A guest post from Hugh Pastoll, my good friend and long-time intellectual sparring partner. Hugh's introduction follows, and then his article.
Michael and I met while studying PPE at the University of Cape Town. Like him, I’ve completely changed direction since then and am now doing a PhD in Computational Neuroscience at the University of Edinburgh.
As part of my postgraduate studies I’ve been fortunate enough to use an exciting new and truly revolutionary technology known as optogenetics. Optogenetics permits fine-grained control of brain activity with light, dramatically increasing the range of interesting experiments we can do. Since it is likely that it will soon become the technology of choice for investigating brain function, Michael has invited me to give a short primer on optogenetics in general and channelrhodopsins in particular.
As a computational neuroscientist I am ultimately motivated by understanding how neural activity determines behavior. Frustratingly, for a long time even attempting to answer this sort of question has been pretty much impossible. This has been a major barrier to understanding how brains work… until recently.
To see why we have been stuck, imagine that I want to test the hypothesis that some pattern of neural activity causes a particular behavior. In order to test this hypothesis I’d need to conduct an experiment where I manipulated the animal’s neural activity and observed its behavior (simply noticing that the pattern and behavior both occur when I give the animal a stimulus only establishes correlation, not causality). Now, we’ve been able to control neural activity for quite a while - the sticking point was that we weren’t able to do it with the millisecond fidelity, neuron type specificity and sub-cubic-millimeter spatial precision we need to test most of our important hypotheses.
To illustrate, say I hypothesize that synchronized firing of excitatory neurons in the subthalamic nucleus at 20 Hz is responsible for akinesia (deficit in movement initiation). Testing this typical hypothesis would require me to synchronize only sub-thalamic excitatory neurons without changing their overall firing rate or affecting activity in nearby brain areas while the animal is behaving. I can’t think of any way we would have to able to accomplish this with drugs, electrical stimulation or any other standard technique for controlling neural activity.
Thanks to the recent development of optogenetics, though, such control is not only possible, but relatively easy. I can’t really exaggerate how completely cool this is - it is going to allow the field of computational neuroscience to hit its stride and start delivering the kinds of insights we need to understand what’s really going on in the brain.
So how does optogenetics work? To understand this, you need to know how ion channels control action potentials in neurons. Very briefly, ion channels are specialized protein channels that, when open, conduct ions (charged molecules) across cell membranes. The brief rise in membrane potential during an action potential is due to positive ions rapidly moving from the outside to the inside of a cell. Channelrhodopsins are ion channels that open when you shine blue light on them! This means we can force the membrane potential of a neuron to become more positive and generate an action potential. This is the ‘opto’ part of optogenetics.
Channelrhodopsin-2 (ChR2 - the most useful original kind) was first described in a species of green algae called Chlamydomonas reinhardtii in 2002 and found to work in mammalian neurons. Since then, genetic engineers have found that strategically mutating different amino acids changes the kinetics of the channel (how quickly it opens and closes). So, now there are different versions that allow different types of control. The fastest type (named ChETA ) opens in about 2 milliseconds and closes after about 5ms; fast enough to pulse blue light at 200 Hz and have the neuron fire at virtually every pulse. Another type (ChR2-C128S) usually takes minutes to close but shuts off very quickly if you shine green light on it. This means it can act as a kind of bi-stable on-off neuron switch. With such fine-grained control we can manipulate neuron spiking in pretty much any way we like.
Now for the ‘genetic’ part of optogenetics: Different kinds of neurons make different kinds of proteins. Since channelrhodopsin is a protein, we can use the cellular machinery that determines whether a protein is expressed in a neuron to restrict channelrhodopsin expression to a specific type of neuron.
This allows us to make one type of neuron in an area fire, without directly disrupting the normal activity of other types in the same area, giving us the neuron sub-type specificity we need for our experiments.
Furthermore, we can restrict channelrhodopsin expression to a very small area of the brain. Since we know that genes code for proteins, if only cells in one area have the channelrhodopsin gene only those cells in that area will respond to light. We can accomplish this by infecting a group of neurons with a non-replicating retrovirus that carries the channelrhodopsin gene. This gene will then be integrated into the genome of the infected neurons and expressed, introducing channelrhodopsins with spatial specificity.
However, although this combination of temporal, neuron sub-type and spatial specificity will enable a wide range of experiments, even more is possible. Another class of membrane proteins, known as halorhodopsins, have the opposite effect to channelrhodopsins. Halorhodopsins are not passive channels - they actively pump negative ions into a cell when illuminated with yellow light, making it more negative and stopping it from firing. Additionally, proteins that pump positive hydrogen ions out of cells to make their interior more negative have been described recently. These proteins are more effective than some types of halorhodopsins at preventing neurons from firing and different types respond to a different light colors – allowing researchers to pick colors that don't interfere with other rhodopsins the animal may also be expressing.
With such powerful optogenetic tools at our disposal we can imagine performing complex experiments, orchestrating neural activity with an array of different color intra-cranial LEDs. Although such experiments will be technically challenging, at the moment it feels like we are only limited by our imagination.
Nagel, G. et al. (2002) "Channelrhodopsin-2, a directly light-gated cation-selective membrane channel," PNAS, doi:10.1073/pnas.193619210.
Boyden, E. et al. (2005) "Millisecond-timescale genetically targeted optical control of neural activity," Nature Neuroscience, doi:10.1038/nn1525
Berndt, A. et al. (2008) "Bi-stable neural state switches," Nature Neuroscience, doi:10.1038/nn.2247
Gradinaru, V. et al. (2009) "Optical deconstruction of Parkinsonian neural circuitry," Science, doi:10.1126/science.1167093
Chow, B. et al. (2010) "High-performance genetically targetable optical neural silencing by light-driven proton pumps," Nature, doi:10.1038/nature08652
Gunaydin, L. et al. (2010) "Ultrafast optogenetic control," Nature Neuroscience, doi:10.1038/nn.2495