Olfactory coding from the periphery to higher brain centers in the Drosophila brain
Odor information is processed through multiple receptor-glomerular channels in the first order olfactory center, the antennal lobe (AL), then reformatted into higher brain centers and eventually perceived by the fly. To reveal the logic of olfaction, it is fundamental to map odor representations from the glomerular channels into higher brain centers.
We characterize odor response profiles of AL projection neurons (PNs) originating from 31 glomeruli using whole cell patch-clamp recordings in Drosophila melanogaster. We reveal that odor representation from olfactory sensory neurons to PNs is generally conserved, while transformation of odor tuning curves is glomerulus-dependent. Reconstructions of PNs reveal that attractive and aversive odors are represented in different clusters of glomeruli in the AL. These separate representations are preserved into higher brain centers, where attractive and aversive odors are segregated into two regions in the lateral horn and partly separated in the mushroom body calyx.
Our study reveals spatial representation of odor valence coding from the AL to higher brain centers. These results provide a global picture of the olfactory circuit design underlying innate odor-guided behavior.
KeywordsDrosophila melanogaster Insect olfaction Antennal lobe Mushroom body Lateral horn Whole cell patch-clamp
Olfactory sensory neuron
Single sensillum recording
A common feature of the olfactory system in insects, mammals, and most other higher animals possessing a sense of smell is an organization where the first order olfactory center (the antennal lobe (AL) and the olfactory bulb) comprise spherical neuropils called glomeruli, each of which receives convergent input from olfactory sensory neurons (OSNs) expressing a given odorant receptor (OR) [1, 2]. Most odors are detected by multiple olfactory receptor classes with various binding affinities. This enables animals to discriminate many more odors than the number of receptor classes expressed. However, although one of the fundamental questions in olfaction is to understand the central processing underlying such combinatorial coding, the answer still remains elusive.
The Drosophila and the vertebrate olfactory system share a similar organization from the first order processing areas up to higher brain centers [23, 24]. In mice, axonal projections of mitral and tufted cells (equivalent to PNs) from individual glomeruli spread diffusely to the piriform cortex without apparent spatial preference, while they show relatively stereotyped innervation to the cortical amygdala. Since the piriform cortex is associated with olfactory learning, and the cortical amygdala plays an important role in innate behavior, this dichotomous organization resembles the MB and LH structures in insects [23, 24]. Comparing such similarities and differences between vertebrates and insects helps in shedding light on general principles of the olfactory system.
How does a fly interpret odor information? Previous studies have revealed a relationship between odor representations in the AL and innate behaviors . Several ecologically relevant odors use a dedicated pathway leading to an innate behavior, such as the male-produced pheromone 11-cis-vaccenyl acetate (cVA) [26, 27], the mating enhancing pheromone methyl laurate , CO2 , acids , ammonia and amines , parasitoid odors , and the microbial odorant geosmin . In contrast, other odors that also induce innate attraction or aversion are processed via a combinatorial code comprising multiple glomeruli. For instance, a highly attractant mixture induces a specific activation pattern among a combination of glomeruli . Knaden et al. screened 110 monomolecular odorants for their innate odor valence, thereby revealing a valence-related spatial representation often comprising combinatorial glomerular activation patterns . Although another study demonstrated that individual glomeruli, rather than the entire pattern of active glomeruli, mediate innate olfactory attraction and aversion , a recent study demonstrated that flies’ innate odor responses can be predicted by a model summing normalized glomerular responses, in which each glomerulus contributes a specific, small amount to odor preference . How does such odor information of different innate valence converge and interact at higher brain centers? Without a precise neural map and a functional map to place on top of it, it is difficult to assess how glomerular activation patterns would be integrated into higher brain centers.
In this study, we used in vivo whole cell patch-clamp recordings in Drosophila melanogaster and mapped odor response profiles of 71 PNs emerging from 31 glomeruli (covering 60% of ~50 glomeruli in total) to 17 different odors including innately attractive and aversive odors. We addressed the following questions: How is odor information transformed from OSNs to PNs? How is odor information mapped at the level of AL glomeruli and transformed into higher brain centers? And, finally, how is this map related to the fly’s perception and innate behavior?
Many odors are encoded by PNs in a combinatorial manner
We characterized odor response profiles of 71 PNs using in vivo whole cell patch-clamp recordings and subsequently stained them to identify the innervated glomerulus for each individual neuron (Fig. 1b–d). A total of 67 out of the recorded 71 PNs represent uniglomerular PNs that extended their primary dendrites into a single glomerulus and sent their axons to the calyx of the MB and the LH through the medial antennal lobe tract (mALT) also known as the inner antenno-cerebral tract (IACT) (Fig. 1a). The remaining four PNs innervated either two glomeruli or the region posterior to the AL (data not shown). In this study, we focused our analyses on the 67 uniglomerular PNs innervating one out of 31 glomeruli and therefore covering ~60% of 52 olfactory glomeruli present in the Drosophila AL. We tested 17 behaviorally relevant monomolecular odors that flies encounter under natural conditions (Additional file 1: Table S1). These odors induce innate attractive or aversive behaviors, which was earlier demonstrated using diverse behavioral assays, such as the trap assay , the T-maze assay [13, 32, 37], or the FlyWalk [38, 39]. We defined the 17 odors as attractive or aversive if flies showed attractive or aversive odor-guided behavior toward these odors in at least one of the behavioral assays (see Additional file 1: Table S1).
PNs showed a spike firing increase when activated by an odor (Fig. 1d). To evaluate odor-induced neural activities, we quantified spike numbers during the 1-s odor stimulus as an indicator of odor response intensity. We observed some slight variations in odor-induced temporal firing patterns, such as lengths of firing phases (Fig. 1d, e), but confined our analysis to the odor response intensity emphasizing rate coding on the system. Temporal patterns might convey critical information on odor stimulus features as demonstrated in other olfactory systems , but we address this issue later (see the section on Temporal dynamics of odor representation).
