The development of optogenetics, a family of methods for using light to control neural activity via light-sensitive proteins, has provided a powerful new set of tools for neurobiology. These techniques have been particularly fruitful for dissecting neural circuits and behaviour in the compact and transparent roundworm Caenorhabditis elegans. Researchers have used optogenetic reagents to manipulate numerous excitable cell types in the worm, from sensory neurons, to interneurons, to motor neurons and muscles. Here, we show how optogenetics applied to this transparent roundworm has contributed to our understanding of neural circuits.
One of the fundamental goals of neuroscience is to understand how neural circuits create behaviour. Caenorhabditis elegans, a 1 mm-long roundworm, possesses a number of unique advantages as a model for addressing this question. It has a compact nervous system containing 302 neurons (in the adult hermaphrodite) and is the only animal for which the complete ‘wiring diagram’ or ‘connectome’—the complete atlas of synaptic connectivity—has been mapped . This wiring, and the overall anatomy of the nervous system, are highly conserved between individual C. elegans. Despite its relative anatomical simplicity, the worm's neurochemistry and genetics are remarkably similar to those of mammals. Its nervous system signals in part through the classical neurotransmitters glutamate, acetylcholine, gamma-aminobutyric acid (GABA) and biogenic amines, along with a large number of neuropeptides.
One of its most important assets of C. elegans as a model has been its optical transparency. Using differential interference contrast microscopy, Sulston et al. traced the organism's complete cell lineage during development . Transparency enabled the targeted killing of specific cells in intact worms using a laser microbeam coupled through a microscope; this approach has generated a great deal of our knowledge of C. elegans nervous system function . The worm was the first animal in which the green fluorescent protein (GFP) was expressed, helping to kindle a revolution in live biological imaging . More recently, early work in monitoring neural and muscle activity through fluorescence of calcium reporters was performed in worms . In 2005, worms became one of the first animals in which optogenetic manipulation of excitable cells was performed , and they continue to be important for the development and testing of new opsins.
In this review, we aim to summarize the insights gained about neural circuits from one decade of optogenetics in worms.
2. Optogenetic methods
(a) Optogenetics strategies
The earliest approaches for genetically conferring light sensitivity to cells were based on components of the Drosophila visual phototransduction machinery [7,8] or synthetic photoisomerizable groups coupled to receptors [9,10]. Around the same time, Boyden et al.  reported a strategy based on expression of microbial opsins. Heterologous expression of the cation channel Channelrhodopsin-2 (ChR2) from the single-cell alga Chlamydomonas reinhardii  conferred blue-light sensitivity in cultured neurons. The first in vivo demonstrations of ChR2 function were performed in chick embryo  and in C. elegans body wall muscles .
The single-component optogenetic strategy exemplified by ChR2—expression of microbial ion channels and pumps—is the most widely adopted by the neuroscience community, probably owing to its simplicity compared to the multiple-component methods. We therefore focus on this approach. However, researchers continue in efforts to develop strategies to perturb specific intracellular signalling pathways, as reviewed below.
(b) Opsins for stimulation
ChR2 consists of the apoprotein channelopsin-2 (chop-2) bound to the cofactor retinal, a form of vitamin A, which isomerizes from all-trans to 11-cis upon illumination. In addition to acting as the chromophore, retinal also stabilizes the ChR2 complex with respect to degradation . In some organisms, including mice, sufficient endogenous retinal is available, given a standard laboratory diet, for ChR2 to function. In C. elegans, retinal must be added to the E. coli food bacteria. This requirement for retinal supplementation provides a convenient way to control for an animal's intrinsic sensitivity to illumination: researchers can compare optogenetic experiments to otherwise equivalent ones performed in the absence of retinal. In addition, researchers can alter retinal concentration to manipulate the level of opsin activity.
Several laboratories have developed ChR2 variants with altered kinetics, conductance, optical spectra and other properties. When illuminated with blue light, wild-type ChR2 shows a large decrease in current within about 7 ms. The gain-of-function mutant ChR2(H134R) exhibits reduced inactivation and therefore higher steady-state currents relative to ChR2(wt)  and is thus the most commonly used ChR2 variant in C. elegans.
Upon cessation of illumination, wild-type ChR2 deactivates within about 20 ms. In some applications, a longer period of activation may be desired. The variants ChR2(C128X) deactivate with time constants varying between 2 and 100 s, depending on the specific mutation . Owing to their slow deactivation, these opsins are known as step function opsins (SFO).
