In-Flight Entertainment

One of the reasons why we choose to study the brains of flies, rather than those of our closer vertebrate relatives, is the relatively small number of neurons in these tiny brains, and the fact that single neurons can be reproducibly identified and targeted across different animals. This allows us to dissect specific sets of neurons – known in the business as circuits – that are involved in specific tasks, to unveil how each neuron contributes to the processing in the brain, and to work out why this is relevant to the animal’s needs and evolutionary pressures.

One example of such a circuit is that in the fly brain that processes visual motion. In the fly brain, the areas responsible processing information from the eyes are the optic lobes. Here, the third layer of visual processing is the lobula plate, and neurons here, known as tangential cells, respond to motion across the visual field (that is, their electrical activity is changed when motion is presented in their visual field). Many studies have shown that these cells in fact respond to optic flow: the motion of a visual scene caused by moving through the environment. Try walking forwards: the objects around you appear to move past your eyes in a very stereotyped pattern, which is

The fly visual system - the third layer is composed of the lobula and the lobula plate.

The fly visual system – the third layer is composed of the lobula and the lobula plate.

different from the pattern when you spin in circles. For flies, this kind of visual information is very important, as it tells the fly about its own motion, which is essential for online in-flight corrections. Distinct sets of lobula plate tangential cells respond to different components of optic flow – some, for example, to optic flow caused by forward translation, and others to rotations about specific body axes. The VS cells are a group of tangential cells that respond during such rotations. Detecting these rotations allows the fly to correct for them, using its wings and neck to stabilise its body, in much the same way as we can unconsciously correct for the effect of a sudden gust of wind whilst walking.

The optic flow field generated by rotation.

When not in flight, this kind of rotation does not affect the fly, and so detecting these components of optic flow is not so essential. As such, when the fly is buzzing around, signalling by the VS cells is strongly enhanced compared to when it is on solid ground. First, the resting membrane potential of the cells is increased – which means that it is much easier to produce spikes, the electrical signals that neurons use to transmit information. Also, the response of the neurons to visual motion becomes much stronger. This demonstrates how the processing of sensory information is strongly dependent on the state of the animal, so that valuable processing power is dedicated only to that information that is relevant to the current behaviour.

The mechanisms that generate this modulation of VS cell responses during flight were the subject of a recent paper from the lab of Michael Dickinson. They developed a sophisticated setup, where the fly is fixed in place from above, but is free to move its wings, as if flying. A screen in front of the fly presents visual stimuli, a camera below tracks the fly’s motion, and a puffer jolts the animal into flight using a short air puff. Electrical recordings were used to measure the responses of VS cells to motion presented on the screen under different conditions. Flight enhances their responses to visual motion at certain speeds (or temporal frequencies, a measure of the speed at which bars are moving across the screen).

Octopamine is a molecule used for communication between some neurons in insects and other invertebrates, much like dopamine or serotonin, which are found in the human brain. In locusts, octopamine levels rise during flight, and this is thought to produce the dramatic changes in their physiology seen when they fly. So, thought the authors of this study, octopamine is a likely candidate to cause the in-flight modulation of VS cell responses in flies. First, they showed that applying octopamine to the brain has a similar effect on VS cells responses as flight itself. Next, they look at the activity of neurons that release octopamine, and, sure enough, they become more active during flight. Expressing a heat-sensitive channel specifically in these octopamine-expressing neurons allow them to be activated by heat, and doing so also increases VS cell responses to visual motion. Conversely, inactivating these same neurons prevents the modulation of VS cell responses by flight. Therefore, overall the release of octopamine is sufficient to boost VS cell responses, and is also necessary for this boost.

Flight (blue) boosts the responses of VS neurons to visual motion.

Flight (blue) boosts the responses of VS neurons to visual motion.

Interestingly, the effect of octopamine seems to be restricted to the boost of responsiveness to visual motion during flight. The shift in baseline membrane potential of VS cells during flight appears to be independent of octopamine, suggesting that different mechanisms converge to increase VS cell responsiveness during flight. One possibility is that this separate effect is due to input from mechanoreceptors, which detect air flow on the body surface during flight, a reliable indicator of active motion.

This neat study adds to the wealth of literature showing that, from the nematode worm to the primate brain, neuromodulators (chemicals that modulate neural signals rather than directly transmitting them) play an essential role in altering how the brain processes information depending on the state of the animal. This ensures that precious resources are dedicated primarily to those tasks that are important for the current behaviour, and to those behaviours that will keep the animal fit and healthy, ensuring its survival and reproduction. After all, the ability to survive and reproduce is what drives evolution – and so all the intricate and complex mechanisms in our brains have evolved to do just this.



In an electrical circuit, it is the connections between components that determine what the circuit can do. When driving from one city to another, you need to know not only the roads that you will take, but where these roads connect. The human brain contains some 85 billion neurons, and these neurons communicate through well over 100 trillion connections, or synapses. If we want to begin to understand how networks of neurons process information and generate all of our varied and complex behaviours, then clearly an essential prerequisite is to know how the network is wired up – which neurons are connected to one another, and how strongly. Doubtless you have heard of the genome – the full complement of genes in your DNA – and perhaps even the microbiome – all of the microorganisms that live and grow in and on your body. Well, in this vein, the complete set of connections between single neurons is known as the ‘connectome’, and neuroscientists are currently investing a lot of effort into unravelling this tangled web of connectivity, as an early step into understanding how the brain performs all of its amazing feats.

Traditionally, two main approaches have been used to analyse single connections within the nervous system. The first uses recordings of the electrical activity of neurons: stimulate one neuron, and record whether this leads to a response in another neuron. The delay

An electrical current is injected into the ‘layer VI’ neuron, and a response is recorded in the ‘nRT’ neuron. From Gentet & Ulrich, 2004.

between the stimulation and the response indicates whether this is a direct (monosynaptic = one synapse) or an indirect (polysynaptic) connection; alternatively, there are several ways to prevent polysynaptic transmission, so that only direct connections remain. This approach also allows us to measure the strength of the connections. However, there are several limitations to this method: first, each recording measures only a single connection, so looking at even a small fraction of the connections in the brain would require a huge effort. Secondly, it is often impossible to record from identified neurons in intact brains, particularly in vertebrates, so brain slice preparations are used instead – but these slices sever many of the very connections we are trying to uncover.

Part of a neuron seen by electron microscopy. Synapses indicated by arrows.


The reconstructed C elegans connectome.


The other traditional approach to the connectome is electron microsopy. This is a form of microscopy that uses electrons in the place of light, allowing us to see structures on a scale smaller than a single cell, which would be indistinguishable to a light microscope. By taking a series of images at successive levels of brain tissue, we can trace the course of a neuron through the brain, and see every synapse that it makes along the way. The synaptic strength can be roughly estimated from the amount of transmitter stored therein.

In fact, we already have the full wiring diagram for one species – the nematode worm C. elegans. This worm has only 302 neurons, and around 7000 synapses, but even so this connectome required a monumental effort, due to the sheer amount of time required to prepare each micrograph and to reconstruct a series of them into a 3D picture. Therefore, scaling up these techniques to more complex nervous systems – even those of flies or fish – is simply not practical. For these reasons, neuroscientists have recently been racking their brains to come up with new approaches to the connectome.

Several of these novel approaches use light microscopy. Just looking at neurons through a light microscope can only tell us whether two neurons are within close proximity, but not whether they make a connection. However, GRASP technology can overcome this limitation. Green Fluorescent Protein, or GFP, is a molecule naturally produced by jellyfish that fluoresces brightly in green when illuminated with blue light. Since expression of GFP can be genetically targeted to specific cell types, it is widely employed in biology as a label. With GRASP, the GFP molecule is split in two, and each half of the molecule is targeted to one of the neurons of interest. Fluorescence is only seen when the two halves come together – which only happens when the cells come into very close contact, as at a synapse. But there is a limit to how specifically neurons can be genetically labelled, and so not all possible pairs of neurons can be studied using GRASP.

The three steps of sequencing the connectome. From Zabor et al, 2012.

In a recent issue of PLoS Biology, Zabor et al a new approach that they believe could revolutionise the study of connectomics. This approach relies on recent advances in DNA sequencing technology: a DNA sequence of one million As, Cs, Ts and Gs, which would have cost over US$5000 to sequence just ten years ago, would now cost less than 50¢. Their concept is composed of three components. First, each neuron is individually labelled by a short DNA sequence – its unique ‘barcode’. Next, the DNA barcodes of directly connected neurons must be associated with one another. One way to do this could employ viruses that can package the DNA barcode of a ‘donor’ cell, and travel to directly connected ‘recipient’ cells, taking the barcode with them. Thus each neuron would contain its own barcode, plus those of directly connected neurons. The final step is to connect all of the barcodes within each neuron together into a single piece of DNA. Then, all that’s left is to read the sequences of all of these pieces. If two barcodes are found together on the same piece of DNA, then the neurons identified by these barcodes must be connected. Sequencing these pieces from every neuron in the brain would give us a complete picture of which neurons are connected to one another.