There are in total ~150 uniglomerular PNs and 52 glomeruli; the number of PNs innervating the same glomerulus is on average three but is variable between glomeruli in that the glomeruli innervated by narrowly tuned OSNs seem to possess a larger number of PNs, while the glomeruli innervated by broadly tuned OSNs possess a smaller number of PNs [4, 41, 42]. A previous study reported that these sister PNs have highly correlated patterns of activity in odor responses as well as spontaneous synaptic inputs . Therefore, it can be assumed that sister PNs should be identical in terms of odor responses, thereby resulting in ~50 PN classes. In the present study, we investigated one PN per fly; therefore, it was necessary to confirm that odor responses of each PN class would be reproducible under our recording conditions, even when recorded in different individuals. Indeed we found that odor response profiles of the same PN class (i.e., innervating the same glomerulus) were highly similar (Fig. 1e). The degree of variation in the odor responses of the same PN class is shown in Additional file 2: Figure S1, illustrating the caveat that although the response patterns were similar, there was inter-neuron variability in the magnitude of odor responses among the same class of PNs. Furthermore, we calculated the Pearson’s correlation coefficient based on odor response intensities to 17 odors between the pairs of all PNs and corroborated that the responses in pairs of PNs innervating the same glomerulus were highly correlated (median r = 0.9402, n = 46), while the responses in pairs of PNs innervating different glomeruli were not (median r = –0.0634, n = 2034, p < 10–24, Wilcoxon rank sum test; Fig. 1f). From this we concluded that characteristics of PNs investigated in different flies could be directly compared.
A global picture of odor transformation between OSNs and PNs
How is the odor information transformed from OSNs to PNs? A previous study comparing neurons innervating seven glomeruli revealed that odor tuning curves tend to be broader in PNs than in the corresponding OSNs that innervate the same glomerulus . By contrast, narrowly tuned odor response profiles are conserved between OSNs and PNs in the case of dedicated channels mediating odors of specific ecological significance [28, 32, 46]. As these investigations point in different directions, we asked if there is a general rule for odor transformation using our set of 17 odors recorded in PNs innervating more than half of the AL glomeruli.
First, in order to characterize the odor response profiles of OSNs corresponding to our PN recordings, we performed single sensillum recordings (SSRs). In D. melanogaster, 17 types of sensilla have been described, including trichoid, basiconic, and coeloconic types; each houses one to four OSN classes [5, 47]. Via a system-wide SSR screen from all OSN classes [32, 48], we were able to characterize odor response profiles of 29 OSN classes corresponding to 29 of the 31 glomeruli that were covered by our PN recordings (n = 3 for each OSN class, Fig. 2c, Additional file 4: Table S3). Odor response profiles of two OSN classes could not be characterized; one OSN class projecting to DC4 is housed in grooved sensilla that project to the interior of the sacculus that are not accessible by SSR , and the other OSN class projecting to VA7m has not been assigned to a specific functional class . Our SSRs show that many odors activate more than one OSN class and many OSN classes respond to more than one odor (Fig. 2c). There are thus similar combinatorial patterns as observed for the PN recordings. Only 13.8% of the odor-OSN pairs investigated (n = 68/493) showed ≥50 spikes during the 1-s odor stimulus and 18.6% (n = 90/473) exceeded ≥30 spikes to the set of odors tested in the present study (Fig. 2d).
Next, we reconstructed spatial odor response maps of OSNs and PNs in the AL and compared these maps with each other (Additional file 5: Figure S2). We found a remarkable similarity between these two maps, suggesting that odor response patterns are fairly conserved between the two processing levels. Then, to evaluate how odor tuning is transformed from OSNs to PNs in the same glomeruli, we calculated the Pearson’s correlation coefficient based on odor response intensities to the 17 odors between the corresponding OSN and PN classes for the 29 glomeruli, where both OSN and PN odor response profiles were available. The correlation was moderately high (r = 0.72 ± 0.38, n = 29, p < 0.01 for the 25 glomeruli except for DA1, VA1lm, DL3, and VA1d, Pearson’s correlation; Fig. 2e, f). When the OSN and PN glomerular combinations were randomized, the high correlation was dramatically decreased (r = 0.14 ± 0.038, n = 29, 100 runs). Notably, for the four glomeruli DA1, VA1lm, DL3, and VA1d, a low correlation was found (r = 0.19 ± 0.07, n = 4, p > 0.1 for all four glomeruli, Pearson’s correlation). These glomeruli are innervated by OSNs present in trichoid sensilla, and the specific ligands for the OSNs and PNs targeting these were not included in our odor set [28, 49]. Except for these low response glomeruli, we found a similar range of correlation (r 2 = 0.37–0.99) in the 25 glomeruli as previously reported (r 2 = 0.26–0.81 in the 7 glomeruli ), supporting the idea that odor response profiles are generally conserved between OSNs and PNs, but likely modulated by lateral interactions between glomeruli [50, 51].
Our results thus provide a global perspective of the transformation in the AL by covering more than half of the glomeruli and their innervating input and output elements. Further, we reveal a non-uniform trend of odor tuning transformation between OSNs and PNs. As in most olfactory studies, a general caveat is that our odor set included only 17 odors, which will inevitably cause some bias. We do, however, know that these odors are behaviorally significant to the fly.
A functional odor representation map in the AL
We also performed a hierarchical cluster analysis for our 17 odors based on response intensities of the 31 PN classes using Euclidean distances (Additional file 7: Figure S4). Similar divisions were obtained as when using the cosine distance, thereby confirming that the odors can be classified into three groups. However, since the Euclidean distances depend on absolute intensities of odor response, one cluster appeared to be formed partly based on the low response intensities elicited by these odors (Additional file 7: Figure S4a, colored in red).
We next applied a principal component analysis (PCA) to visualize the 31-dimensional odor space in three dimensions (Fig. 4d). Consistent with the hierarchical cluster analysis, we found that attractive and aversive odors clustered separately. The PC1, accounting for 27.8% data variance, roughly separated attractive and aversive odors, as demonstrated by a previous study (Knaden et al. 2012) , and the PC2 (21.3%) further separated the two aversive odor groups. In summary, these results indicate that attractive and aversive odors are significantly separately represented by different combinations of glomeruli (p < 0.001, one-way analysis of similarities, ANOSIM).
Implication of the AL functional map
By measuring the distances among odors in PN neural space, we found three odor clusters that were represented by separate, but partly overlapping, groups of neighboring glomeruli (Fig. 4a–c). Since we have shown that the odors within the same cluster induced similar glomerular activation patterns among PNs (Fig. 4c), we asked how these odors are chemically related. We obtained physicochemical properties of these 16 odors (except for geosmin) calculated by the Dragon software with 32 descriptors from  and applied cluster analysis based on Euclidean distances among the physicochemical properties of the odors (Additional file 8: Figure S5a, b). Some pairs or combinations of odors that have similar physicochemical properties clustered closely, such as acetophenone and benzaldehyde, 1-octanol, 1-hexanol, and 1-octen-3-ol (Additional file 8: Figure S5a), also clustered in the PN neural space (Fig. 4a). We performed a PCA to visualize the 32-dimensional physicochemical space in three dimensions (Additional file 8: Figure S5c). This map showed similar but slightly modified distribution of the odors in the physicochemical space compared to those in neural space represented by PN odor responses (Fig. 4d and Additional file 8: Figure S5c). These results suggest that PN neural space largely preserves relationships as measured by physicochemical properties of odors. However, the space was not a simple reflection, but was modified. The degree of similarity of the representations was tested by comparing the correlation between physicochemical distances and PN response distances for each odor pair (Additional file 8: Figure S5d). We found a moderate correlation between these two parameters (r = 0.518; p = 1.43 × 10–9). Thus, the physicochemical space is reorganized into PN neural space, where odors are encoded by a combination of AL glomerular groups.
