Neuroengineering with photostimulation
One common engineering technique to understand how a system works is to tinker with individual parts and see how that one change affects the system as a whole. The artificial stimulation of individual neurons in primates has been technologically feasible for some time (i.e., see Salzman et al, 1990), but the invasiveness of those electrode-based techniques stymied progress. Recent advances in genetics have allowed for the possibility of expressing invertebrate proteins ectopically in vertebrate cells. In 2001, Zemelman et al attempted to program cells with the mRNA that encodes for the proteins with defined roles in fruit fly phototransduction. After they transfected the three proteins identified as absolutely necessary for the job, hippocampal neuron cultures exposed to white light produced action potentials. However, the latency between the light exposure and the action potentials was too troublesomely variable, ranging from a few hundred milliseconds to several tens of seconds. Further work in their lab (Zemelman et al, 2003) improved upon their initial technique by heterologously expressing ligand gated ion channels (either P2X2 or TRPV1) to sensitize the cell to stimulation. This method works through a photochemical lock and key mechanism. The agonists that typically gate the ion channels are rendered biologically inert by attaching chemical blocking groups which sterically prohibit binding to the ligand. These blocking groups are photoremovable upon absorption of a photon in the UV-light range, which frees the agonist to bind with the appropriate receptor and depolarize the cell. Their improved method had the advantage of coupling the kinetics of the neural responses more tightly to the stimulus, reducing the latency variability of their first attempt. It also allowed the frequency of the firing rate to be altered by the concentration of the agonist.
Other labs soon discovered other photoreceptors that functioned in vitro when expressed ectopically. In 2003, Nagel et al heterogously expressed channelrhodopsin-2 (ChR2), a sensory photoreceptor that orients the swimming movements of the green algae Chlamydomonas reinhardtii, in neurons. When the rhodopsin absorbs a photon (with a maximum wavelength of 460 nm), it isomerizes the all-trans-retinal to the 13-cis-retinal, which directly opens the cation channel. As a proof of principle, the researchers expressed ChR2 in human embryonic kidney cells in vitro. These cells were consistently depolarized within 1-3 milliseconds upon illumination with blue light. One of the reasons for its rapid depolarization of the cell is that it does not act through coupling to a signal-accelerator protein like a G-protein, which is the norm among most rhodopsins in other vertebrates (Sakmar et al, 2002). Another reason that ChR2 has caught on as the optical method of choice is that, unlike other techniques, it does not require the addition of exogenous chemicals in order to depolarize the cell in mammals, because sufficient baseline levels of all-trans-retinal are already available (Zhang et al, 2006).
Although the in vitro work was exciting, the next logical step was to determine how the selective activation of neurons could be accomplished in vivo. In 2005, Lima and Miesenbock published their work on this task in the model system Drosophila. They prepared transgenic flies that expressed the ligand gated ion channel PX2 and injected blocking group binded ATP into the brain 10-60 minutes before analysis, using the lock and key strategy described above. They targeted the giant fiber neuronal system, which is responsible for escape mechanisms such as jumping and rapid flight, because these stereotyped behaviors would be easy to detect. UV illumination of 150-250 milliseconds in flies expressing PX2 in one restricted set of neurons elicited the expected escape behavior 82% of the time. To demonstrate that visual signals were not responsible for the activation of the escape responses, they blinded their flies, and the effects of direct photostimulation remained the same. They had accomplished the remote control of behavior by the manipulation of neurons.
Following these basic proofs of principle, research into the field has exploded and these original techniques have been iterated upon in novel and creative ways. Banghart et al (2004) tethered a photoswitchable chemical blocker to the Shaker potassium channel, the activation of which can hyperporalize the cell and thus silence neural activity. Yoshimura et al (2005) used the photostimulated uncaging of glutamate molecules to generate action potentials in small populations of neurons. But while the methods in and of themselves are interesting, it is the applications that will advance our fundamental understanding of the brain. This paper will briefly review three of these applications: characterizing the role of specific neurons, identifying the circuits responsible for behavior, and enhancing the methods of operant conditioning.