In order to enable differential optogenetic stimulation of distinct cell populations, to monitor intracellular calcium activity in the presence or the absence of stimulation, or to intrinsic effects of blue-light stimulation [16,17], researchers have developed opsins with red-shifted activation spectra. These include the chimaeric channelrhodopsin C1V1 , VChR1 from the alga Volvox carteri , Chrimson  and many others . Red-shifted opsins also have the advantage of improved tissue penetration owing to reduced endogenous absorption.
(c) Opsins for inhibition
After the development of ChR2 for optical stimulation, complementary strategies for optical inhibition rapidly followed. The first of these was NpHR/Halo, a light-powered chloride pump from the archaebacterium Natronobacterium pharaonis [21,22]. Improved versions of this opsin addressed limitations related to trafficking of the opsin to the plasma membrane [23,24]
Chow, Han et al. used a cross-kingdom functional molecular screen to identify a number of light-driven proton pumps that can mediate neural silencing . These include yellow/green-light-drivable archaerhodopsin-3 (Arch) from Halorubrum sodomense and blue/green-light-drivable Mac from Leptosphaeria maculans.
ChR2, normally used as an excitatory opsin, can also play an inhibitory role in two ways. First, in addition to its capacity as a light-gated cation channel, ChR2 has an outward proton pump activity acting to hyperpolarize the cell . Under normal conditions the effect on the membrane potential of this pump activity is small compared with the cation-mediated depolarization effect.
Second, researchers applying ChR2 in mammalian systems have reported that repeated stimulation of cells can cause them to stop firing entirely, and therefore have the effect of inhibiting rather than exciting them [27,28]. This so-called depolarization block may result from the kinetics of ChR2 and/or disruption of ionic gradients during excessive stimuli. Depolarization block causes a transition between excitatory and inhibitory effects as a function of ChR2 stimulus time  and can confound interpretation of optogenetics experiments. Susceptibility of cells to depolarization block varies from one cell type to another and on temperature . While depolarization block has not yet been demonstrated in C. elegans, it is possible that it may occur under some conditions. It is therefore important for researchers to empirically determine the response of cells, circuits and/or behaviours to the specific stimulation parameters to ensure that illumination has the intended effect.
(d) Spectrally distinct combinations of opsins
The different spectral properties of various opsins make it possible to address different populations of cells in a spectrally dependent manner. For example, the Deisseroth and Boyden groups used blue- and green-light excitation of ChR2 and NpHR/Halo, respectively, to perform excitation and inhibition of body wall muscles in the same worms [21,22]. More recently, dual excitation bands with very little crosstalk have been reported using ChR2 and the red-shifted opsin Chrimson , or by combining Chrimson with blue-light-sensitive CoChR from Chloromonas oogama . Dual inhibition can be approached by pairing a blue-light-sensitive inhibitory opsin with a green- or yellow-light-sensitive inhibitory opsin.
(e) Optogenetics using other signalling pathways
Manipulation of membrane potential via light-activated ion channels and pumps has become nearly synonymous with the term optogenetics. However, this strategy is only one of many approaches for using light to change neuronal function. Here, we briefly review optogenetics methods for perturbing other signalling pathways.
Lima and Miesenbock developed a ‘phototrigger’ scheme in Drosophila based on a key-and-lock mechanism. In this method, researchers express the ionotropic purinoreceptor P2X2 or the capsaicin receptor TRPV1 in neurons, and add caged versions of their agonists. Upon illumination by a flash of light, agonists are activated, and then bind to the receptors, stimulating the cells .
Another class of optogenetic tools uses ligands attached to photoisomerizable azobenzene tethers, such that specific neurotransmitter receptors can be activated in response to light [9,10]. This system has the advantage of being able to target specific receptor types.
Cyclic AMP (cAMP) is a common second messenger in intracellular signalling; it is a component of G-protein signalling cascades, and it is involved in synaptic transmission. Photoactivated adenlyate cyclase alpha (PACα) from Euglena gracilis is a receptor flavoprotein that generates cAMP in response to blue-light illumination . When expressed in C. elegans motor neurons, PAC enhances neurotransmitter output and causes worms to increase in speed upon illumination . An infrared-sensitive adenylyl cyclase has also been developed and shown to be effective in C. elegans motor neurons .
(f) Genetic and spatial targeting
To express opsins in C. elegans, most researchers use standard methods for transgenesis . The choice of promoter determines the tissues expressing the opsin. In cases where no single promoter provides adequate specificity, researchers can apply intersectional promoter schemes based on recombinases FLP or Cre [34,35]. Another method for cell-specific transgene induction is to stimulate heat shock in single cells by a pulsed laser beam [36–38].
