Sensory identity usually remains constant across a large intensity range. Vertebrates use lateral inhibition to match the sensitivity of retinal ganglion cells to the intensity of light. A new study published in Journal of Biology suggests that lateral inhibition in the Drosophila antennal lobe is similarly required for concentration-invariant perception of odors.
Adaptation is a fundamental neural mechanism for stable sensory perception in a changing environment. For example, our perception of the contrast between the black text and the white background of a page remains constant under a variety of illumination conditions ranging from indoor lighting to bright sunlight. In a manner similar to this, the olfactory system must be able to perceive the same odor identity across a wide range of concentrations. Why might this be important?
To navigate towards an odor source, a Drosophila larva must be able to recognize odor intensity as well as concentration-invariant odor identity. From physiological studies, however, we know that the odor response of odorant receptor neurons (ORNs) normally saturates within one or two orders of magnitude . In addition, the number of ORNs activated by an odor increases with odor concentration – thus creating a shifting odor representation in the antennal lobe. Despite this, Drosophila larvae can navigate towards an attractive odor source across a much broader range of odor concentrations . How does the olfactory system accomplish this? In this issue of Journal of Biology, Asahina et al.  use a highly effective synthesis of genetics, behavioral analyses and calcium imaging to uncover a neural circuit at early stages of the olfactory system for concentration-invariant odor perception. The data  suggest that properties of lateral inhibitory neurons are the key to understanding how perceptual constancy is achieved in olfactory circuits.
How might lateral inhibitory connections support adaptive functions in a sensory system? The well-studied vertebrate retinal circuitry is one of the best examples for sensory adaptation. Horizontal cells, a type of lateral inhibitory interneuron in the retina, integrate inputs from many cone photoreceptors and make inhibitory synapses back onto the presynaptic terminals of each cone photoreceptor. Neural activity in horizontal cells thus represents ambient light intensity and presynaptic inhibition of the corresponding cone photoreceptor scales with the ambient light intensity (Figure 1a). This effectively results in the transmission of information about the difference between local and ambient light intensity to the corresponding bipolar and retinal ganglion cells .
Figure 1. Similarities between lateral inhibition in the vertebrate retinal system and the insect olfactory system. (a) Ambient light intensity modulates light sensitivity in retinal ganglion cells. Modified from . (b) Similarly, high odor concentrations recruit more odorant receptor neurons, down shifting the sensitivity of the corresponding projection neuron in Drosophila antennal lobe. Modified from Figure 7 of Asahina et al. , with the dashed line indicating potential projection neuron response in normal larvae.
Analogous to the horizontal cells in the vertebrate retina, lateral inhibitory interneurons in the Drosophila adult olfactory system receive inputs from many ORNs of different glomeruli  and feed back onto presynaptic ORN terminals through inhibitory connections. Two recent studies show that these local interneurons (LNs) provide a gain control mechanism to modulate olfactory sensitivity [6,7]. Drosophila larvae have a relatively simple olfactory system, with only 21 ORNs in the dorsal organ, each of which expresses a unique odorant receptor gene . ORNs make synapses with specific projection neurons (PNs) in the antennal lobe, which carry olfactory information to higher brain centers for further processing. Despite its simpler anatomical organization, the larval antennal lobe also contains GABAergic LNs that innervate different glomeruli.
Using a novel larval preparation that is amenable to calcium imaging, Asahina et al.  confirm the observations in adult flies – that a given odorant excites multiple ORNs and a given ORN responds to multiple odorants [1,9]. In principle, a combinatorial code using the glomerular pattern can encode more odors than the number of receptor types available. However, higher concentrations of a given odorant may also activate more ORNs. Therefore, odor identity is potentially confounded by a change of concentration , which is a problem that has attracted much speculation from researchers in the field of olfaction. Indeed, in this study , Asahina et al. report that high concentrations of the attractive odorant ethyl butyrate excite three ORNs – those expressing the olfactory receptor gene Or35a, Or42a or Or42b. Yet the response thresholds of these three ORNs are orders of magnitude apart, with the Or42b and Or35a ORNs showing the highest and lowest sensitivities to ethyl butyrate, respectively. Thus, depending on the ethyl butyrate concentration, the number of recruited glomeruli can switch from one to three.
In order to study the physiological and behavioral response properties of isolated ORN channels, Asahina et al.  created larvae with only one functional ORN using an elegant genetic trick. The Or83b gene is normally expressed and required for odor detection in all larval ORNs. Targeted expression of the wild-type Or83b in the Or83b mutant background generates larvae with just one functional ORN type, which they term OrX-functional. These Or83b rescue experiments allowed them to show that one functional ORN is sufficient for odor navigation towards an attractive odorant. In addition, the behavioral threshold for each isolated ORN channel is similar to its physiological threshold as measured by calcium imaging in this study and electrophysiology in a previous work .
This study  and a recent publication  provide unprecedented resolution on olfactory behaviors of Drosophila larvae with the Or42a and Or42b ORNs. Both studies show that control larvae can navigate toward attractive odorants over a large range of concentrations. Both studies also report that loss of function in certain ORNs causes a reduction in attraction or even avoidance to high concentrations of odorants. Kreher et al.  showed that Or42b mutant larvae exhibit reduced attraction to low concentrations of ethyl acetate, whereas Or42a mutant larvae avoid high concentrations of ethyl acetate. One interpretation of the avoidance behavior offered by the authors  is that hyperactivation of the Or42b ORN or downstream neurons, which is normally balanced by the activation of the Or42a ORN, causes a switch from attraction to aversion.
However, Asahina et al.  show that simultaneous functional restoration of Or42a and Or42b is not sufficient to recapitulate the wild-type attraction behavior. This result led them to investigate other cell types in the antennal lobe. They discovered that LNs fail to respond to odor stimulation when only one ORN channel is present, but that LNs respond to the summed stimulation of Or42a and Or42b ORNs. In parallel, they showed that PN output is suppressed by the simultaneous activation of these two ORNs (Figure 1b).
Together, these results paint a picture in which the firing rate of the GABAergic LNs scales with the number of receptor inputs and serves as a mechanism to dampen PN response. It is interesting to note that the LN response in the Or42a+Or42b-functional larvae is still less than that of wild-type larvae. Based on the data in these two papers [3,11], one might imagine a model whereby Or42a ORNs offer the greatest contribution towards balancing out Or42b hyperactivation through inhibitory LNs. Contributions from the total ORN ensemble may be necessary for sufficient LN recruitment to suppress hyperactivation. Future experiments to investigate the role of inhibitory LNs in the behavioral switch from attraction to aversion will be necessary for a conclusive answer.
The traditional view of GABAergic LNs in the olfactory system is that they serve to increase contrast between odors of similar glomerular patterns by lateral inhibition . The findings of Asahina et al. , together with two other recent studies [6,7], offer compelling evidence that inhibitory LNs may instead mediate automatic gain control to expand the dynamic range of odor responses. Although horizontal cells in the retina and inhibitory LNs in the olfactory system seem to share functional similarities, their exact synaptic wiring diagrams may have differences. Besides targeting the axonal terminal of ORNs [6,7], LNs also synapse with PNs. LNs, in principle, can thus mediate both feedback and feedforward inhibition. GABAB receptor-mediated feedback inhibition is important for efficient odor-tracking behaviors . Feedforward inhibition may be mediated by postsynaptic GABA receptors that reduce dendritic excitability. Future efforts to assess the relative contributions from feedback and feedforward mechanisms in perceptual constancy and adaptation will be crucial for understanding how early olfactory processing shapes incoming olfactory information and how this information is used to generate behavior.