Monday, September 2, 2013

Why optogenetics deserves the hype

Optogenetics has come in for some stick lately, with a number of people criticising the hype that this technique generates in some quarters. That’s fair enough, I suppose – there have no doubt been some claims made about what can be accomplished with this technique that are, at the very least, premature. I’m all for bashing hype (see The Trouble with Epigenetics 1 and 2, for example), but criticising the technique for what it’s not good for seems to be missing the point to me.

To me, optogenetics will revolutionise neuroscience. It is the tool that will finally let us meaningfully integrate the cellular with the systems level. Not by itself, of course – we’ll still need all the electrophysiology and pharmacology and neuroimaging and lesion studies and model organisms and whatever you’re having yourself. And not without some teething problems and over-interpretation of early findings, which will no doubt earn more tongue-lashings from the hype-police. But it will let us ask questions we have not been able to ask before – the right questions, at the level of cell types, the fundamental functional units of the nervous system.

Before I go on, a brief primer on how optogenetics works: this technique takes advantage of a number of proteins found in various species of algae that respond to light of certain wavelengths by opening a channel in their cell membrane to allow electrochemical ions (like sodium or chloride) to flow in or out of the cell. Controlling the flow of such ions along their fibres is also how neurons conduct electricity. If you take the gene that encodes the light-sensitive channel from algae and force neurons to express it, then they will become responsive to light – if you shine a light on them they will “fire” an electrical signal, or “action potential”. If you turn the light off, they will stop firing action potentials. And if you use a different channel protein, you can silence the neurons and stop them firing action potentials. This gives very tight, reversible control over the activity patterns of the neurons expressing these channel proteins (called channelrhodopsins).

The trick, and the power of the technique, comes from the specificity with which you can direct that expression. This is based on the fact that different types of cells express different sets of genes. All genes have two main parts – one part is basically the recipe or code for a particular protein. The other part, which is encoded on a neighbouring piece of DNA, is the regulatory region – the instructions for when and where to make that protein and how much to make. Those two regions can be separated. You can cut out the DNA that makes up just the regulatory piece of one gene and hook it up to the protein-coding region for any other gene you like (in this case, a channelrhodopsin protein). Now you can take that fusion gene and introduce it to cells or transgenically introduce it to animals, like worms or flies or mice. Such animals will now express channelrhodopsin only in the cell types directed by the regulatory piece of DNA you chose. A variety of other molecular methods can also be used to achieve this goal, including resources based on binary systems like the Cre-LoxP recombinase system. (Fibre optics can then be used to target light to those cells in particular brain regions).

So how many different cell types are we talking about? Going by the kinds of animations common in science fiction movies, many people apparently think of the inside of the brain as a network of effectively identical cells, randomly placed in a sponge-like layout, connecting simply to their nearest neighbours. Nothing could be further from the truth. We have known since the time of Ramon y Cajal and Golgi that there are many distinct types of neurons, which are distributed in a highly organised fashion in different brain regions and interconnected with exquisite specificity. And when I say many, I mean many hundreds, possibly thousands of types.

The retina alone has over 60 distinct, recognised neuronal cell types and more subtypes are being defined all the time. Those 60 cell types are arranged in four or five distinct layers, with multiple subtypes in each layer. There are at least a dozen parallel pathways across these several layers, processing various aspects of the visual stimulus – colour, form, direction, motion and many others. If you want to understand how the retina works – to reverse engineer it – you need to know what the functions of these cell types are within the context of the circuit in which they are embedded.

The importance of cell types as functional classes is blindingly obvious for the retina, but the same principle applies to any area of the brain. Subsets of cells in any area not only have discrete jobs to do within that area, making unique contributions to the computations carried out there, they also often connect in distinct, cell-type-specific, parallel circuits with other brain areas.

In the cerebral cortex, different excitatory cell types are arranged into six obvious layers, but these often have several sublayers. And within each layer, there are multiple subtypes of excitatory neuron, intermingled. In layer 5, for example, some neurons project across the corpus callosum to the other hemisphere, some within the cortex on their own side and others to subcortical targets. Each of these types contains multiple subclasses carrying information to distinct targets. That cellular complexity is multiplied by the number of cortical areas – subcortically-projecting layer 5 neurons in motor cortex are molecularly distinct from those in visual cortex, for example.