Characterizing the role of specific neurons
Ever since Santiago Ramon y Cajal discovered the structure of the neuron, scientists have attempted to classify them into discrete groups. Camillo Golgi grouped them into neurons with axons and neurons without axons, a dichotomy that is still in use. Today, classifying all neuronal cell types is a major goal in neuroscience, especially among those in circuit neuroscience attempting to reverse engineer the brain (Yuste, 2008). Genetic techniques can silence neuron types and attempt to determine the structure, but these work on the scale of hours to days. It is difficult to isolate the effect of a neuron without temporal precision, and since optogenetics can stimulate or inhibit a neuron’s activity on the scale of milliseconds, it is uniquely suited to this task. There are already some examples where researchers have exploited this new tool to characterize the function of specific groups of neurons.
For example, Hwang et al (2007) used optogenetic techniques to pinpoint the role of multidendritic (MD) sensory neurons in fruit flies. The morphology of these neurons resembles that of vertebrate nociceptors, and larvae whose MD neurons have been silenced do not respond to noxious stimuli, suggesting that some classes of these neurons are responsible for nocioception. But they may be responsible for other behaviors as well. For example, previous work with genetic silencing of class I MD neurons found that they were responsible for certain kinds of muscle contraction during normal larval locomotion. By genetically reducing the synaptic vesicle release of select classes of MD neurons, the researchers determined that blocking the output of class IV neurons alone eliminated pain-related defensive behavior. But that does not prove that these neurons are nocioreceptors. Instead, they could provide propioreceptive feedback necessary for control of the defensive behavior. To isolate the causal effects, they generated flies that express ChR2 in neurons of various classes. Light-stimulated activation of all classes of neurons at once, or stimulation of class I, II, and III neurons individually, mostly just caused the contraction of all muscles in the body. But light stimulation of class IV neurons caused the defensive pain-related behavior only, at high speeds of consistently less than 100 milliseconds, which is faster than the noxious stimulus itself. This data allowed the authors to conclude with authority that class IV neurons are in fact nocioreceptors.
Other researchers have use optogenetics or hybrid approaches to explore the function of individual cells in the brain. Adamantidis et al (2007) expressed ChR2 in in hypocretin producing neurons of the lateral hypothalamus of mice, the loss of which had been previously associated with narcolepsy. They coupled an optical fibre with a diode laser and surgically inserted it in a small tube to deliver light stimulation to neurons in the lateral hypothalamus. Specific activation of hypocretin neurons at frequencies of 5 HZ and above led to an increased probability of waking from both slow wave and REM sleep, suggesting a causal link between action potentials in hypocretin neurons and sleep-to-wake transitions. In another demonstration of this capability, Nagel et al (2005) expressed ChR2 in the mechanosensory neurons ALM, PLM, AVM, and PVM of C. elegans. When stimulated with blue light for one second, transgenic nematodes raised in the presence of all-trans-retinal performed the expected withdrawal behavior 72% of the time on the first trial as opposed to 10% for the control animals, a significant increase. Thus the authors were able to show that these neurons probabilistically determined behavior, indicating the utility of optogenetics in studying neuronal function even in basic model systems such as C. elegans.
Identifying the role of specific neurons is a tricky business, and as Hwang et al show, researchers have to consider all of the strategies at their disposal in order to account for the random noise and plasticity of the brain. Optogenetics represents a powerful new part of that toolbox.
Identifying neural circuits responsible for behavior
Photostimulation techniques also have the ability to examine the role and mechanism of neurons that are functionally connected yet located in different regions of the brain. Indeed, one of the advantages that optogenetics has over methods like voltage clamping with electrodes is that it can simultaneously activate populations of neurons that all express the same photoactive protein. This allows researchers to focus on the big picture while controlling for the individual variation of neuron output. Additionally, the ability to accurately control the activation of neurons and stop stimulating them when necessary allows for novel explorations of connectivity. For example, photostimulating neurons in one brain region and measuring the electrical activity of anatomically distant neurons would allow researchers to see if the measured neurons are connected downstream and/or in the same neural circuit of the stimulated ones. Scientists have begun to apply this technology to various avenues of inquiry.
In one example of this application, Douglass et al (2008) set out to characterize the stimuli encoding patterns of the sensory pathway in developing zebrafish. They expressed ChR2 in two somatosensory neurons, the Rohon-Beard and trigeminal cells. Upon photostimulation their zebrafish elicited an escape response, qualitatively similar to the typical response following tactile stimulation, both quickly (less than 30 milliseconds) and consistently (arithmetic mean 79% +/- 4% of the time). When they restricted stimulation to single ChR2 expressing neurons within these populations, light illumination of 11 out of 33 the neurons triggered a muscle contraction capable of driving a full escape behavior on their own. Finally, provoking only one action potential in trigeminal cells with ChR2 caused an escape response, which the authors claim is the first documentation in any animal that single spikes in primary sensory neurons can directly elicit behavior. Moreover, the researchers speculate that single action potentials in trigeminal cells may be translated to a downstream Mauther cell, which has also been shown to drive escape behavior through a single action potential, although that was in goldfish (Nissanov et al, 1990). By enabling the hypothesize-and-test model of reverse engineering, optogenetics allows insight into the individual neurons responsible for escape behavior in zebrafish.
