People often seek a cool breeze on a hot summer's day, and delight in a roaring fire on a cold winter's evening — the desire to stay in a thermal 'comfort zone' is a powerful driver of our behaviour. Because temperature is both a sensory stimulus and a factor that directly affects metabolic processes, it is not surprising that nearly all animals avoid unfavourable temperatures1. But what is the biological basis of temperature preference? Two studies2,3 in this issue track the flow of temperature information into the brains of fruit flies (Drosophila melanogaster). They reveal that the brain extracts different qualities of temperature at the very first connection, and does so using distinct mechanisms for hot and cold.
Fruit flies have long been the workhorse of genetic studies, but the animals' behavioural repertoire is also noteworthy. Like most arthropods, flies cannot regulate their core temperature internally. Instead, they move in search of favourable temperatures, which makes them ideal for studies of thermosensing. D. melanogaster exhibits exquisite sensitivity to small temperature changes4, and can even associate thermal conditions with other senses, such as sight and smell, to remember locations and avoid ominous odours5.
Studies have defined the organs6, neurons7,8 and receptor proteins9,10 that fruit flies use to avoid unfavourable temperatures. Their primary sensory neurons are in the antenna, which is a veritable Swiss army knife of sensory functions. Individual thermoreceptor neurons are mainly excited by small temperature changes — hot thermoreceptors are strongly excited by heating and weakly inhibited by cooling, and cold thermoreceptors have the opposite response3,8. Genetically silencing these peripheral thermoreceptor neurons, or removing their hot or cold receptor proteins, prevents the fly from avoiding harmful heat or cold7,8. Thermoreceptor neurons project to an area of the fly brain called the proximal antennal protocerebrum (PAP), in which their output terminals (axons) segregate into hot and cold zones8.
To define temperature-processing circuits, Frank et al.2 (page 358) and Liu et al.3 (page 353) used tools for controlling expression patterns of introduced genes11,12 to identify and then manipulate projection neurons, which convey thermal sensory information from the PAP into the central brain. Frank and colleagues surveyed the potential targets of thermoreceptor neurons, and found groups of projection neurons that send axons to three brain regions. The major projection is to the lateral protocerebrum, an understudied region that receives inputs from many other sensory systems. The other two projections innervate the mushroom body and lateral horn, regions involved in learned and innate olfactory behaviours, respectively.
Frank et al. genetically manipulated flies such that the animals' PAP projection neurons expressed a fluorescent indicator of neural activity. The authors then used in vivo imaging to probe the function of these different types of projection neuron. They observed distinct responses to thermal stimulation — most of the projection neurons were narrowly tuned, responding exclusively to either heating or cooling, but a subset was broadly tuned, responding to both stimuli (Fig. 1a). Frank and colleagues demonstrated that some of these broadly tuned neurons are required for normal temperature-preference behaviour.
The researchers then showed that the thermal-response properties of each type of projection neuron can be largely predicted on the basis of whether it innervates the hot or cold zone of the PAP. Broadly tuned projection neurons seem to be driven by both hot and cold stimuli, whereas hot projection neurons and cold projection neurons have been proposed to act as 'labelled lines', which convey information from only one class of receptor neuron into the central brain.
Liu and colleagues' results remind us that brains rarely use the simplest solution to a problem. The authors exploited similar tools to Frank et al. to identify and label projection neurons, but used intracellular electrophysiological techniques to measure the responses of single neurons to thermal stimulation. Again, Liu et al. identified diverse response profiles in different types of projection neuron. They then chose to closely examine one subset of narrowly tuned projection neuron.
The researchers found that cold projection neurons are primarily driven by excitatory input from cold thermoreceptors, consistent with the labelled-lines hypothesis. However, they report that hot projection neurons not only receive direct excitatory input from hot sensory neurons, but are also indirectly excited by the cold pathway. Through a series of experiments to tease apart the contributions of different sensory neurons onto the projection neurons, the authors showed that the cold pathway excites the hot projection neurons in response to heating by reducing the inhibitory activity of an intermediate connection linking the cold and hot pathways (Fig. 1b).
Why should such a system exist? Liu et al. revealed that the hot projection neurons exploit both the excitatory 'getting hotter' signal from the hot receptors and the inhibitory 'getting less cold' signal from the cold receptors, and suggest that this not-quite-redundant use of both pathways leads to more-sensitive measurements of temperature change. This circuit motif should suppress any common, temperature-independent excitation between the hot and cold receptor neurons, and reinforce the opposing reactions to changes in temperature encoded by each pathway. This seems so sensible that it raises a question: why do only the hot projection neurons seem to use this extra circuitry?
These two studies are a step towards understanding how temperature information is routed and transformed in the fly brain. Frank et al. outline a blueprint for thermal-information pathways. Taking a complementary approach, Liu et al. uncover a mechanism that enhances sensory function. The differing approaches were made possible by a confluence of tools and resources in Drosophila neuroscience. With a precise anatomical starting point, we can now routinely obtain genetic access to the neuronal types that pass into and out of a brain region.
Applying these same approaches to follow thermal processing one step further into the brain may not be straightforward, because the three downstream brain regions receive multisensory input. Mapping out the functional anatomy of these areas will be essential. Although potentially difficult, this is exactly the puzzle that must be solved if we are to unravel the mechanistic basis of diverse, thermally mediated behaviours. Surely, the way in which these pathways coordinate temperature-preference behaviour will soon be explained. Neuroscientists take note — the drosophilists are just getting warmed up.