And we haven’t even started on the interneurons. These are smaller, more locally projecting cells, which are inhibitory – they put the brakes on excitation in neural circuits. They not only prevent runaway excitation, but also, crucially, control many aspects of information processing, such as filtering, gain control and temporal and spatial integration. In addition, they orchestrate the synchronous firing of ensembles of excitatory neurons, which in turn is a central mechanism in mediating communication between brain areas. Just in the hippocampus, there are twenty-some subtypes of interneurons already known, and, again, more are being defined all the time. Each of these subtypes is distributed in a particular manner, expresses different kinds of ion channels and neurotransmitter and neuromodulator receptors and makes specific kinds of synapses on specific subcellular locations of specific target cells.

We cannot ignore this cellular complexity, but, until recently, we have had few options for really embracing it. As long ago as 1979, the central importance of cell types was recognised. Francis Crick had seen the power of molecular genetic techniques in other areas of biology and knew that, with the right tools in hand, it could be harnessed to help unlock the mysteries of the brain. His article in Scientific American’s, “Thinking about the Brain” explicitly described three needed methods for neuroscience to make real progress: first, a method by which “all the connections to a single neuron could be stained”; second, a method by which “all neurons of just one type could be inactivated, leaving the others more or less unaltered”; and, third, a means to differentially stain each cortical area, “so that we could see exactly how many there are, how big each one is and exactly how it is connected to other areas.”

While connectomics on various scales is addressing the first and third of these, optogenetics provides the means to accomplish the second. Indeed, it surpasses the requirement Crick had in mind, by allowing not just inactivation but also activation, with exquisite temporal control and rapid reversibility. (As it happens, optogenetics is also a fantastic method for mapping functional connections between cell types).

Using optogenetics, we can move beyond the crude methods of lesion studies or stimulation with electrodes inserted into a particular brain region. These methods are hopelessly confounded by the intermingling of cell types within the targeted regions. In any given area, it is typical to find multiple cell types that directly antagonise each other – lesioning them all or stimulating them all may not reveal the complex functions and computations carried out by the region in question. Optogenetics simply provides a much more precise, selective and controllable method to perform these kinds of investigations.

One example is provided by the circuitry controlling appetite. The arcuate nucleus in the hypothalamus is a crucial hub in this signaling, integrating signals from the periphery, such as leptin and insulin levels, and passing these signals on to further hypothalamic regions which mediate feeding behaviours. Lesioning the arcuate nucleus has little effect on feeding behaviour, however. The reason for that was discovered once the leptin receptor and other players in this system were cloned and molecular genetic characterisations revealed two major cell types intermingled in the arcuate nucleus. These directly antagonise one another and communicate opposite signals to downstream areas – it is the balance between their activities which controls behaviour. Several recent optogenetics studies have now greatly increased our understanding of this system, mapping connectivity to specific cell types in downstream target regions, revealing the hierarchy of their functional relationships and directly demonstrating short- and longer-term effects on behaviour of activity of these different neuronal classes.

These experiments are not just elegant and precise, they are powerful and incisive – they are the right experiments to do to understand this system because they interrogate the system at the right level: the distinct cell types that make up the fundamental computational units.

Another reason I am so excited by optogenetics is it provides one means to integrate analyses at very different levels, uniting what have been disparate areas of neuroscience. The characteristics of individual neurons or specific synaptic connections are traditionally analysed by molecular and cellular neuroscience and electrophysiology. The roles of specific neurotransmitters or receptors are probed with pharmacology. The functions and interactions of brain areas are studied using field recordings, electroencephalography, neuroimaging, lesions and other systems neuroscience methods. These approaches have traditionally been carried out by different people with different skills and different mindsets.

While we may have learned a lot of details at each level, integrating knowledge across those levels has remained a huge challenge. As a result, we have had little real understanding of how the functions of any brain area emerge from the interactions of its component cells.

Optogenetics provides a method to connect those levels. By inhibiting or activating entire classes of neurons within a region and analysing the effects on activity in other cells or regions or the effects on behaviour of the animal, on a moment-to-moment basis, we can discern the functions which these cells and circuits have evolved to perform.