In another example of this novel strategy, Zhang and Oertner (2007) show that optogenetics is able to induce long-term potentation (LTP) of synaptic connections in CA1. After inducing LTP with paired photostimulated postsynaptic excitations, the excitatory current increased to 168% +/- 29% of the baseline amplitude, reproducing the results of other LTP experiments with connected pairs of pyramidal cells. Although their results were as expected, it is further indication of the potential for investigating neural circuitry through optogenetics. Recently, Petreanu et al (2009) utilized ChR2 and photostimulation as part of an in vitro method to characterize the circuitry of the barrel cortex in mice. In the case of exploring neural circuitry, optogenetic techniques have not allowed for many approaches that would have otherwise been technologically infeasible, but they have made those approaches more versatile and convenient. The technology is still young, so it will be up to other enterprising neuroscientists to make the most of its capabilities.
The holy grail of operant conditioning
The psychological literature on operant conditioning is extensive, and relies on reinforcement or punishment through external stimuli to shape an animal’s behavior. But theoretically, it should be possible to bypass the external reward and instead reinforce behavior through the direct stimulation of neurons in the reward centers of the brain. Such an approach would require temporally precise stimulation of neurons in order to ensure the proper contingency between stimulus and reward. Since ChR2 works on a scale of milliseconds, it is quite capable of accomplishing such a feat.
Airan et al (2009) displayed the power of optogenetics to control mice’s movement during an open field test. They developed chemeric proteins that express normal rhodopsin extracellularly but replaced the intracellular loops with specific G-protein coupled receptors that activate various signal transduction pathways. When they genetically replaced rhodopsin’s intracellular loops with the human alpha-1 adrenergic receptor, it led to a significant upregulation of inositol trisphosphate (IP3). Light stimulation of this same receptor complex increased network activity in vivo in transduced nucleus accumbens cells in mice, which led the researchers to hypothesize its activation could be used to shape reward-dependent behavior. To test this assertion, they used optical stimulation as part of a three-day conditioned place preference assay. On day one they collected baseline data for chamber preference. On day two they delivered light pulses with a laser diode coupled optic fiber at 10 HZ to accumbens neurons when the mice entered the designated chambers to simulate a strong reward. Finally, on day three they blindly scored the time spent in each chamber, and there was a significant increase in preference for the rewarded chamber in optogenetically stimulated mice, but not in controls, p < 0.01. This result shows that mice do not discriminate between actual reward and reward circuit activation, which presumably will be true across species. Needless to say, the ability to directly activate reward circuits opens up many doors for future research in operant conditioning.
Yet even this technology can be improved upon. Hira et al (2009) have developed a technique to make mice skulls transparent by removing the overlying skin and treating it with acrylic resin, through which light can stimulate ChR2 expressing neurons in the motor cortex. This allows researchers to vary the spatiotemporal pattern of neuron activation in freely moving mice, because they need not surgically implant an optic fiber. Theoretically, this would allow researchers to elicit one behavior through stimulation of one set of neurons while simultaneously rewarding that behavior through the stimulation of another set of neurons, in what can only be described as the holy grail of operant conditioning.
Optogenetics is currently one of the most exciting areas of neuroscience and behavior research. In this paper I have attempted to emphasize the ideas that it forces us to consider as opposed to the specific technologies involved, because those will probably change and be iterated upon as the field advances. It is possible that optogenetics will eventually have some clinical applications. Bi et al (2006) recently showed that expressing ChR2 in mice with retinal degeneration restored its ability to transduce light and send the information to the visual cortex. Additionally, it has been demonstrated that stable and functional expression of ChR2 in mammalian systems can last up to one year (Zhang et al, 2006). But that timescale would probably have to be extended in order to be of clinical use, and overall the immediate clinical prospects of optogenetics are not overwhelming. There are also potential complications, as mouse and human immune systems are different, so the foreign opsin proteins that do not adversely affect mice might harm humans. Finally, there are even some ethical concerns concerning the direct control of human behavior.