An alternative to the solely genetic approach to stimulating specific cells is to target the illumination itself to the desired cells or tissues. Light sources are patterned by a digital micromirror device or liquid crystal display and coupled onto worms through a microscope. Guo et al.  used such a system to perform simultaneous optogenetic illumination and imaging of neural activity by calcium-sensitive fluorescent proteins in immobilized worms. Trojanowski et al. [40,41] used a similar system to perform single-cell excitation and inhibition in the pharyngeal nervous system. Illumination of selected cells in freely behaving worms requires computer vision algorithms that infer the position of targeted neurons based on the outline of the animal. Several laboratories have used such tracking systems to target neurons in live behaving C. elegans [42–44].
(g) Intrinsic light sensitivity
Even without optogenetic manipulation, C. elegans is sensitive to light in several ways. Researchers should consider these intrinsic light sensitivities when designing and performing optogenetics experiments.
In response to illumination with ultraviolet (UV), violet or blue light, worms accelerate their locomotory rhythms, an effect mediated by LITE-1, a member of the gustatory receptor family, and dependent on cAMP and diacylglycerol signalling [16,17]. This behaviour is thought to help worms avoid potential damage and dehydration owing to sunlight exposure. Worms have good reason to avoid short-wavelength light: long-term illumination by bright UV, violet or blue light causes paralysis and, after approximately 30 min, death .
Bhatla & Horvitz  showed that bright light inhibits C. elegans feeding. Like the light-induced locomotory acceleration, this effect is mediated by LITE-1 and also another gustatory receptor analogue, GUR-3.
To control for intrinsic light sensitivity, experiments should incorporate control worms lacking retinal and/or lacking the opsin transgene. For cases in which the intrinsic behaviours cause problems for behavioural assays, researchers can perform optogenetics experiments in lite-1 mutants defective for endogenous light sensitivity. A recently published blue-light-sensitive opsin from Chloromonas oogama (CoChR) has a fivefold higher sensitivity than ChR2 . This allows CoChR activation at light intensities that do not trigger the intrinsic light response of C. elegans.
Illumination also causes some degree of warming in worms and their substrates owing to light absorption. Worms are highly sensitive to temperature and avoid warming at temperatures higher than that of their recent experience . Experimenters can test whether behavioural responses are thermal in nature by analysis of thermo-insensitive mutants.
3. Overview of the Caenorhabditis elegans nervous system
Caenorhabditis elegans has two nearly independent nervous systems: (i) the somatic nervous system containing 282 neurons (in the adult hermaphrodite) in 118 classes (figure 1) and (ii) the nervous system of the pharynx (feeding organ), containing 20 neurons in 14 classes . These two nervous systems are connected by a single electrical synaptic connection between the somatic RIP neurons and the pharyngeal I1 neurons. Each C. elegans neuron is identified by a 2- or 3-letter name and in some cases a number, often followed by one or more of L (left), R (right), D (dorsal) or V (ventral) to specify anatomical position.
The worm's neurons can be classified as sensory neurons, interneurons and motor neurons, based on their anatomical features and synaptic connectivity (figure 2). Sensory neurons are those that appear to have sensory endings, whether or not they have shown to be functional. Motor neurons are those that make synapses onto muscle cells. Interneurons are neurons that make many connections with other neurons . These labels are to some degree arbitrary, since many neurons span two or more categories. For example, the B-type excitatory motor neurons form synapses with several other neuron types and have been shown to have a proprioceptive sensory function .
4. Sensory systems
Optogenetics has played an important role in dissecting the physiology of C. elegans sensory systems. About 85 of the hermaphrodite worm's 302 neurons have been anatomically classified as sensory cells; some of these have been found to detect odors, tastes, temperature, mechanical stimuli and light. Laser killing and genetic experiments have been instrumental in dissecting sensory modalities in the worm. Optogenetics has provided a complementary approach to directly test the behavioural consequences of neuronal activation or inhibition.
Optogenetic activation of sensory neurons that mediate quick avoidance responses has been particularly robust in recapitulating the endogenous behaviour. Caenorhabditis elegans exhibits a stereotyped response to non-nociceptive touch to its body: touch to the anterior part of the body induces backward locomotion while touch to the posterior part of the body induces accelerated forward locomotion. Mechanosensation of gentle touch is mediated by six touch receptor neurons (TRNs) that extend long processes throughout different regions of the body . Subsets of the TRNs are necessary to generate these responses: the anterior touch response requires the ALM and AVM neurons, while the posterior touch response requires the PLM neurons. A functional role for PVM has not yet been demonstrated.
Selective optogenetic illumination of touch cells in freely moving worms showed that optogenetic activation of individual TRNs elicits avoidance responses indistinguishable from those observed with touch, demonstrating that depolarization of the TRNs is sufficient for the induction of avoidance response [42,43]. Furthermore, optogenetic inhibition of premotor interneurons that drive backward locomotion can reduce the probability of reversals induced by ALM and AVM activation .