We next addressed the question about how neighboring glomerular groups (defined in Fig. 4c) represent specific odor groups. One hypothesis was that neighboring glomerular groups possess similar odor response profiles because the OSNs projecting to these glomeruli might preserve similar OR protein sequences. Therefore, we examined similarity of OR protein sequences by generating a phylogenetic tree of ORs. The phylogenetic tree of 60 ORs and the distance matrices among these ORs indicated that ORs that project to the same glomerular groups (labeled in the same color) did not cluster but rather spread widely (Additional file 9: Figure S6a, b). Thus, these glomerular groups were not targeted by OSNs expressing similar ORs. Further, the correlation between sequence similarity and glomerular anatomical distances (r = 0.168; p = 0.00360) and the correlation between sequence similarity and PN response distance (r = 0.211; p = 0.000246) were very weak (Additional file 9: Figure S6c, d). These results show that neighboring glomerular groups building up the AL functional map are not targeted by OSNs expressing similar OR protein sequences. This relationship can be partly explained by the fact that ORs are so divergent that even within the same species sequence similarity does not necessarily indicate that similar odor response profiles are conserved. Taken together, the similarity of odor response profiles of the PNs innervating the three AL glomerular clusters can thus not simply be explained by OR protein sequence similarity but might be due to other, so far unknown, factors.
Conversion of the odor representation map from the AL to higher brain centers
How are the odor representations in the AL transferred to higher brain centers, where both convergence and divergence might take place? Jefferis et al. reconstructed the virtual odor response map by combining an anatomical map of PN axonal projections with OSN odor responses . However, no study so far has combined an anatomical projection map with actual odor responses of PNs to map odor representation in the MB calyx and LH. In the present study, we mapped virtual odor representation at the input level of the LH and MB by using the actual odor responses and morphology obtained from multiple PN classes.
Odor valence segregation and integration in the MB and LH
Next, we weighted our measured odor response intensities on the density maps of axonal projections of the 28 PNs (Fig. 5c). Then, we summed odor responses from the 28 PN classes and generated a virtual activity map for each odor (Fig. 5d).
Several previous studies suggested that the LH could be directly linked to innate behaviors [13, 18, 20, 21, 22]. Thus, reconstructing the odor representation map in the LH is key to understanding how flies perceive odor valence. Our reconstructed functional maps demonstrated that the attractive and aversive odors were classified into two separate clusters in the LH (Fig. 6a, b). Aversive odors activated ventral regions, while attractive odors activated posterior-dorsal regions (Fig. 6c). These results suggest that odor valence is integrated into two separate regions in the LH. The exceptions were hexanoic acid and propionic acid, which had been categorized as attractive odors in behavioral assays  but grouped with the aversive odors (Fig. 6a). This is likely due to our under sampling of acid responding glomeruli (ionotropic receptor projected glomeruli ) in the present study, as revealed by rather small total activities induced by these two odors (Additional file 5: Figure S2). Our results further showed that the two aversive odor groups separated in the representations in the AL converged to the ventral regions of the LH. This area might thus form an integration site for information brought by these aversive odors.
The MB, on the other hand, plays a role in circuit plasticity and is involved in learning and memory. Several studies have indicated that the MB is dispensable for innate odor-guided behavior [18, 19, 20, 21]. The reconstructed activity maps of the MB calyx revealed that the odors were classified into two clusters; one including exclusively attractive odors and the other including both attractive and aversive odors (Fig. 7a, b). In the MB, attractive odors in the first cluster activated the entire calyx regions including ventral portions, while aversive odors in the second cluster activated more focal regions near the base of the calyx (Fig. 7c). Interestingly, a gradual transition was observed from broad to focal activation patterns in the second cluster, corresponding to an attractive-to-aversive transition. These results suggest that there might be a concentric organization, where the peripheral region of the calyx contributes to attractive behavior while the focal region contributes to aversive, and thereby also suggest that the MB might play an auxiliary role underlying innate odor-guided behavior.
Finally, to address the question of whether labeled line pathways and combinatorial pathways that carry information of similar valence converge in higher brain centers, we focused on the aversive pathway, as our coverage of attractive odor processing channels was less comprehensive than that of aversive ones. Geosmin is processed via a dedicated pathway through the DA2 glomerulus, while eight other aversive odors (benzaldehyde, acetophenone, methyl salicylate, 2-methylphenol, 1-octanol, linalool, 1-octen-3-ol, and 1-hexanol) are represented by combinatorial activation of glomeruli [32, 34] (Fig. 2a, Additional file 5: Figure S2a). In the LH, the eight aversive odors activated the ventral part, while geosmin activated a ventral-posterior region. Although the regions activated by the eight odors extended to the entire region of the ventral LH, there was an overlap at the ventral-posterior region (Fig. 6c). This suggests that the activation patterns of these two different types of aversive odors partially overlap in the LH. In the MB, the geosmin activation pattern was more similar to other aversive odors (Fig. 7c). Our results thus show that the geosmin-activated regions in the LH and the MB calyx are not isolated, but overlap to some degree with regions activated by other aversive odors (Figs. 6c, 7c).