However, the relative lack of promise in clinical applications is of little concern because there is much to learn via animal models. By increasing our knowledge about the brain and the proteins that it is made of and the genes that code for those proteins, we will be able to create new technologies, which will feed forward into more knowledge. For example, it is possible that the optogenetic activation of Class IV MD neurons, exposed to be nocioreceptors by Hwang et al, could be used as a substitute for the unconditioned stimulus in an olfactory operant conditioning paradigm. Research into neuroscience and behavior is one of the most exciting fields in contemporary science, and optogenetics is one of the most new technologies available to neuroscientists. Although an advance in research capabilities alone may not seem of tremendous import to a layman, those in the field understand that it is from this basic research that we are able to formulate theories of the brain and develop the clinical advances that will affect all of our lives.
Nagel G, Szellas T, Huhn W, Kateriya S, Adeishvili N, Berthold P, Ollig D, Hegemann P, Bamber E. 2003 Channelrhodopsin-2, a directly light-gated cation-selective membrane channel. Proc Natl Acad Sci USA 100:13940-13945.
Salzman CD, Britten KH, Newsome WT. 1990 Cortical microstimulation influences perceptual judgements of motion direction. Nature 346: 174-177.
Zemelman BV, Lee GA, Ng M, Miesenbock G. 2001 Selective Photostimulation of Genetically ChARGed Neurons. Neuron 33:15-22.
Zemelman BV, Nesnas N, Lee GA, Miesenbock G. 2003 Photochemical gating of heterologous ion channels: Remote control over genetically designated populations of neurons. PNAS 100:1352-1357.
Sakmar TP, Menon ST, Marin EP, Awad ES. 2002 Rhodopsin: insights from recent structural studies. Annu Rev Biopys Biomol Struct 31:443-84.
Lima SQ, Miesenbock G. 2005 Remote Control of Behavior through Genetically Targeted Photostimulation of Neurons. Cell 121:141-152.
Nissanov J, Eaton RC, DiDomenico R. 1990 The motor output of the Mauthner cell, a reticulospinal command neuron. Brain Research 517:88-98.
Douglass AD, Kraves S, Deisseroth JJ, Schier AF, Engert F. 2008 Escape Behavior Elicited by Single, Channelrhodopsin-2-Evoked Spikes in Zebrafish Somatosensory Neurons. Current Biology 18:1133-1137.
Petreanu L, Mao T, Sternson SM, Svoboda K. 2009 The subcellular organization of neocortical excitatory connections. Nature 457:1142-1145.
Adamantidis AR, Zhang F, Aravanis AM, Deisseroth K, de Lecea L. 2007 Neural substrates of awakening probed with opteogenetic control of hypocretin neurons. Nature 450:420-424.
Banghart M, Borges K, Isacoff E, Trauner D, Kramer RH. 2004 Light-activated ion channels for remote control of neuronal firing. Nature Neuroscience 7:1381–1386.
Zhang F, Wang LP, Boyden ES, Deisseroth K. 2006 Channelrhodopsin-2 and optical control of excitable cells. Nature Methods 3:785–792.
Bi A, Cui J, Ma Y, Olshevskaya E, Pu M, Dizhoor A, Pan Z. 2006 Ectopic expression of a microbial-type rhodopsin restores visual responses in mice with photoreceptor degeneration. Neuron 50:23-33.
Airan RD, Thompson KR, Fenno LE, Bernstein H, Deisseroth K. 2009 Temporally precise in vivo control of intracellular signalling. Nature 458:1025-1029.
Yuste R. 2008 Circuit neuroscience: the road ahead. Fronteirs in Neuroscience 2:6-9.
Nagel G, Brauner M, Liewald JF, Adeisvhili N, Mamberg E, Gottschalk A. 2005 Light activation of channelrhodopsin-2 in excitable cells of Caenorhabditis elegans Triggers Rapid Behavioral Responses. Current Biology 15:2279-2284.
Miller G. 2006 Optogenetics: Shining new light on neural circuits. Science 314:1674–1676.
Hira R, Honkura N, Noguchi J, Maruyama Y, Augustine GJj, Kasai H, Matsuzaki M. 2009 Transcranial optogenetic stimulation for functional mapping of the motor cortex. Journal of Neuroscience Methods 179:258-263.