The C. elegans mechanosensory neurons have served as a model for neuronal regeneration after injury. Axons of touch cells (and many other neuronal types), after being severed using a pulsed laser beam, are capable of spontaneous regrowth . Sun et al.  showed that periodic stimulation by ChR2 can enhance neuronal regeneration after axotomy, suggesting a novel strategy for beneficial neurotherapy .
Analysis of animals deficient in gentle touch response showed that animals that fail to respond to gentle touch can still respond to harsh (nociceptive) touch. Laser ablation studies identified the FLP and PVD neurons as the main sensors of harsh touch [53–56], with additional neurons (SDQ, BDU, ADE, AQR, PDE, PHA and PHB) contributing to the response . FLP and PVD are multi-dendritic neurons similar to mammalian nociceptors in both morphology and function. The analysis of harsh touch response proved difficult since harsh mechanical stimuli also activate the TRNs that mediate the gentle touch response. Optogenetics allowed the analysis of the harsh touch neurons and downstream circuits without interference by other mechanoreceptor cells. While photoactivation of the FLP initiated backward movement, activation of the PVD induced forward movement similar to the endogenous response to harsh touch [50,57]. The optogenetic activation of PVD paved the way for the identification of the DEG/ENaC and TRP channels required for its function .
The nociceptive ASH neurons detect several noxious stimuli and are required for avoidance responses to nose touch, high osmolarity and chemical repellents . Despite the advantages of targeted illumination in live behaving animals, the spatial resolution can still be limiting factor when no cell-specific promoters are available and optogenetic proteins are expressed in cells that are close together. An alternative approach uses FLP or Cre recombinase and combinatorial expression using promoters whose expression uniquely overlaps in the cell(s) of interest. Using this approach, it was shown that ASH photoactivation triggers withdrawal behaviours mimicking the endogenous response [50,59]. The strength of the optogenetic activation of the ASH neurons directly correlated with the magnitude of the behavioural response [50,60].
Sensory neurons required for chemosensation trigger compound behavioural responses over long time scales that allow the animal to move up or down chemical gradients. Caenorhabditis elegans will navigate up or down salt gradients towards salt concentrations corresponding to concentrations at which they were raised . The ASE neurons are the principal sensory neurons that detect salt concentration . ASER is depolarized by decreases in salt concentration, while ASEL responds to increases in concentration [63,64]. Optogenetic activation of the ASER neuron induced a transient increase in turning for animals on salt concentrations below the one they were raised at, but a decrease when the animal navigates above the one they were raised at. Thus, the ASER neuron mediates salt avoidance and attraction by regulating the frequency of reorientation movements in response to the gradient [61,65].
Caenorhabditis elegans prefers oxygen concentrations ranging from 5 to 12% and avoids habitats with low (less than 5%) or high (21%) oxygen concentrations. The principal oxygen-sensing neurons are the URX, AQR and PQR neurons, which have sensory endings in the animals' body fluid [66,67]. Optogenetic activation of the URX, AQR and PQR neurons increases locomotion, while inhibition of the same neurons strongly reduces movement, indicating that tonic activation of the oxygen-sensing neurons stimulates fast movement required for the escape of unfavourable oxygen concentrations . The primary oxygen sensors, the URX neurons, make gap junction and reciprocal connections with the RMG interneurons. Light-induced activation of the RMG inhibits spontaneous reversals and induces rapid forward movement even in animals in which the URX, AQR and PQR neurons were ablated , indicating that RMG activation is sufficient to induce an arousal state, probably through the activation of other downstream interneurons.
The combination of optogenetics with calcium imaging or electrophysiology in semi-constrained animals has been particularly powerful for unravelling sensorimotor circuits that mediate behavioural responses to stimuli. For instance, optogenetic activation of ASH induces strong calcium transients in the AVA and AVD premotor interneurons that drive backward locomotion. Electrophysiological recordings from AVA show an increase in excitatory currents as the strength of light-induced ASH activation increases . This indicates that the functional connection between ASH and AVA is graded such that stronger noxious stimuli elicit stronger escape responses. Similarly, optogenetic activation of the thermosensory AFD neurons combined with recordings of the downstream interneuron AIY reveal that AFD can tonically excite or inhibit AIY : inhibition of AFD activity using halorhodopsin leads to strong activation of AIY, while strong activation of AFD results in weaker activation of AIY. Therefore, the level of activation of AFD determines whether it evokes warm-seeking or cold-seeking behaviour .