There are, of course, limitations to this approach. For one thing, it would not allow us to measure the strength of the connections, or whether they are excitatory or inhibitory. Also, we would not obtain any spatial information regarding the layout of neurons, which is important because the shape and size of a neuron affects its electrical properties. Nevertheless, this approach is promising, partly because it would identify long- and short-range connections equally well, whereas conventional methods are very poor at tracing long-range connections.

Hurdles remain to be overcome before the sequencing approach to the connectome sees the light of day – how to get these barcodes into every neuron, how to couple connected barcodes and how to join these barcodes together. Still, this idea has generated a lot of interest amongst neuroscientists, not least because it offers the possibility of characterising the connectomes of animals in very short times and for a very small cost – something that will be essential to understand the complex interactions between the web of neurons that generates all of our thoughts and actions. But of course, as C. elegans has shown us, even knowing all of an animal’s neurons and all of their connections will not tell us how the nervous system works. For this, we need to know the properties of every neuron and of every connection, and the information coming into the system. Clearly, understanding how brains work is going to be a long, hard problem to crack. But it’s new ideas like “sequencing the connectome” these that will help us on our way.


Gentet LJ & Ulrich D (2004). Electrophysiological characterization of synaptic connections. Eur J Neurosci 19(3), 625-33.

Zador AM, Dubnau J, Oyibo HK, Zhan H, Cao G & Peikon ID (2012). Sequencing the Connectome. PLoS Biol10, e1001411.

Tasting Blood.

Here’s a question: why do you choose to eat certain foods at certain times, and not other things at other times? Why will a mouse work so hard for a sugar reward, and why does it keep coming back? Moreover, why does it ever stop eating? New research in flies is beginning to uncover how they detect food sources in the environment and evaluate their nutritional values, and how they decide whether or not to dive in and chow down.

Animals employ a diverse arsenal of tools to tell whether a food source is worth pursuing. A fly’s sense of smell helps it to locate a piece of rotting fruit and fly towards it. Once it gets there, the fly uses its sense of taste to what the food contains: sugars, toxins, salts and so on. But these senses are not completely infallible: artificial sweeteners, for example, taste great but have very little nutritional value. It is only in the last few years that we have begun to see that even after eating a food, animals are capable of assessing its nutritional value, assigning this value to a particular food source, and using this value in future decisions about what to eat. What’s more, we are getting closer to understanding what an internal nutrient sensor really is, and how it is integrated with other senses to guide feeding decisions.

Last year, three groups independently demonstrated that flies can detect the nutritional content of food independently of its taste. Dus et al (2011) allowed flies to make a choice between two different food sources – one containing red dye, and the other blue – and assessed their food choices based on the colour of the flies’ abdomens after 2 hours.
Given a choice between a nutritious sugar (sucrose) and agar containing no food source, flies will usually choose sucrose. The researchers then made use of mutant flies that lack the ability to detect external sugars, due to an absence of the taste receptor molecules Gr5a and Gr64a, or due to an absence of all taste receptor organs. After 5 hours of starvation, these flies eat much less sugar, and many more of them eat nothing at all, since they cannot taste the sucrose. However, after 22 hours of starvation, these flies eat just as much sucrose as normal flies. This works for different sugars, such as fructose and glucose, but not for non-nutritious D-glucose or sucralose. This suggests that in a starved state, the flies rely much more on their internal assessment of a food’s nutritional benefits, rather than an external sense of taste, to guide their choices.

Fujita & Tanimura (2011) and Burke & Waddell (2011) both both demonstrate that flies can learn an attraction to a food source based purely on its nutritional value, without the attraction of sweet taste. If an odour is presented together with a sweet-tasting food, flies will rapidly learn to approach this odour when it reappears. Fujita & Tanimura showed that flies can learn such an association even when the sugar is tasteless (sorbitol), as long as it is nutritious. Burke & Waddell used a similar odour-pairing test, and showed that while short-term memory is based mainly on sweet taste – less nutritious sweeteners such as arabinose were as effective as fructose as rewards – long-term odour memories are formed much more effectively by sugars that provide a nutritional benefit. While long-term memories formed by non-nutritious but sweet arabinose are very weak, supplementing this sugar with nutritious but tasteless sorbitol allows the formation of strong memories, showing that flies can assess the nutritional content of a food after eating it and use this to guide future decisions.

To understand how a fly integrates this nutritional information into its picture of a food source, Stafford et al (2012) made use of a fly café. The CAFE (CApillary FEeding assay) is not like your regular greasy spoon: the flies are provided with tubes extending from their ceiling, each containing liquid fly food. As the levels of the liquids change, one can track the amount of food eaten by the six-legged customers. They showed that flies’ initial food choices can be explained by their taste alone, but that over time their consumption shifts to reflect the nutritional content of the food sources. For example, given the choice between a sweet, non-nutritious sugar and a less sweet but highly nutritious sugar, the flies initially choose the former, but over time they increasingly choose the latter. Indeed, this shift to nutritious sugars occurs much more rapidly in hungry flies, beginning within 20 minutes. So it seems that the role of the internal nutrient sensor is to refine the initial evaluations of food sources by the taste system. But still, the identity of this internal nutrient sensor remains elusive..

… Or at least it did, until last week. In the most recent issue of Cell, Miyamoto et al identify a molecule essential to the functioning of an internal sugar sensor in flies. First, they show that the taste receptor Gr43a is a specific receptor for fructose, a sugar found, surprisingly, in fruits. As well as in taste organs, Gr43a is also found in flies’ brains. Flies lacking Gr43a in the brains cannot base their food choices on the nutritional sugar content of foods, but rely on their perceived sweetness alone. Thus, flies detect changes in the level of fructose in the haemolymph (fly blood) following a meal as a measure of the food’s nutritional content. But the main sugars in the haemolymph are trehalose and glucose – so why use fructose? Well, the levels of trehalose and glucose have to be very tightly regulated, to keep their levels stable so that all organs receive just the right amounts of each. But fructose, as a minor constituent of the haemolymph, is free to vary much more widely, so it serves as a much more reliable indicator of the amount of sugar entering the body from a meal.

ImageGr43a is found in only 6-8 neurons of the brain (out of around 100 000 neurons in all). Miyamoto et al show that the effect of activating these neurons depends on the state of the fly. They use genetically implanted channels that render just these few neurons sensitive to temperature, so that by raising the temperature they activate these neurons at specific times. A similar odour-pairing test is used as above – one odour is paired with activation of these neurons, while the other odour remains neutral, and the fly is then given a choice between the two.


These experiments show that in full flies, activation of these neurons is perceived as negative, and they will avoid activating them. In contrast, hungry flies prefer to activate the neurons. In practice, this would mean that fed flies would avoid eating nutritious foods, while hungry flies would eat in order to raise their fructose levels and activate these neurons. So not only do these neurons act as internal nutrient sensors of sugars in the haemolymph using Gr43a, but activating them can be either rewarding or repulsive, depending on the feeding state of the fly, and thus they can guide useful feeding decisions to keep the fly well-fed but prevent overeating. Now if only we humans could make good decisions like those. We have a lot to learn from the humble fruit fly.

Dus M, Min S, Keene AC, Lee GY & Suh GSB (2011). Taste-independent detection of the caloric content of sugar in Drosophila. PNAS 108, 11644–11649.

Fujita M & Tanimura T (2011). Drosophila Evaluates and Learns the Nutritional Value of Sugars. Current Biology 21, 751–755.

Burke CJ & Waddell S (2011). Remembering nutrient quality of sugar in Drosophila. Curr Biol 21, 746–750.

Stafford JW, Lynd KM, Jung AY & Gordon MD (2012). Integration of taste and calorie sensing in Drosophila. J Neurosci 32, 14767–14774.

Miyamoto T, Slone J, Song X & Amrein H (2012). A Fructose Receptor Functions as a Nutrient Sensor in the Drosophila Brain. Cell 151, 1113–1125.

Food For Thought

How do animals – and that includes us humans – choose which foods to eat? How does that jumble of neurons in your head tell you when it’s meal time? Neuroscientists are beginning to uncover some of the mechanisms that govern feeding decisions in all kinds of animals – and with the rising rates of obesity both in the Western world and in developing nations, there’s never been a greater need to understand what makes us hungry.

Animals, just like humans, love sweet food. Even a tiny fruit fly will go crazy for a drop of sugar water. What’s more, flies are attracted to many of the same artificial sweeteners as us, even though they provide no nutritional benefit [1]. As well as decaying fruits, these flies also consume yeast, and just like us humans they are especially fond of beer [2].
To locate a tasty piece of rotten fruit, a fly principally uses two of its senses: smell and taste. As it buzzes around its environment, olfactory receptors on the fly’s antennae detect minute amounts of odours in the air. Like Scooby to his snacks, the fly follows the odour to its source, and lands on the delicious morsel that awaits. But the fruit is probably buried in amongst a heap of things the fly definitely doesn’t want to eat – and so here’s the great part. The end of a fly’s leg is covered with taste receptor cells; so as the fly walks around, it can tell whether the substance beneath it is edible. But here there’s a problem: imagine finding yourself on a piece of fruit that’s hundreds of times your own size. How are you going to eat it? Well, of course, instead of bringing the food to your mouth, you bring your mouth to the food.

Flies, like other insects, are equipped with a wonderful piece of feeding machinery: the proboscis. When the receptor cells on a fly’s legs detect an attractive taste, the fly’s trunk-like proboscis shoots out from its head towards the food. On the end of the proboscis are the labellar plates, which perform much the same role as our tongue: they are covered with taste receptor cells that allow the fly to assess whether a food source is delicious or potentially toxic before committing to ingesting it. If it likes what it tastes, the fly opens out the labellar plates, sucking food and liquid through the proboscis and into the digestive system. Here too, in the pharynx, lie more taste receptor cells that make a further assessment of the food as it is ingested. So by now the fly has a pretty good idea of what it’s eating. Attractive substances – many sugars, glycerol from yeast fermentation, low concentrations of salt – will be consumed, while foods contaminated with bitter substances or high levels are salt are rejected [3].

Just as with attractive stimuli, many of the substances that flies avoid are also bitter to us. Quinine, for example, is used to give tonic water its bitter taste; caffeine makes coffee bitter; and denatonium, the most bitter compound known to man, is added to household cleaning products to stop kids from drinking them. All of these compounds make foods less attractive to flies, and the decision of whether or not to eat ultimately depends on the balance of attractive and repulsive tastes, and how hungry the fly is – as it progresses from mere hunger to starvation, a fly’s tolerance of bitter substances increases.
How do we know so much about what tastes good to a fly? Well, in the lab we can use the fact that a fly extends its proboscis in response to stimuli touching its legs or labellar plates to probe its sense of taste. Touch the legs with a drop of sugar solution, and the proboscis extends. Add a bitter compound to this solution, and the probability of proboscis extension decreases. This simple behaviour has been used for over 50 years to characterise the responses to a range of tastes, how these responses change with the state of the fly, and the molecules and neurons responsible for tasting and co-ordinating the appropriate response. And it’s this experimental setup that I use each day to try and understand more about how our senses are modified by the state of our brains and bodies, and how networks of neurons allow us to make the most important decisions in life: to eat, or not to eat.

[1] Gordesky-Gold B, Rivers N, Ahmed OM & Breslin PAS (2008). Drosophila melanogaster Prefers Compounds Perceived Sweet by Humans. Chem Senses 33, 301–309.
[2] Wisotsky Z, Medina A, Freeman E & Dahanukar A (2011). Evolutionary differences in food preference rely on Gr64e, a receptor for glycerol. Nature Neuroscience 14, 1534–1541.
[3] Amrein H & Thorne N (2005). Gustatory Perception and Behavior in Drosophila melanogaster. Current Biology 15, R673–R684.

Welcome to my world

In the summer of 2012, I left the grey skies of my home country, travelling over land and sea to settle in warmer climes here on the southern tip of Europe. As well as my first step into this new country, this was also my first step into the world of science, the first step on a journey in which I hope you will join me, as I learn about the world around us, and how to probe its secrets. I will be trying to unlock the inner workings of our brains, by dissecting those of flies, using all kinds of neat tools that those before me have developed. Along the way, I will share insights from things that I read, the scientists I meet, and my own work on how flies’ brains integrate sensory and internal information to make some of the most important decisions of their short lives – what, when and how much to eat. So take a look, open your mind and get in touch.
Peace out.