Temporal dynamics of odor representation
Although previous studies demonstrated that not only spatial but also temporal patterns of the neural code are essential for fine odor discrimination , we so far employed only spike numbers during 1-s odor stimuli as an indicator of odor response intensity. Now, we asked whether temporal patterns of PN responses add additional coding capacity to the system under study or not. First, we examined how odor representations evolve in the ensemble PN population over the course of the odor presentation by virtually assembling 31 PN class responses to 17 odors. We described trajectories of three representative odors (ethyl butyrate, acetophenone, and 1-octen-3-ol) by reconstructing PN population responses and visualizing them in a three-dimensional PC space (Additional file 12: Figure S9a). The trajectories showed that odor representations developed rapidly to reach a peak firing rate at 150 ms and slowly returned to the baseline. Next, we reconstructed trajectories for all odors in a two-dimensional PC space. Odors were rapidly separated with a similar pattern as observed by the clustering in Fig. 4a (Additional file 12: Figure S9b). Inter-odor distances between all odor pairs showed that separation peaked at 150 ms and gradually declined to the baseline at 2 s after the onset of odor stimulation. (Additional file 12: Figure S9c). Since previous studies have revealed that slow temporal patterns through de-correlation of mitral cells ensemble activities represent the key factor for fine odor discrimination (e.g., ), we examined developing changes of distance matrices during the course of olfactory stimulation. Distance matrices retained a similar degree of distinction during the 1-s odor stimulus, and the pattern disappeared at 1 s after the odor offset (Additional file 12: Figure S9d). We also confirmed similar time course changes in the distance matrices mapped on the LH and MB (Additional file 12: Figure S9e, f). Taken together, these results suggest that in the Drosophila AL, slow temporal patterns may not contribute to fine odor discrimination through de-correlation ensemble representation, as has been shown in the zebrafish olfactory system . Therefore, analyzing response intensity during 1-s odor stimulation was a suitable compromise to evaluate odor representation without losing critical information for fine odor discrimination. However, note that we could not address another important temporal factor, “oscillatory synchrony,” which was considered less prominent compared to other olfactory systems but demonstrated in a previous study in the Drosophila AL , as precise synchronization must be analyzed by local field potentials (LFPs) being recorded simultaneously as a reference.
We used in vivo whole cell patch-clamp recordings and stainings, mapped odor response profiles of 31 glomerular channels, and followed the flow of information into higher brain centers. We characterized odor response profiles of OSNs and PNs in ~30 glomeruli and found that the majority of odors are indeed represented in multiple glomeruli in a combinatorial manner (Fig. 2). Odor representation is thus largely combinatorial, while a few single odors highly important for survival and/or reproduction are processed through dedicated pathways. This organization enables the fly to discriminate a vast number of different odors, while maintaining a very robust and specific odor processing system for a few crucial, directly fitness-related odors. Our results show a rather conservative transformation between OSNs and PNs (Fig. 2) [55, 56]. Earlier studies focused on a few specific glomeruli and generalized the results to the entire system [44, 45]. Our study expands the scale to more than half of the glomerular channels and demonstrates that transformation of odor tuning curves is not homogeneous but glomerulus-dependent (Fig. 3).
Although our PN samplings were comprehensive, covering ~60% of glomerular channels, the depth of replication (the number of samples for each glomerulus) was moderate, as it was limited by technical difficulties. Therefore, inter-individual variation that may exist both in physiological odor responses (Additional file 2: Figure S1) and morphological axonal projection (Additional file 10: Figure S7) should be taken into account. However, our efforts to minimize such biases by the selection of analytical methods and application of registration techniques minimize ambiguity.
Whether a topographical representation that reflects the feature of olfactory stimuli is present is a central issue for olfactory coding. Interestingly, we found a few glomerular groups that are relevant for odor valence representation, often comprising neighboring glomeruli (Fig. 4). This organization tempts us to postulate that through evolution these glomeruli have divided from the same ancestral glomerulus, or one glomerulus has derived from the neighboring glomerulus. However, we found no evidence that these neighboring glomeruli are targeted by OSNs expressing phylogenetically related ORs (Additional file 8: Figure S5). Future studies should address the developmental mechanisms by which a glomerulus can be divided, which might provide an explanation as to why the AL has a topographical map.
Odor valence thus seems to be represented by different clusters of glomeruli in the AL. How is this information transformed into higher brain centers? When we reconstructed a functional map combining the anatomical map with odor response characteristics, we discovered segregated odor valence representation in the LH and partly in the MB calyx.
Several studies have proposed that the LH is the center underlying innate olfactory-dependent behavior [13, 18, 20, 21, 22]. Thus, it is postulated that the LH consists of a topographic map where each subregion is dedicated to a different type of innate behavior. An obvious example is found in pheromone channels that project separately from general odor channels in the LH (lateral protocerebrum in other insects) [11, 57, 58]. In Drosophila, dedicated pathways processing pheromone-related odors have been studied in detail [27, 28, 59, 60, 61, 62]. Other than pheromones, dedicated pathways processing aversive CO2 or acids via glomeruli V and DC4 respectively have been shown. These two pathways converge to the medial side of the LH, while information regarding ammonia and amines that are attractive to flies is processed via the VM1 glomerulus, for which PNs converge in the lateral posterior part of the LH . Our group recently showed that dedicated pathways processing aversive geosmin and parasitoid odor information via the DA2 and DL4 glomeruli respectively converge onto very similar regions in the LH . These odors are ecologically important for flies and directly trigger specific behaviors via dedicated neural pathways.
On the other hand, it is largely unknown how more general odors that induce innate odor-guided behavior are represented in the LH. A recent study revealed a topographical LH map relating to hedonic odor valence by imaging inhibitory PNs and ventro-lateral protocerebrum (vlpr) neurons (third order neurons projecting to the vlpr from the LH) . However, how odor information carried by excitatory PNs is represented in the LH has remained unknown. Our results for the first time show that information conveyed by excitatory PNs converges into two regions that could be integration sites for attractive and aversive odors. Attractive odor stimulation activated the posterior-dorsal region of the LH (Fig. 6). This region appears to correspond to the fruit odor processing region shown in . Conversely, aversive odors activated the ventral regions of the LH (Fig. 6). We found that two categories of aversive odors, represented in different clusters of AL glomeruli (Fig. 4a, c), activated largely overlapping regions in the ventral part of the LH (Fig. 6c). Furthermore, the labeled line pathway mediating geosmin aversive information also activated the ventral region of the LH, showing a partial overlap with the activation pattern for other aversive odors that are represented by combinatorial activation of glomeruli. Third order neurons that convey information from this region to the vlpr were also demonstrated in recent studies [12, 13]. In another study, third order neurons conveying attractive information from the dorsal region of the LH to the superior medial protocerebrum were found .
The MB consists of numerous Kenyon cells (KCs), among which combinatorial activation patterns of PNs are converted into a sparse representation [64, 65]. Since the connection patterns between PNs and KCs are likely to be random, there would be no hard-wired circuit connecting specific glomeruli to specific KCs [16, 66]. However, a recent study suggests that a KC receives inputs from several different glomeruli and integrates activation evoked by similar odors . In addition, a zonal organization has been found in the MB. The different zones are innervated by different types of KCs, and corresponding coarse regional PN projection patterns were also revealed [15, 17]. Our results revealed that odor valence segregation is partly observed also in the MB calyx (Fig. 7). The role of the MB for innate odor preference is less understood compared to its well-established role for olfactory learning. A previous study demonstrated that the MB is relevant for innate odor attraction, while it is irrelevant for innate odor aversion [18, 68]. Our results suggest that the MB is also able to partly contribute to categorize hedonic odor valence via a concentric organization. Such an organization in the calyx of the MB has already been observed at neuroanatomical levels, where it is proposed that the termination zones schematized as concentric circles from the inner to outer MB calyx correspond to a progression from the anterior-ventral to posterior-dorsal LH [11, 15]. Recent studies demonstrated that ensemble activity of MB output neurons can determine odor-guided behavior and that response profiles of these neurons are individually modulated, possibly by experience [69, 70, 71]. It is not clear how the odor map revealed in the MB calyx is read out by different types of KCs and then by the MB output neurons, but the MB output should be important to modulate innate odor-guided behavior by associative learning. Taken together, our results suggest that there is an organization in the LH and MB that allows the insect to categorize odor valence information. Beyond these anatomically defined areas, positive and negative odor information has to be integrated and put into the context of other sensory input and of the physiological state of the insect. This integration can then form the basis for a behavioral decision.
A fundamental question in olfactory coding is to understand how odor information initially separated by multiple receptor-glomerulus channels is integrated and read out by higher brain centers. In addition, little is known about how odor maps at higher processing centers are related to the animal's perception and innate behavior. The present study provides a major breakthrough to answer these questions, by mapping the olfactory information flow from the periphery to higher brain centers in unprecedented detail, using the Drosophila olfactory system. We found that attractive and aversive odors were separately represented by different clusters of glomeruli in the AL, and these representations were segregated into two different regions in the LH and partly separated in the MB. Since it has been increasingly evident that the anatomical organization of the olfactory system in mice has a similar architecture to that of the Drosophila olfactory system, our results may provide an important stepping stone toward understanding the common principle of the olfactory circuit design underlying innate olfactory behaviors.
Flies were maintained on conventional medium under a 12-h light/12-h dark cycle at 25 °C. Adult females 1–3 days after eclosion were used in all experiments. Fly lines used in this study were Canton-S (n = 42), GAL4-GH146/UAS-mCD8GFP (n = 25), GAL4-NP5221/UAS-mCD8GFP (n = 3), and GAL4-NP7217/UAS-mCD8GFP (n = 1).
In vivo whole cell patch-clamp recording
Patch-clamp recordings were performed with the same configuration as previously described (Seki et al. 2010) , except that an in vivo preparation was used and odor stimuli were given (Fig. 1b). The in vivo preparation was similar to the one used in imaging experiments (Stökl et al. 2010) . Briefly, flies were anesthetized on ice and fixed in a Plexiglas stage using a copper plate (G220-5, Athene Grids). The neck was fixed with a minutien pin (#26002-10, Fine Science Tools, Foster City, CA, USA), and the back of the head was glued with colophony resin (Royal Oak Rosinio, Royal Oak, Germany) to prevent moving of the head. The antennae were pulled forward with a fine metal wire (Rediohm-800, HP Reid Co., Palm Coast, FL, USA). After a few hours, plastic cover glass (L4193, Plano-em, Germany) with a hole covered by stretched parafilm was put on the head. The parafilm was ripped to expose the surface of the head, and the gap between the parafilm and the head capsule was filled with two component silicone (KWIK-SIL, World Precision Instruments, Sarasota, FL, USA). Then the head capsule was removed to expose the brain. At the same time, the brain was immersed with ringer solution containing (in millimoles) 130 NaCl, 5 KCl, 2 MgCl2, 2 CaCl2, 36 sucrose, and 5 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES); pH 7.3. Tracheae and muscles were removed for the both AL to be exposed, and the neurolemma was carefully removed with fine forceps. Cell bodies of PNs were visualized under an infrared DIC microscope (BX51WI Olympus, Hamburg, Germany) (Fig. 1c). Whole cell recording was made from a cell body of PN with a glass pipette containing the internal solution: (in millimoles) 140 potassium aspartate, 10 HEPES, 1 KCl, 4 Mg-ATP, 0.5 Na3GTP, 1 ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA), and 1 Lucifer yellow CH or 7 biocytin; pH 7.3. The membrane potential was recorded in current clamp mode using an EPC 10 patch-clamp amplifier (HEKA Elektronik, Lambrecht, Germany) with Patch-master software. The membrane potential was kept around –50 mV when injecting a small current if necessary. One neuron per brain was stimulated, recorded, and stained.
Odorants were diluted (10–2 v/v) except for geosmin (mostly 10–3, only three PNs were tested with 10–2 and they did not respond to 10–2 geosmin) in H2O (for acetic acid, propionic acid) or mineral oil (for all other odors). Odors used for PN recordings were acetic acid, benzaldehyde, 1-octanol, 2,3-butanedione, linalool, acetophenone, 1-octen-3-ol, ethyl butyrate, methyl salicylate, isopentyl acetate, hexanoic acid, 2-methylphenol, ethyl acetate, geranyl acetate, 1-hexanol, propionic acid, and geosmin. We tested all PNs with all odors and three controls (mineral oil, H2O, empty pipette), “20 stimuli as one set of stimuli,” at least one time with the fixed order as written above (i.e., acetic acid, benzaldehyde … geosmin, mineral oil, H2O, empty pipette) and proceeded the second round of stimuli until the recording stopped. If we lost recording without completing the whole odor set, we did not use those data for analysis and did not try recording another PN in the same preparation. The stimulus duration was 1 s, and the inter-stimulus interval was ~40 s. 10 μl of the diluted odors was put on the filter paper (Whatman), placed inside a glass Pasteur pipette. A stimulus controller (Stimulus Controller CS-55, Syntech, Kirchzarten, Germany) was used to produce a continuous airstream (1.2 L/min), which was joined with the airstream (0.12 L/min) from an empty pipette at 8 cm from the end of the delivery tube, and to switch to the stream from an odor stimulus pipette (0.12 L/min), which was joined to the continuous airstream at 6.5 cm from the end of the delivery tube. The end of the delivery tube, which was 5 mm in diameter, was placed 8 mm from the fly. The applied odor stimuli were thus additionally diluted approximately 10 times when merging to the continuous airstream.
Single sensillum recording (SSR)
SSR measurements were performed as described previously . The recording electrode and the reference electrode (inserted into the eye) were positioned under a microscope (Olympus BX51W1). The recording electrode was positioned by using a motorized, piezo-translator-equipped micromanipulator (Märzhäuser DC-3 K/PM-10, Wetzlar-Steindorf, Germany). The signal was amplified (Syntech UN-06), digitally converted (Syntech IDAC-4), and finally visualized and analyzed by using Syntech AutoSpike v3.2.
To visualize glomerular structures and regions in the higher brain centers, nc82 immunostainings were performed as previously described (Seki et al. 2010) . Additionally, biocytin-injected neurons were labeled with (1:500) Alexa Fluor 555 streptavidin (S-32355, Invitrogen), which was incubated with the secondary antibody for nc82, (1:200) goat anti-mouse Alexa Fluor 633 (A21052, Invitrogen). Images were taken with a Zeiss LSM510 META confocal microscope (Carl Zeiss, Jena, Germany). Axons of the stained PNs were reconstructed using the Amira 4.1.1 or 5.3.3 software (Visage Imaging, Berlin, Germany) as previously described . For neuron reconstruction the Skeleton plugin  of Amira was used. This module allowed the tracing process and created vertex models in which neurites were approximated by cylinders of a particular diameter. The innervated glomeruli were identified based on the glomerular map in , except that the “Vm7” and the unnamed “1” glomerulus in  were called “Vm7d” and “Vm7v” respectively, according to the recent nomenclature .
Registration method for PN axonal projection
Axonal projections of a single PN were reconstructed, and the MB calyx and LH structures were segmented in each brain. The MB calyx and LH were registered to a template brain using non-linear surface-matching methods. Image stacks were imported to Amira 5.6 (Fei), and neuronal arborizations and neuropils were manually segmented. For registration of datasets a label template of the central brain (MB calyx and LH) was chosen out of n = 58 preparations as described in . The warping of segmented labels of the MB calyx and LH onto the template was done in a two-step process: an affine transformation with 12 degrees of freedom (DOF) followed by an elastic registration using modules of Amira. The calculated transformation matrix was applied to the neuron reconstructions that were thus transformed to the template reference space.
PN responses were evaluated using spike numbers during the 1-s odor stimulus (0.05 s after the onset of stimuli to 1.05 s; 0.05 s was an approximate delay for the odors to reach the antennae and pulps from the onset of the stimulus pulse). Spike numbers used for analysis were made by subtraction with the control stimulus (either mineral oil or H2O, depending on the solvent for the odors). The response amplitude (spike numbers/s) was calculated using the responses in the first round of stimuli to achieve precise comparison among different PNs under virtually the exact same condition. The mean response amplitude was calculated if more than two PNs were recorded from the same glomerulus. Recording data were analyzed in Igor Pro (Wavemetrics, Lake Oswego, OR, USA) and MATLAB (The MathWorks, Natick, MA, USA) using custom software.
where N = number of odors (17 in this study), and r j is the odor response intensity of the neuron to odor j. Any values of r j <0 were set to zero before computing lifetime sparseness.
The density maps of axonal projections of PNs in the MB calyx and LH were generated similarly to the process in . (See Fig. 5, Additional file 11: Figure S8.) Each neuron was divided into segments of length equal to 0.5 μm by setting the interval of vertices as 0.5 μm. To generate a density map in the space of the MB calyx and LH, the voxel size was set to 1 μm3. The central points of each segment (vertex) were binned onto the grid of 135 × 135 × 105 voxels. A three-dimensional histogram was generated for each PN axonal innervation in the MB calyx region (55 × 46 × 32 μm) and the LH region (52 × 53 × 43 μm) by counting the vertices in each 1 μm3 voxel. Then the histogram was filtered three dimensionally with a boxcar filter (9 pixels). To visualize a two-dimensional projection, each column of voxels was integrated along the projection axis.
We generated the predicted odor response map as a functional map in the MB calyx and LH. (See Figs. 6 and 7). On the basis of density maps generated for each PN class, the odor response intensity value (any values <0 were set to zero) was multiplied by the density value and then summed with all the PN classes. The similarities of functional maps of each odor were compared by using cluster analysis with correlation distances and Ward’s classification method.
Physicochemical properties of 16 odors calculated with a set of 32 descriptors using Dragon were obtained from .
Phylogenetic tree and odorant receptor sequence analysis
Sixty aligned OR amino acid sequences of D. melanogaster were obtained from the Database of Olfactory Receptors (DOR; http://caps.ncbs.res.in/DOR/index.html) . The distance matrix and the phylogenetic tree were generated using Molecular Evolutionary Genetics Analysis version 7 (MEGA7) . Analyses were conducted using the Jones-Taylor-Thornton (JTT) matrix-based model . The analysis involved 60 amino acid sequences. All positions containing gaps and missing data were eliminated. The evolutionary history was inferred using the neighbor-joining method . The percentage of replicate trees in which the associated ORs clustered together in the bootstrap test (1000 replicates) is shown next to the branches . The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree.
The PCA and hierarchical cluster analysis were performed using MATLAB. Hierarchical cluster analysis was performed using cosine distance (Fig. 4a), Euclidean distance (Additional file 7: Figure S4, Additional file 8: Figure S5, Additional file 9: Figure S6) or correlation distance (Figs. 6, 7 and Additional file 11: Figure S8), and Ward’s classification method. One-way ANOSIM (Bray-Curtis similarity, sequential Bonferroni correction for ties, 10,000 permutations) was done with Paleontological STatistics (PAST) software (https://folk.uio.no/ohammer/past/).
Data are given as “mean ± s.d., n number”, unless otherwise noted.
This study was supported by the Max Planck Society (Max-Planck-Gesellschaft). We thank Johanna Langner for assisting with morphological reconstructions. We thank Silke Trautheim and Sabine Kaltofen for excellent technical assistance. The nc82 antibody generated by E. Buchner was obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the National Institute of Child Health and Human Development (NICHD) and maintained by the University of Iowa.
This study was funded by the Max-Planck-Gesellschaft (to BSH) and the Bundesministerium für Bildung und Forschung (to SS). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Availability of data and materials
The values of odor response intensity for PNs and OSNs used and analyzed in this study appear in Additional file 3: Table S2 and Additional file 4: Table S3 respectively. The other datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
YS helped with study conception and design, performed whole cell patch-clamp experiments, helped with acquisition of data and with analysis and interpretation of data, and drafted and revised the article; HD performed SSR experiments and helped with data acquisition and data analysis and interpretation; JR performed morphological analysis, analyzed and interpreted data, and revised the article; DW helped with study conception and design, interpretation of data, and revising of the article; SS helped with conception and design, interpretation of data, and revising of the article; BSH helped with conception and design, interpretation of data, and drafting and revising of the article, and provided intellectual and financial support. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 4.Masse NY, Turner GC, Jefferis GS. Olfactory information processing in Drosophila. Curr Biol. 2009;19(16):R700–13.Google Scholar
- 5.Couto A, Alenius M, Dickson BJ. Molecular, anatomical, and functional organization of the Drosophila olfactory system. Curr Biol. 2005;15(17):1535–47.Google Scholar
- 6.Fishilevich E, Vosshall LB. Genetic and functional subdivision of the Drosophila antennal lobe. Curr Biol. 2005;15(17):1548–53.Google Scholar
- 7.Silbering AF, Rytz R, Grosjean Y, Abuin L, Ramdya P, Jefferis GS, Benton R. Complementary function and integrated wiring of the evolutionarily distinct Drosophila olfactory subsystems. J Neurosci. 2011;31(38):13357–75.Google Scholar
- 8.Grabe V, Strutz A, Baschwitz A, Hansson BS, Sachse S. Digital in vivo 3D atlas of the antennal lobe of Drosophila melanogaster. J Comp Neurol. 2015;523(3):530–44.Google Scholar
- 9.Wong AM, Wang JW, Axel R. Spatial representation of the glomerular map in the Drosophila protocerebrum. Cell. 2002;109(2):229–41.Google Scholar
- 10.Marin EC, Jefferis GS, Komiyama T, Zhu H, Luo L. Representation of the glomerular olfactory map in the Drosophila brain. Cell. 2002;109(2):243–55.Google Scholar
- 11.Jefferis GS, Potter CJ, Chan AM, Marin EC, Rohlfing T, Maurer Jr CR, Luo L. Comprehensive maps of Drosophila higher olfactory centers: spatially segregated fruit and pheromone representation. Cell. 2007;128(6):1187–203.Google Scholar
- 13.Strutz A, Soelter J, Baschwitz A, Farhan A, Grabe V, Rybak J, Knaden M, Schmuker M, Hansson BS, Sachse S. Decoding odor quality and intensity in the Drosophila brain. Elife. 2014;3:e04147. doi: 10.7554/eLife.04147.
- 14.Tanaka NK, Endo K, Ito K. Organization of antennal lobe-associated neurons in adult Drosophila melanogaster brain. J Comp Neurol. 2012;520(18):4067–130.Google Scholar
- 15.Tanaka NK, Awasaki T, Shimada T, Ito K. Integration of chemosensory pathways in the Drosophila second-order olfactory centers. Curr Biol. 2004;14(6):449–57.Google Scholar
- 16.Caron SJ, Ruta V, Abbott LF, Axel R. Random convergence of olfactory inputs in the Drosophila mushroom body. Nature. 2013;497(7447):113–7.Google Scholar
- 17.Lin HH, Lai JS, Chin AL, Chen YC, Chiang AS. A map of olfactory representation in the Drosophila mushroom body. Cell. 2007;128(6):1205–17.Google Scholar
- 18.de Belle JS, Heisenberg M. Associative odor learning in Drosophila abolished by chemical ablation of mushroom bodies. Science. 1994;263(5147):692–5.Google Scholar
- 20.Heimbeck G, Bugnon V, Gendre N, Keller A, Stocker RF. A central neural circuit for experience-independent olfactory and courtship behavior in Drosophila melanogaster. Proc Natl Acad Sci U S A. 2001;98(26):15336–41.Google Scholar
- 21.Parnas M, Lin AC, Huetteroth W, Miesenbock G. Odor discrimination in Drosophila: from neural population codes to behavior. Neuron. 2013;79(5):932–44.Google Scholar
- 22.Min S, Ai M, Shin SA, Suh GS. Dedicated olfactory neurons mediating attraction behavior to ammonia and amines in Drosophila. Proc Natl Acad Sci U S A. 2013;110(14):E1321–9.Google Scholar
- 26.Kurtovic A, Widmer A, Dickson BJ. A single class of olfactory neurons mediates behavioural responses to a Drosophila sex pheromone. Nature. 2007;446(7135):542–6.Google Scholar
- 27.Ruta V, Datta SR, Vasconcelos ML, Freeland J, Looger LL, Axel R. A dimorphic pheromone circuit in Drosophila from sensory input to descending output. Nature. 2010;468(7324):686–90.Google Scholar
- 28.Dweck HK, Ebrahim SA, Thoma M, Mohamed AA, Keesey IW, Trona F, Lavista-Llanos S, Svatos A, Sachse S, Knaden M, et al. Pheromones mediating copulation and attraction in Drosophila. Proc Natl Acad Sci U S A. 2015;112(21):E2829–35.Google Scholar
- 29.Suh GS, Wong AM, Hergarden AC, Wang JW, Simon AF, Benzer S, Axel R, Anderson DJ. A single population of olfactory sensory neurons mediates an innate avoidance behaviour in Drosophila. Nature. 2004;431(7010):854–9.Google Scholar
- 30.Ai M, Min S, Grosjean Y, Leblanc C, Bell R, Benton R, Suh GS. Acid sensing by the Drosophila olfactory system. Nature. 2010;468(7324):691–5.Google Scholar
- 31.Ebrahim SA, Dweck HK, Stokl J, Hofferberth JE, Trona F, Weniger K, Rybak J, Seki Y, Stensmyr MC, Sachse S, et al. Drosophila avoids parasitoids by sensing their semiochemicals via a dedicated olfactory circuit. PLoS Biol. 2015;13(12):e1002318.Google Scholar
- 32.Stensmyr MC, Dweck HK, Farhan A, Ibba I, Strutz A, Mukunda L, Linz J, Grabe V, Steck K, Lavista-Llanos S, et al. A conserved dedicated olfactory circuit for detecting harmful microbes in Drosophila. Cell. 2012;151(6):1345–57.Google Scholar
- 33.Stökl J, Strutz A, Dafni A, Svatos A, Doubsky J, Knaden M, Sachse S, Hansson BS, Stensmyr MC. A deceptive pollination system targeting drosophilids through olfactory mimicry of yeast. Curr Biol. 2010;20(20):1846–52.Google Scholar
- 35.Semmelhack JL, Wang JW. Select Drosophila glomeruli mediate innate olfactory attraction and aversion. Nature. 2009;459(7244):218–23.Google Scholar
- 36.Badel L, Ohta K, Tsuchimoto Y, Kazama H. Decoding of context-dependent olfactory behavior in Drosophila. Neuron. 2016;91(1):155–67.Google Scholar
- 37.Stensmyr MC, Giordano E, Balloi A, Angioy AM, Hansson BS. Novel natural ligands for Drosophila olfactory receptor neurones. J Exp Biol. 2003;206(Pt 4):715–24.Google Scholar
- 38.Steck K, Veit D, Grandy R, Badia SB, Mathews Z, Verschure P, Hansson BS, Knaden M. A high-throughput behavioral paradigm for Drosophila olfaction —The Flywalk. Sci Rep. 2012;2:361.Google Scholar
- 39.Thoma M, Hansson BS, Knaden M. Compound valence is conserved in binary odor mixtures in Drosophila melanogaster. J Exp Biol. 2014;217(Pt 20):3645–55.Google Scholar
- 41.Jefferis GS, Marin EC, Stocker RF, Luo L. Target neuron prespecification in the olfactory map of Drosophila. Nature. 2001;414(6860):204–8.Google Scholar
- 42.Grabe V, Baschwitz A, Dweck HK, Lavista-Llanos S, Hansson BS, Sachse S. Elucidating the neuronal architecture of olfactory glomeruli in the Drosophila antennal lobe. Cell Rep. 2016;16(12):3401–13.Google Scholar
- 44.Wilson RI, Turner GC, Laurent G. Transformation of olfactory representations in the Drosophila antennal lobe. Science. 2004;303(5656):366–70.Google Scholar
- 45.Bhandawat V, Olsen SR, Gouwens NW, Schlief ML, Wilson RI. Sensory processing in the Drosophila antennal lobe increases reliability and separability of ensemble odor representations. Nat Neurosci. 2007;10(11):1474–82.Google Scholar
- 47.Shanbhag S, Müller B, Steinbrecht R. Atlas of olfactory organs of Drosophila melanogaster: 1. Types, external organization, innervation and distribution of olfactory sensilla. Int J Insect Morphol Embryol. 1999;28(4):377–97.Google Scholar
- 48.Dweck HK, Ebrahim SA, Kromann S, Bown D, Hillbur Y, Sachse S, Hansson BS, Stensmyr MC. Olfactory preference for egg laying on citrus substrates in Drosophila. Curr Biol. 2013;23(24):2472–80.Google Scholar
- 49.Olsen SR, Wilson RI. Lateral presynaptic inhibition mediates gain control in an olfactory circuit. Nature. 2008;452(7190):956–60.Google Scholar
- 50.Olsen SR, Wilson RI. Lateral presynaptic inhibition mediates gain control in an olfactory circuit. Nature. 008;452(7190):956–60.Google Scholar
- 54.Tanaka NK, Ito K, Stopfer M. Odor-evoked neural oscillations in Drosophila are mediated by widely branching interneurons. J Neurosci. 2009;29(26):8595–603.Google Scholar
- 57.Homberg U, Montague RA, Hildebrand JG. Anatomy of antenno-cerebral pathways in the brain of the sphinx moth Manduca sexta. Cell Tissue Res. 1988;254(2):255–81.Google Scholar
- 58.Seki Y, Aonuma H, Kanzaki R. Pheromone processing center in the protocerebrum of Bombyx mori revealed by nitric oxide-induced anti-cGMP immunocytochemistry. J Comp Neurol. 2005;481(4):340–51.Google Scholar
- 59.Grosjean Y, Rytz R, Farine JP, Abuin L, Cortot J, Jefferis GS, Benton R. An olfactory receptor for food-derived odours promotes male courtship in Drosophila. Nature. 2011;478(7368):236–40.Google Scholar
- 60.Datta SR, Vasconcelos ML, Ruta V, Luo S, Wong A, Demir E, Flores J, Balonze K, Dickson BJ, Axel R. The Drosophila pheromone cVA activates a sexually dimorphic neural circuit. Nature. 2008;452(7186):473–7.Google Scholar
- 62.Ronderos DS, Lin CC, Potter CJ, Smith DP. Farnesol-detecting olfactory neurons in Drosophila. J Neurosci. 2014;34(11):3959–68.Google Scholar
- 64.Honegger KS, Campbell RA, Turner GC. Cellular-resolution population imaging reveals robust sparse coding in the Drosophila mushroom body. J Neurosci. 2011;31(33):11772–85.Google Scholar
- 65.Wang Y, Guo HF, Pologruto TA, Hannan F, Hakker I, Svoboda K, Zhong Y. Stereotyped odor-evoked activity in the mushroom body of Drosophila revealed by green fluorescent protein-based Ca2+ imaging. J Neurosci. 2004;24(29):6507–14.Google Scholar
- 66.Murthy M, Fiete I, Laurent G. Testing odor response stereotypy in the Drosophila mushroom body. Neuron. 2008;59(6):1009–23.Google Scholar
- 68.Wang Y, Chiang AS, Xia S, Kitamoto T, Tully T, Zhong Y. Blockade of neurotransmission in Drosophila mushroom bodies impairs odor attraction, but not repulsion. Curr Biol. 2003;13(21):1900–4.Google Scholar
- 69.Aso Y, Sitaraman D, Ichinose T, Kaun KR, Vogt K, Belliart-Guerin G, Placais PY, Robie AA, Yamagata N, Schnaitmann C, et al. Mushroom body output neurons encode valence and guide memory-based action selection in Drosophila. Elife. 2014;3:e04580.Google Scholar
- 72.Seki Y, Rybak J, Wicher D, Sachse S, Hansson BS. Physiological and morphological characterization of local interneurons in the Drosophila antennal lobe. J Neurophysiol. 2010;104(2):1007–19.Google Scholar
- 75.Nagarathnam B, Karpe SD, Harini K, Sankar K, Iftekhar M, Rajesh D, Giji S, Archunan G, Balakrishnan V, Gromiha MM, et al. DOR — a database of olfactory receptors — integrated repository for sequence and secondary structural information of olfactory receptors in selected eukaryotic genomes. Bioinform Biol Insights. 2014;8:147–58.PubMedPubMedCentralGoogle Scholar
- 76.Sudhir Kumar, Glen Stecher, Koichiro Tamura. MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets. Mol Biol Evol. 2016;33(7):1870–1874.Google Scholar
- 77.Jones DT, Taylor WR, Thornton JM. The rapid generation of mutation data matrices from protein sequences. Comput Appl Biosci. 1992;8(3):275–82.Google Scholar
- 78.Saitou N, Nei M. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol. 1987;4(4):406–25.Google Scholar
- 79.Felsenstein J. Confidence limits on phylogenies: an approach using the bootstrap. Evolution. 1985;39(4):783–91.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.