These calcium imaging or electrophysiological experiments used semi-constrained animals, which may lack proper behaviour feedback mechanisms required for proper neuronal processing. Novel systems that combine targeted optogenetic stimulation with calcium imaging in unrestrained freely behaving animals have now been developed that circumvent this caveat. Using this technique, it was shown that optogenetic activation of the anterior TRNs induces calcium transients in the AVA premotor interneurons that correlate with the worm's backward velocity. Since the excitation spectra of ChR2 and GCaMP overlap, this combination can only be used if the neurons' bodies are sufficiently far apart. However, recently developed red-shifted ChR2 or RCaMP (a red fluorescent calcium reporter) variants with more distinct excitation spectra [18,19,72] may overcome these limitations.
In C. elegans, sensory information is transduced into behaviour by a shallow network of interneurons. Only two or three layers of interneurons separate sensory neurons from premotor interneurons and motor neurons that regulate motor output. In the case of mechanosensory avoidance responses, it has been possible to use laser ablation to establish the feed-forward pathways that connect sensory stimulation to movement . However, for complex navigational behaviours like chemotaxis and thermotaxis, it has been challenging to connect sensory neurons to the regulation of turning movements, forward movements and backward movements that organize behavioural strategies. Removing specific interneurons from the circuit may reduce the overall ability to effect chemotaxis and thermotaxis, but understanding the role of each interneuron in patterning the sensorimotor transformations that drive behaviour requires direct physiological analysis.
Optogenetics provides a means of acutely activating or inactivating interneurons during C. elegans navigation, and thus a means of assigning a role to each interneuron in mediating specific behavioural outputs. For example, the AVA premotor interneuron has been shown to be active during backward movement, and killing AVA reduces the spontaneous reversal rate during locomotion. Optogenetic activation of AVA can evoke backward movement. AVA has a number of gap junctions with the RIM interneuron. Guo et al.  showed that activating the RIM interneuron using cell-specific expression of ChR2 also evokes AVA activity and thereby triggers reversals. Taken together, these data show that RIM may be part of the circuit for initiating reversals. Interestingly, killing RIM has the opposite effect to killing AVA, causing animals to have a lower frequency of spontaneous reversals. One possibility is that RIM removal from the circuit by laser killing has more complex consequences than RIM inactivation. These results indicate that optogenetics can yield information that is complementary and distinct from traditional methods of circuit dissection such as ablation analysis.
In C. elegans, interneurons are highly interconnected with one another and with downstream circuit pathways. Thus, many effects might be evoked by optogenetic manipulation of single neurons. A striking example is the AIY interneuron, which is downstream of numerous chemosensory and thermosensory neurons. In a study of olfactory chemotaxis, Kocabas et al.  activated interneurons downstream of the AWC olfactory neuron with inputs that might resemble those experienced by an animal trying to navigate towards food. By stimulating or inhibiting the AIY interneuron every time the head undulated to one side, they showed that it is possible to steer the animal towards either the dorsal or ventral sides. This might be related to a ‘klinotactic’ mechanism, by which the animal augments the curvature of its movements in one direction depending on whether it encounters improving or declining conditions during a specific phase of its undulation cycle. Similar results were obtained in a study of salt chemotaxis using phasic optogenetic stimulation of AIY. Asymmetric activation of any neuron in a pathway downstream of AIY (AIY → AIZ → RIM → SMB → RME) also evoked steering . Thus, it is possible that a cascade of phasic activation of neurons from sensory neurons to motor neurons generates klinotactic steering.
Interestingly, activating or inactivating AIY using optogenetics can also lower or raise, respectively, the rate of reversals. The regulation of reversal rate by AIY appears to require its postsynaptic partner, AIZ. Activating AIY can also increase the speed of forward movement, but this effect requires a different postsynaptic partner, RIB . Taken together, these results suggest that single interneurons in C. elegans can have multiple roles in behavioural output, roles that are differentiated by the detailed activity pattern of the interneuron and the downstream pathways that are thereby activated.
Interneurons in C. elegans not only receive input from upstream sensory layers, but also receive feedback from downstream motor layers. This presents a challenge to interpreting the behavioural phenotypes that might be evoked by activating or inactivating individual neurons. The AIB interneuron is downstream of a number of sensory neurons including the AWC olfactory neuron. To understand how sensory response in the interneuron AIB is influenced by the overall network state, Gordus et al.  used optogenetic stimulation in combination with calcium imaging. Optogenetic activation of AVA, the premotor interneuron for reversals, by the red-light-sensitive cation channel was shown to spread to both AIB and RIM, influencing the response of these neurons to olfactory stimulation.
Another complexity in understanding interneuron processing in C. elegans is interpreting the effects of optogenetic perturbation in the context of behavioural state. A behavioural state of an animal is thought to involve the organized activity of many brain circuits towards a specific behavioural goal. As described below, it has been shown that C. elegans' behavioural state can be modulated by optogenetic stimulation of specific neurons. It has also been shown that how optogenetic manipulation modulates behaviour depends on behavioural state.
Caenorhabditis elegans alternates between two behavioural states during foraging. During roaming the animal has low reversal rates, while reversals increase during dwelling. Flavell et al.  showed that serotonergic neuronal activity underlies dwelling. Optogenetic activation or inhibition of serotonergic neurons, including HSN and NSM, upregulated and downregulated dwelling, respectively. Remarkably, these effects could also be mimicked by directly manipulating the activity of serotonergic targets that expressed the MOD-1 serotonin receptor (AIY, RIF and ASI). Optogenetic activation or inhibition of these downstream targets downregulated or upregulated dwelling, respectively. Analogous effects could be produced by optogenetic manipulation of a distinct circuit that regulates foraging state. The PDF neuropeptide promotes roaming state, and optogenetic activation of PDF receptor-expressing cells promotes the roaming state.
During larval development, the worm regularly enters lethargus, a quiescent sleep-like state during which much of the nervous system becomes inactive [76,77]. Switching between quiescence and wakefulness in C. elegans has been shown to involve the RIA and RIS interneurons that appear to play opposite roles. Ablation of RIA neurons promotes quiescence, whereas ablation of RIS disrupts quiescence. Interestingly, cell-specific optogenetic activation of RIA during quiescence can activate movement, akin to waking the worm up . Optogenetic activation of RIS promotes quiescence . Lethargus can also be interrupted by optogenetic activation of premotor interneurons that evoke reversals, but this effect requires the synchronous activation of several interneurons, e.g. AVA, AVD and AVE. Activation of just AVA does not evoke movement during lethargus . Thus, optogenetic manipulation of a core circuit for quiescence appears to switch the nervous system between a sleep-like state and wakefulness.
The worm also has distinct behavioural states depending on ambient oxygen levels. Worms avoid atmospheric oxygen levels (21%). Exposing worms to high oxygen causes an increase in speed and reduction in reversal rates that facilitate escape. The URX oxygen-sensing neurons and their synaptic partner RMG play a critical role in the entry to the arousal state upon exposure to high oxygen levels. Laurent et al.  showed that selective optogenetic activation of RMG suffices to trigger the arousal state. Interestingly, whether the animal is in the arousal state or not also affects the behavioural effects of optogenetic activation of other neurons that regulate movement patterns. For example, the AIA interneuron was shown to have higher calcium levels during forward than backward movement, whether the animal is exposed to high or low levels of oxygen. However, optogenetic inhibition of AIA has distinct effects at high and low oxygen levels. Optogenetic inhibition of AIA at high oxygen causes an increase in reversal rate, and optogenetic disinhibition at low oxygen causes a reduction in reversal rate.
The emerging picture from studies of interneurons using optogenetics is complex and subtle. The raising and lowering of activity in any one neuron can have multifarious effects on the nervous system and behaviour. To be useful, optogenetics must be judiciously combined with other powerful tools available in C. elegans including ablation analysis, genetics, imaging and behavioural analysis.
6. Motor neurons
The 113 cells classified as motor neurons in the adult hermaphrodite include excitatory cholinergic and inhibitory GABAergic neurons . The majority of motor neurons are located in the ventral nerve cord (VNC) and innervate the body wall muscles. Other motor neurons include those controlling head movements, pharyngeal muscles, defecation and the egg-laying musculature.
(a) Synaptic transmission
ChR2-mediated stimulation of motor neurons has been used in analyses of synaptic transmission at the neuromuscular junction (NMJ). Prior to the advent of optogenetic methods, NMJ physiology was studied by extracellular stimulation of the entire VNC during patch clamp recording from muscle cells. The non-specific nature of the stimulus made it impossible to discern roles for specific motor neuron types. Optogenetics has made it straightforward to stimulate specific neuronal types in combination with electrophysiological recording. Thus, optogenetics has added to the power of C. elegans to elucidate the molecular machinery of the synapse and synaptic plasticity.
Quantitative behavioural assays found a decrease in worm body length during stimulation of excitatory neurons, which induces contraction of body wall muscles (BWMs). Similarly, stimulation of GABAergic motor neurons induces muscle relaxation, which transiently increases body length. In dissected worms, electrophysiological recordings monitored the effect of optogenetic perturbation on neuromuscular currents. Using these assays, Liewald et al. examined mutants with defects in presynaptic and postsynaptic machinery . One surprising result was that stimulation of cholinergic neurons in presynaptic mutants led to contractions with larger amplitude than wild-type, potentially owing to compensatory effects in muscle cells. Kittelmann et al.  conducted a study of synaptic recovery after hyperstimulation of motor neurons (ChR2-mediated stimulation until paralysis occurred). Analysis revealed different time scales of recovery ranging from 8 to 60 s, which were related to endocytosis, scission and regeneration of synaptic vesicles. Optogenetic stimulation has also been used to study acetylcholine  and GABA receptors .