And that’s the key, really – evolution has built the mammalian brain by elaborating on basic plans already present in our distant ancestors. In simpler organisms it is possible to identify not just types of cells, but individual neurons – in nematodes, the 302 neurons have all been named. In insects, you can see the equivalent individual neurons repeated in each segment of the ventral nerve cord. Those nervous systems function based on the actions of individual neurons and their interconnections. Mammalian brains function more at the level of ensembles of neurons, but the basic logic is similar – evolution has built these brains by expanding the numbers of cells of ancestral types, so that what was once a single neuron is now a population of neurons of the same “type”. Evolution has also increased the diversity of subtypes, which are deployed and combined in myriad ways to generate the incredibly complex circuitry we seek to understand. 

That is the reason I argue that cell types are the fundamental units of the nervous system and why optogenetics is such a powerful method to help move neuroscience from crude and fragmented approaches to a united field capable of explaining how the operations of the mind emerge from the workings of the brain.

Let me add a few notes:

First of all, I don’t have a dog in this fight. I have no stake in any optogenetic technologies and don’t currently use the method, though I certainly hope to in the future. I’m simply really excited by its potential. I don’t get giddy over new techniques very often, but when I saw Karl Deisseroth present his team’s work at the first Wiring the Brain meeting in 2009, I was blown away by its potential – along with the rest of the audience of hard-to-impress neuroscientists.

Second, optogenetics alone is not the answer to all things – it is a method that is suitable for asking specific kinds of questions. There are, in addition, numerous conceptually similar molecular genetic techniques now being used or developed, which greatly expand our arsenal of tools for monitoring and manipulating patterns of neuronal activity.

Third, let’s consider a few of the common and recent critiques of the method:

1.     The drivers we are using do not target real cell types, because they depend on the expression pattern of single genes, while real cell types are defined in a combinatorial fashion by the expression of multiple genes. That is absolutely true, but intersectional strategies (which drive expression only where two genes intersect) are greatly increasing the specificity possible. Also, combining transgenic drivers with viral systems that can be delivered to specific brain regions can address many of these issues.

2.     We don’t know what stimulation protocols to use. Just blasting some neurons so they fire like crazy does not recapitulate the real patterns of firing seen in vivo. Also true, though that criticism applies to traditional electrical stimulation as well. But molecular genetic tools designed to monitor and measure these patterns have also been developed and such patterns can be retransmitted through the sensitive, rapid and reversible optogenetic drivers.

3.     It’s not good for studying neuromodulation – the slow signaling which is so important for changes in the functions of neural circuits over longer timeframes. This is just wrong. You just need to target the neuromodulatory neurons – the ones that release dopamine or serotonin in response to action potentials that they fire. Many of the most exciting early papers using optogenetics have taken this approach. In addition, new techniques, like DREADD, have been developed to directly activate G-protein-coupled receptors in a way that closely mimics neuromodulatory effects.

4.     It can’t replace lesion studies. Yes, it can. Or at least it can provide a crucial complement. Lesion studies are great for studying the effects of lesions to specific areas (of obvious clinical importance) but limited for inferring how the functions of those areas are mediated, for the reasons outlined above.

5.     We can’t use it for therapies because we don’t know which brain regions to target. Well, first off, deep brain stimulation is currently in use for conditions like obsessive-compulsive disorder and Parkinson’s disease and is showing great promise for depression. Optogenetic approaches may provide a more sophisticated method to control neural activity, which is directed to specific cell types within the target region. This is likely a long way in the future and would involve the complex issue of transfecting human brain cells with viruses, but it is clearly a theoretical possibility and an interesting avenue to explore. Secondly, optogenetics is primarily a research tool – one that we hope will lead us to a greater understanding of brain circuit function and dysfunction, which, in turn, will allow us to develop new therapeutic approaches. When people like Karl Deisseroth talk about its relevance to psychiatric disease, this is what they mean: “Despite the enormous efforts of clinicians and researchers, our limited insight into psychiatric disease (the worldwide-leading cause of years of life lost to death or disability) hinders the search for cures and contributes to stigmatization. Clearly, we need new answers in psychiatry.” As quoted and misrepresented here.

Finally, to end on a positive note, here are a few of my personal favourites from the recent literature where optogenetics approaches have generated real and novel insights into the organisation and function of specific brain circuits: