, the flies did begin to avoid the odor in which they had experienced the illumination. Through many such experiments, the researchers were able to narrow down the identity of the ‘critic’ to a group of 12 dopaminergic cells that send their output to a structure called the mushroom body. They suggest that this structure may be the ‘critic’ (or reinforcement signal generator) in the Drosophila decision-making system [55, 56] ).

Zebrafish, which are semi-transparent and genetically well characterized, is another popular model organism in optogenetics. In 2008, researchers used light to activate two populations of ChR2-expressing somatosensory neurons, and found that activation of either population triggered an escape response similar to that triggered by tactile stimulation [57]. In the case of one-third of the neurons, light activation of a single cell was sufficient to trigger the escape response, while for trigeminal neurons, a single action potential was enough.

The organism that has been used most frequently in optogenetics experiments, however, is probably the mouse. Indeed optogenetics owes its popularity at least in part to the way it complements and exploits advances in the field of mouse genetics. An ever-expanding range of transgenic mice showing stable opsin expression in specific cell types is available for purchase, while the relatively extensive behavioral repertoire of this species—combined with the ease with which it can be kept in the lab—has made mice the first choice when modeling many human diseases. While it is likely that applications of optogenetics will move beyond the brain, rodent studies have tended to focus on human neurological and psychiatric disorders [58]. Abdo H et al generated Plp-ChR2 and Sox10-ChR2 mice to investigate the role of nociceptive Schwann cells in pain perception [59]. CckCRE_ChR2-tdTomato and CckCRE_Halorhodopsin-YFP mice derived from LSL_ChR2-tdTomato Jackson Lab stock 012567 and LSL_Halo-YFP Jackson Lab stock 014539, respectively, were used to investigate the synaptic formation between neuropod cells and vagal neurons [60]. Adam Y et al construct OptoPatch3 transgenic mice to enable the simultaneous optical perturbation and measurement of membrane action potential [61]. Carta I injected directly AAV1-SynChR2-YFP from Penn Vector Core into deep cerebellar nuclei to investigate the projections from cerebellum to the ventral tegmental area and modulation of the reward circuitry by these projections [62].

Anxiety is characterized by a sustained state of vigilance even in the absence of any immediate or impending threat. The role of the basolateral amygdala (BLA) in anxiety is well-established; however, this structure projects to many other brain regions and the relative importance of each of these pathways is unclear. To address this question, Tye et al used channelrhodopsin to activate BLA neurons and showed that doing so induced anxiety in mice. By contrast, selectively activating only those BLA neurons with axons that project to the central nucleus of the amygdala (CeA) had the opposite effect [40]. These data thus reveal a striking dissociation between the consequences of activating neurons that occupy the same brain region, but which have distinct projection targets. In a more recent study, efferent projections from the bed nucleus of the stria terminalis to different brain regions were likewise shown to have distinct roles in anxiety [63].

Optogenetics can also be used to selectively target particular cell types within a single structure. For example, activation of granule cells within the dorsal part of the hippocampal dentate gyrus mediated encoding of contextual fear memories, whereas activation of granule cells in the ventral dentate gyrus had no effect on contextual fear learning, but suppressed innate anxiety [64].

The causes and underlying biology of depression are complex, with multiple neural circuits and neurotransmitter systems implicated in the disorder. A number of mouse models of depression have been developed, mostly based on the principle of learned helplessness. This is the idea that a depressed individual will be less motivated to try to escape an unpleasant situation than a healthy individual, and will more quickly become resigned to their fate. In the ‘tail suspension test’, healthy mice that are restrained by their tails will struggle and try to get away; ‘depressed’ mice will remain largely still. Healthy mice that are placed in a bucket of water will swim around continuously in search of an escape route; ‘depressed’ mice will float motionless on the surface.

With the aid of pharmacological or genetic approaches to induce a depression-like phenotype in mice, and tests such as the tail suspension test as an output measure of behavior, optogenetics can then be used to probe the underlying disease pathology. It can also be used to gain a better understanding of the mechanisms by which antidepressant treatments work. Thus, illumination of local ChR2-expressing neurons within the medial prefrontal cortex produced an antidepressant effect in mice [65], while bidirectional control of specific midbrain dopamine neurons bidirectionally modulated symptoms of depression in response to chronic stress [66]. Optogenetic stimulation of phasic, but not tonic, firing of dopaminergic neurons in the ventral tegmental area (VTA) induced susceptibility to depression in mice that were experiencing social defeat stress [67]. These effects were pathway-specific: phasic activation of VTA neurons projecting to the nucleus accumbens, but not to the medial PFC, induced susceptibility to stress. Inhibition of VTA-accumbens projections, on the other hand, induced resilience.

Drug addiction is a chronic relapsing condition marked by compulsive drug seeking that continues despite harmful consequences. Addictive substances are thought to ‘take over’ the brain’s natural reward system—this is centered on the mesolimbic dopamine system, which extends from the ventral tegmental area (VTA) via the medial forebrain bundle to the nucleus accumbens. However, the precise details of how this circuitry mediates reward and addiction are unclear.

Cre-mediated recombination has been used to target opsin expression to specific populations of dopaminergic cells. Behavioral tests such as conditioned place preference (CPP)—in which a rodent quickly develops a preference for an environment that has been associated with a rewarding event over one that has not—can then be used to study the effects of optogenetic manipulation of dopaminergic circuitry.

Using this technique, phasic but not tonic stimulation of ChR2-expressing dopaminergic cells within the VTA was shown to be sufficient to induce CPP [68]. Within the nucleus accumbens, activation of neurons that express the D1 subtype of dopamine receptors increased cocaine CPP, whereas activation of D2 expressing neurons had the opposite effect [69]. Precisely timed upregulation of G protein signaling within the accumbens using an OptoXR was likewise sufficient to drive CPP [70], while silencing the cocaine-induced activation of cholinergic interneurons within the accumbens was enough to block CPP [71].

It has been known for almost 60 years that animals will repeatedly press a lever to deliver an electric current into the brain’s reward centers (intra-cranial self-stimulation) [72], but it has not been clear which cell types mediate this effect. Optogenetic experiments revealed that excitatory projections from the basolateral amygdala to the nucleus accumbens support self-stimulation, via a mechanism that is dependent on dopamine D1 signaling. Moreover, optogenetic inhibition of these fibers can reduce sucrose consumption, implicating this pathway in response to natural reinforcers too [73]. Animals have also been shown to self-administer optogenetic stimulation of dopamine neurons in the VTA [74]. If optogenetic self-stimulation could be used as a robust and reliable reinforcer, this would facilitate rodent behavioral testing by allowing animals to be tested over many trials without the need for food deprivation beforehand.

Indeed, optogenetics has already been used to compare the reward value of different reinforcers. Mice were offered a choice between taking sips from a drinking bottle of water—which was paired to optogenetic activation of dopaminergic neurons—and natural or artificial sweeteners. The mice preferred optogenetic activation of dopamine neurons to sucralose, although not to sucrose [75].

Parkinson’s disease is a neurodegenerative disorder characterized by rigidity, tremor and slow movement. Optogenetics has made several valuable contributions to our understanding of the brain circuitry that underpins movement, both in the healthy and the diseased brain. Thus optogenetics provided the first clear empirical evidence for the long-hypothesized role of the direct and indirect pathways in the striatum for the control of movement. By expressing ChR2 in striatal medium spiny neurons in a mouse model of Parkinson’s disease, it was shown that activation of the indirect pathway results in a Parkinsonian-like state, whereas activation of the direct pathway reverses Parkinsonian symptoms [76].

Figure 6. Using optogenetics to study interactions between the graft and host tissue. Stem-cell derived dopaminergic neurons expressing GFP were grafted into an organotypic mouse striatal culture. The grafted neurons were seen to extend complex processes into the surrounding host tissue.

Deep brain stimulation of components of the basal ganglia circuit — in particular, the subthalamic nucleus—has been shown to help alleviate the symptoms of Parkinson’s disease. The cellular basis of this effect is largely unclear, but optogenetic activation of afferent fibers in the subthalamic nucleus also improved motor functioning in a hemi-Parkinsonian rat model [77]. Notably, stimulation of local cell bodies did not have the same effect, suggesting that deep brain stimulation in human patients should predominantly target axons and white matter tracts. Indeed, it may well be the case that the flow of information between brain regions is more pertinent to psychiatric and neurological disease than the activity of individual brain regions per se.

Another promising therapeutic avenue in Parkinson’s disease is the use of dopaminergic neurons derived from stem cells as replacements for dysfunctional tissue in the brains of patients. To investigate how effectively such grafts could integrate into existing circuitry, Tønnesen and co-workers grafted stem-cell derived dopaminergic neurons into organotypic cultures of wild-type mouse striatum [78].

After preparing the cultures, they used viral transduction to introduce channelrhodopsin and halorhodopsin into dopaminergic neurons belonging either to the host or the graft. Activating and inhibiting the cells revealed extensive synaptic connectivity within the graft, as well as complex bidirectional synaptic interactions between host and graft cells.

Figure 7. Using optogenetics to study interactions between the graft and host tissue. Graft–to–host synaptic connectivity in organotypic slice cultures probed with ChR2. Selective activation of ChR2-expressing graft cells with blue light-induced outwards (hyperpolarizing) currents in host cells. Image: adapted from Tønnesen et al. (2011) [4].

Epilepsy, in which excessive synchronous activity arises in one or more brain regions, leading to a seizure (or ‘fit’), is another key target for optogenetics research. Epileptic seizures can result from a loss of inhibitory interneurons, or from reorganization or inappropriate strengthening of excitatory pathways. In either case, the use of halorhodopsins or proton pumps to tone down the aberrant rhythmic neuronal activity that occurs during seizures may offer therapeutic benefits. Consistent with this, activation of NpHR in organotypic hippocampal cultures in which seizure activity had been induced using electrical stimulation, significantly reduced epileptiform bursting [79]. Moreover, in a mouse model of temporal lobe epilepsy, in vivo activation of a subpopulation of GABAergic interneurons, or inhibition of excitatory pyramidal cells, rapidly stopped seizures [80]. Epilepsy can sometimes arise as a result of the reorganization of long-range connections within the brain after a stroke. In rats, optogenetic targeting of thalamic neurons that were connected to areas of injured epileptiform cortex also reduced seizures [81].

Optogenetics has been used to study the neural basis of sleep and sleep disorders. ChR2-mediated activation of orexin/hypocretin-expressing neurons in the lateral hypothalamus of mice increased transitions from sleep to wakefulness [82], consistent with reports that dysfunction of these neurons is associated with the sleep disorder narcolepsy. In another study, researchers used light-induced activation of orexin/hypocretin neurons in the lateral hypothalamus to fragment sleep, and found that this impaired memory performance in a novel object recognition task even when total sleep duration was preserved [83].

Efforts are underway to use optogenetics to study the mechanisms that underpin learning and memory. There is some suggestion, for example, that the pathways that mediate memory formation may show asymmetry within the brain. Kohl et al (2011) used ChR2 to activate axons of CA3 cells originating in either the left or right mouse hippocampus. Activating CA3 inputs from the left hemisphere induced greater strengthening of CA3–CA1 synapses (long-term potentiation) than activating equivalent inputs from the right hemisphere. This may reflect differences in the distribution of glutamatergic NR2B receptors between the two hemispheres [84].

Optogenetics has been used to manipulate brainstem and spinal cord function. Thus, activation of a group of neurons within the ventral medulla region of the brainstem in anesthetized rats (the retrotrapezoid nucleus-parafacial respiratory group) induced active expiration [85]. Spinal cord injury can often lead to respiratory deficits owing to disruption of descending inputs to respiratory motor neurons in the spinal cord. However, in a rodent model of spinal cord injury, transducing neurons in the spinal cord with ChR2 and then activating these neurons with light restored rhythmic movements of the diaphragm. Moreover, the rhythmic activity persisted after the light had been switched off, suggesting a new therapeutic strategy for tackling respiratory impairments in spinal cord injury [86].

Encouraging data have been obtained with respect to potential therapeutic applications of optogenetics in retinitis pigmentosa: a heterologous group of progressive degenerative diseases in which loss of photoreceptors (first rods and then cones) leads to incurable blindness. Retinitis pigmentosa is estimated to affect some 2 million people worldwide. In a mouse model of the disorder, transducing surviving retinal cells with ChR2 and then activating these cells with light restored the ability of the retina to transduce light into electrical signals that were relayed to visual cortex [87]. Similar manipulations improved visually guided behavior in a rat model [88]. Activation of ChR2 in blind transgenic mice and rats that expressed the opsin in only a subset of retinal ganglion cells also restored light sensitivity to the retina and permitted visually guided behavior [5, 89].

Figure 8. Degeneration of the retina in a rat model of retinitis pigmentosa. Left: the intact retina. Middle: degeneration extends throughout the retina. Right: ChR2 (and GFP) are expressed in a subset of retinal ganglion cells. Image: Adapted from Tomita et al., 2009 [5].

Rather than using channelrhodopsins to switch on so-called ON bipolar cells or retinal ganglion cells, an alternative tactic is to use halorhodopsin to target the dysfunctional light-insensitive cones that often remain behind in the retina after the rods have died. Switching cones off with halorhodopsin allowed them to become re-sensitized to light, and restored visually guided behavior in two mouse models of retinitis pigmentosa. It also reactivated light-insensitive photoreceptors in human ex vivo retinas [90].

Results such as these raise the possibility that optogenetics could ultimately be used as a therapeutic tool in humans. But how plausible is this?

Chow and Boyden [91] propose that optogenetics could be used as a therapeutic intervention in conditions for which there is a clear benefit to be obtained from switching on or off specific cell types, and for which there is no better alternative available. In addition to retinitis pigmentosa, an increasing body of in vitro work suggests that this could include Parkinson’s disease and epilepsy, and perhaps even spinal cord injury.

However, there are a number of problems that must be overcome for this to become a reality, as Chow and Boyden discuss. Optogenetic therapeutics will require the delivery of genetic constructs into the body by means of a viral vector, but this technology is far from fully established in humans. Once inside the body, bacterial or algal opsins would need to be stably expressed for weeks, months or even years, and this could trigger an immune response. Human opsins (e.g., melanopsin or rhodopsin) could perhaps be used instead, but these respond more slowly to light than their microbial counterparts. Moreover, while the use of human opsins would avoid the need for inter-species gene transfer, it would not overcome the potential dangers associated with insertional mutagenesis as the engineered constructs integrate into the genome.

Furthermore, it is important to remember that cells are components of highly complex networks. Altering the function of any one cell (or group of cells) is likely to have knock-on effects for the activity of others. This may explain why the light was found to inhibit—rather than activate—one population of ChR2-expressing pyramidal neurons during recordings from the frontal cortex of an awake rhesus macaque [92]. Therapeutic applications of optogenetics in humans must therefore ensure that any beneficial effects are not outweighed by deleterious knock-on effects elsewhere.

Optogenetic techniques have been applied to repair pacemaker activity of damaged myocardium. In particular, ChR2, which can be activated by light, has been used to stimulate and depolarize cardiomyocytes in culture and in vivo [93]. Another study has demonstrated stimulation of rat cardiomyocytes by introducing ChR2 gene inserted into an adeno-associated virus 9 vector [94]. A monochromatic light emitting diode 450 nm has been used to produce flashes to the ChR2 transgene, which caused stimulation of the ventricles. Furthermore, optogenetic methods may be applied to terminate ventricular arrhythmias. Effective termination of experimentally induced arrhythmia in Wistar rat myocardiocytes, which expressed red-activatable channelrhodopsin (ReaChR). A cardiotropic viral vector has been used to deliver ReaChR [95]. In line with this data, red light effectively interrupted ventricular arrhythmias in ChR2-transgenic mice [96]. Viral transduction is considered to be the most effective methods to express exogenous proteins in mammary cells. Furthermore, melanopsin has also been suggested for studies of the pathogenesis of arrhythmias. Blue light-mediated activation of melanopsin stimulated phospholipase C, Ca²⁺ release and spontaneous pacemaking effects [97].

Several new therapeutic strategies for diabetes have been developed using optogenetics. Ye at al. have described a blue light-controlled method to control blood glucose level [98]. Melanopsin, which responds to blue light, was expressed in HEK-293 cells. Activated melanopsin stimulated the pathways regulating glucose homeostasis. When the melanopsin-expressing cells were engrafted into a mouse model of diabetes, the animals showed a balanced level of glucose. In addition, an optogenetic device LightOn, which contains a blue light-sensitive domain, has been described by Wang et al [99]. The LightOn device has been introduced into diabetic mice, which showed activation of insulin production in response to blue light. Also, control of insulin production with light irradiation without the involvement of glycolysis has been achieved by inducing ChR2 expression in pancreatic beta cells [100].

Since channelrhodopsin was first used to control the activity of mammalian neurons in 2005, a great deal of progress has been made in a short space of time. Thanks to optogenetics, it has become possible to either increase or decrease the activity of specific neuronal populations in multiple brain regions on demand. Optogenetics has provided mechanistic insights into a range of human psychiatric and neurological diseases, and its influence is likely to expand further as new tools allow the technique to be applied to a wider range of cells. Encouraging data have been obtained from the use of optogenetics as a therapeutic intervention in animal models of human disease, although much work remains before such an approach could be moved into human patients.

Dr. Konstantin Yakimchuk contributed to the section "Potential therapeutic applications of optogenetics in cardiology and endocrinology" in Oct 2018.