Weissenberger et al.  used PACα to increase the synaptic output of C. elegans motor neurons, causing an increased velocity of forward locomotion. The authors highlight this as an advantage of PACα over ChR2-mediated stimulation: whereas illumination of a ChR2-expressing cell causes it to be stimulated, illumination of PACα-expressing cell may cause an increase in the neuron's intrinsic output, and may therefore produce a more physiological output.
Optogenetics has been used to investigate signalling pathways that regulate synaptic plasticity. Jensen et al.  optically stimulated excitatory motor neurons in the VNC using ChR2 in combination with electrophysiological analysis of muscles and imaging fluorescent-tagged acetylcholine receptors. The authors demonstrated activity-dependent synaptic plasticity at the NMJ and showed that it depends on acetylcholine translocation mediated by Wnt signalling. These results showed that Wnt signalling regulates the strength of synaptic signalling. Hoerndii et al.  considered mechanisms of synaptic plasticity in AMPA receptors. The authors showed that repetitive optogenetic stimulation of selected synapses changed the localization of GFP-tagged AMPA receptors in a manner dependent on UNC-43, the homologue of Ca+2/calmodulin-dependent protein kinase II (CaMKII), an important mediator of learning and memory in vertebrates.
(b) Locomotory circuit function
Optogenetics has been helpful for understanding the neural basis of locomotion. Worms move forward by propagating dorsoventral bending waves from anterior to posterior. Navigation occurs in part via brief reversals during which the bending wave proceeds from posterior to anterior.
Motor neurons were hypothesized to be sensitive to stretch based on the presence of undifferentiated processes in each motor neuron ; this sensory feedback has been proposed to underlie the propagation of dorsoventral bending waves during locomotion via a feedback loop. Wen et al.  used static and dynamic microfluidic devices in which portions of the worm were forced to adopt a straight or curved posture to show that the locomotory bending wave indeed propagates via proprioceptive feedback. Using NpHR/Halo-mediated optogenetic inhibition, the authors then showed that this feedback is mediated by B-type cholinergic motor neurons during forward movement.
Caenorhabditis elegans' sinusoidal locomotion depends on smooth contraction and relaxation of its body wall muscles. For many years, it was thought that unlike vertebrate skeletal muscle, which exhibits action potentials, C. elegans body wall muscles have only graded potentials. However, in 2011 researchers reported that worm muscles do in fact have calcium-dependent action potentials , and that synchronization of action potentials depends on gap junction coupling between muscle cells . The authors used ChR2-mediated stimulation of excitatory cholinergic and inhibitory GABAergic motor neurons to generate specific neuromuscular inputs during electrophysiological recordings from body wall muscles.
Many aversive stimuli, including head touch, induce a response consisting of a brief reversal followed by a sharp turn during which the animal forms an ‘omega’ shape. Donnelly et al.  showed that a shift in the balance between ventral and dorsal GABAergic signalling contributes to the execution of a sharp omega turn. Optical activation of the DD motor neurons that innervate the dorsal muscles relaxes the dorsal muscles and induces a ventral body bend. During optical inhibition of the DD motor neurons, the dorsal muscles hypercontract, leading to a dorsal bend. This work was the first to dissect in detail the neural basis of a stereotyped sequential behaviour.
These studies represent important steps towards an integrative understanding of how the locomotory neural circuit coordinates behaviour.
(c) Other motor neurons
The paired hermaphrodite-specific neurons HSN are located near the vulva and innervate muscles involved in egg-laying behaviour. ChR2-mediated stimulation of HSN, which induces egg laying, has been used to identify roles of potassium channels in regulating HSN excitability [42,90].
Defecation in C. elegans is a rhythmic behaviour with a period of roughly 60 s, mediated by the intestinal cells, body wall muscles and enteric muscles. Mahoney et al.  showed that ChR2-mediated stimulation of GABAergic motor neurons rescued expulsive-defective mutants. Wang & Sieburth  used a similar protocol to demonstrate that GABA neuron stimulation rescued expulsion defects in nlp-40 neuropeptide mutants but not in mutants lacking exp-1, an excitatory GABA receptor on enteric muscles. These results showed that rhythmic neuropeptide release from pacemaker cells to downstream neurons can execute rhythmic behaviours.
The body wall musculature of C. elegans consists of 95 cells in four longitudinal rows . These muscles are involved in essentially all worm behaviours except for feeding. Stimulation of all BWMs causes a temporary constricted paralysis and a decrease in body length; conversely, inhibition of BWMs causes flaccid paralysis and an increase in body length. Researchers have used optogenetic stimulation and inhibition of the body wall muscles to test opsins [6,21], probe synaptic function (see above)  and discern the influence of the worm's muscle tone on its biomechanical properties .
8. Pharyngeal neurons
So far we have considered only the somatic nervous system of C. elegans. Optogenetics has also been very fruitful in analyses of the much smaller nervous system of the pharynx, which is responsible for feeding behaviour.
Trojanowski et al.  examined the pathways for excitation of the pharyngeal muscle. Previous work using laser ablation  demonstrated that the motor neuron MC is the only pharyngeal neuron required for rapid pumping. Using a system for cell-specific optogenetic illumination , the authors showed that not only the MC but also the M2 and M4 neurons directly stimulate pharyngeal pumping, and that the MC neurons act via both nicotinic and muscarinic receptors . In addition, the authors showed that the I1 interneuron stimulates pharyngeal pumping via both MC and M2. Complementary experiments using optical inhibition via Mac yielded opposite results, showing that the identified roles of these neurons are physiological. Inhibition of pharyngeal pumping during I1 inhibition is surprising, since laser ablation of I1 has no effect on pumping rate in the presence of bacterial food [94,95], highlighting the limitations of laser ablation approaches for understanding circuit function. These results showed how degenerate neural pathways, in which multiple elements can perform the same function, regulate behaviour in a simple model system.
Dillon et al.  used ChR2-mediated excitation of glutamatergic pharyngeal neurons to investigate the function of metabotropic glutamate receptors (mGluRs). The authors found that glutamate receptors are required for presynaptic modulation of pharyngeal behaviour and electrical activity. These findings highlight the essential role of mGluRs in regulating complex behaviours in C. elegans, as they do in vertebrates.
In an analysis of food-dependent locomotory states, Flavell et al.  showed that ChR2-mediated stimulation of the serotonergic pharyngeal sensory/interneuron NSM induced dwelling behaviour (characterized by a high turn rate and low velocity) and produced an effect mimicking the presence of food. These results showed how a slow neuromodulatory circuit can supplement fast motor circuits to organize an animal's behavioural states.
Optogenetic analysis has become an indispensable tool for dissecting the neural basis of behaviour in C. elegans. Researchers are now able to control individual cell types from sensory neurons to interneurons to motor neurons to muscle cells, and quantify the effects of raising or lowering the activity of individual cells on intact circuits and behaviour. The temporal precision and cellular resolution of optogenetics in C. elegans have allowed the dissection of a variety of behaviourally relevant computations in the nematode nervous system. The rapid expansion of the optogenetics toolkit—more light-sensitive reagents with different spectral and temporal characteristics—is certain to augment the utility of optogenetics in worm neuroscience.
As with any experimental method, the limitations and caveats of optogenetics must be understood to properly design or interpret experiments. Because the degree of optically mediated excitation or inhibition in any particular cell is unknown, it is difficult to know how any particular manipulation compares to physiologically relevant patterns of activity. Optogenetic manipulations must be coupled to other methods in C. elegans, such as calcium imaging, genetic manipulation and quantitative behavioural analysis, to properly gauge their effects. Manipulating neuronal activity with light, while perhaps minimally invasive, is not non-invasive. Any transgenic manipulation can disrupt the properties of a cell in unpredictable ways, and intrinsic light sensitivity may be compounded with the optogenetic perturbation.
In most cases, optogenetics provides new ways to test models of circuit function. Because optogenetic perturbations will rarely recapitulate the normal activity patterns of circuits, they must be interpreted with care. That said, optogenetics often provides the most convenient and direct way of raising or lowering the activity level of any cell type to assess their functional role. Optogenetics is an essential part of the toolbox throughout C. elegans neuroscience, and its importance will only grow as these tools become better understood and more widely used.
We declare we have no competing interests.
C.F.-Y. is supported by the Alfred P. Sloan Foundation, Ellison Medical Foundation, and the National Institutes of Health. M.J.A. is supported by the National Institutes of Health (GM084491). A.D.T.S. is supported by the National Institutes of Health, National Science Foundation and the Human Frontier Science Program.
The authors thank Nicholas Trojanowski and Ni Ji for helpful comments.
One contribution of 15 to a theme issue ‘Controlling brain activity to alter perception, behaviour and society’.
- Accepted June 16, 2015.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.