Animals discriminate stimuli, learn their predictive value and use this knowledge to modify their behavior. In Drosophila, the mushroom body (MB) plays a key role in these processes. Sensory stimuli are sparsely represented by ∼2000 Kenyon cells, which converge onto 34 output neurons (MBONs) of 21 types. We studied the role of MBONs in several associative learning tasks and in sleep regulation, revealing the extent to which information flow is segregated into distinct channels and suggesting possible roles for the multi-layered MBON network. We also show that optogenetic activation of MBONs can, depending on cell type, induce repulsion or attraction in flies. The behavioral effects of MBON perturbation are combinatorial, suggesting that the MBON ensemble collectively represents valence. We propose that local, stimulus-specific dopaminergic modulation selectively alters the balance within the MBON network for those stimuli. Our results suggest that valence encoded by the MBON ensemble biases memory-based action selection.https://doi.org/10.7554/eLife.04580.001
An animal's survival depends on its ability to respond appropriately to its environment, approaching stimuli that signal rewards and avoiding any that warn of potential threats. In fruit flies, this behavior requires activity in a region of the brain called the mushroom body, which processes sensory information and uses that information to influence responses to stimuli.
Aso et al. recently mapped the mushroom body of the fruit fly in its entirety. This work showed, among other things, that the mushroom body contained 21 different types of output neurons. Building on this work, Aso et al. have started to work out how this circuitry enables flies to learn to associate a stimulus, such as an odor, with an outcome, such as the presence of food.
Two complementary techniques—the use of molecular genetics to block neuronal activity, and the use of light to activate neurons (a technique called optogenetics)—were employed to study the roles performed by the output neurons in the mushroom body. Results revealed that distinct groups of output cells must be activated for flies to avoid—as opposed to approach—odors. Moreover, the same output neurons are used to avoid both odors and colors that have been associated with punishment. Together, these results indicate that the output cells do not encode the identity of stimuli: rather, they signal whether a stimulus should be approached or avoided. The output cells also regulate the amount of sleep taken by the fly, which is consistent with the mushroom body having a broader role in regulating the fly's internal state.
The results of these experiments—combined with new knowledge about the detailed structure of the mushroom body—lay the foundations for new studies that explore associative learning at the level of individual circuits and their component cells. Given that the organization of the mushroom body has much in common with that of the mammalian brain, these studies should provide insights into the fundamental principles that underpin learning and memory in other species, including humans.https://doi.org/10.7554/eLife.04580.002
To survive in a dynamic environment, an animal must discover and remember the outcomes associated with the stimuli it encounters. It then needs to choose adaptive behaviors, such as approaching cues that predict food and avoiding cues that predict danger. The neural computations involved in using such memory-based valuation of sensory cues to guide action selection require at least three processes: (1) sensory processing to represent the identity of environmental stimuli and distinguish among them; (2) an adaptive mechanism to assign valence—positive or negative survival value—to a sensory stimulus, store that information, and recall it when that same stimulus is encountered again; and (3) decision mechanisms that receive and integrate information about the valence of learned stimuli and then bias behavioral output. To understand such decision-making processes, one approach is to locate the sites of synaptic plasticity underlying memory formation, identify the postsynaptic neurons that transmit stored information to the downstream circuit and discover how their altered activities bias behavior.
The mushroom body (MB) is the main center of associative memory in insect brains (de Belle and Heisenberg, 1994, Heisenberg et al., 1985; Dubnau et al., 2001; McGuire et al., 2001). While the MB processes several modalities of sensory information and regulates locomotion and sleep (Martin et al., 1998; Liu et al., 1999; Joiner et al., 2006; Pitman et al., 2006; Zhang et al., 2007; Hong et al., 2008; Vogt et al., 2014), MB function has been most extensively studied in the context of olfactory memory—specifically, associating olfactory stimuli with environmental conditions in order to guide behavior. In Drosophila, olfactory information is delivered to the MB by projection neurons from each of ∼50 antennal lobe glomeruli (Marin et al., 2002; Wong et al., 2002; Jefferis et al., 2007; Lin et al., 2007; Vosshall and Stocker, 2007; Yu et al., 2010). Connections between the projection neurons and the ∼2000 Kenyon cells (KCs), the neurons whose parallel axonal fibers form the MB lobes (Crittenden et al., 1998; Aso et al., 2009), are not stereotyped (Figure 1A) (Murthy et al., 2008; Caron et al., 2013); that is, individual flies show distinct wiring patterns between projection neurons and KCs. Sparse activity of the KCs represents the identity of odors (Laurent and Naraghi, 1994; Perez-Orive et al., 2002; Turner et al., 2008). The output of the MB is conveyed to the rest of the brain by a remarkably small number of neurons—34 cells of 21 cell types per brain hemisphere (Figure 1B, Table 1) (Aso et al., 2014).
The information flow from the KCs to the MB output neurons (MBONs) has been proposed to transform the representation of odor identity to more abstract information, such as the valence of an odor based on prior experience (See discussion in Aso et al., 2014). In contrast to KCs, MBONs have broadly tuned odor responses; any given odor results in a response in most MBONs, although the magnitude of the response varies among MBON cell types (Hige et al., unpublished) (Cassenaer and Laurent, 2012). Unlike the stereotyped response to odors of the olfactory projection neurons that deliver odor information to the MB, the odor tuning of the MBONs is modified by plasticity and varies significantly between individual flies, suggesting that MBONs change their response to odors based on experience (Hige et al., unpublished).
For olfactory associative memory in Drosophila, multiple lines of evidence are consistent with a model in which dopamine-dependent plasticity in the presynaptic terminals of KCs alters the strength of synapses onto MBON dendrites. This is thought to provide a mechanism by which the response of MBONs to a specific odor could represent that odor's predictive value. D1-like dopamine receptors and components of the cAMP signaling pathway, such as the Ca2+/Calmodulin-responsive adenylate cyclase encoded by the rutabaga gene, are required specifically in the KCs for memory formation (Livingstone et al., 1984; Zars et al., 2000; Schwaerzel et al., 2003; Kim et al., 2007; McGuire et al., 2003; Gervasi et al., 2010; Qin et al., 2012) and rutabaga was shown to be required for the establishment of the differences in MBON odor tuning between individuals (Hige et al., unpublished). Reward and punishment recruit distinct sets of dopaminergic neurons (DANs) that project to specific regions in the MB lobes (Mao and Davis, 2009; Burke et al., 2012; Liu et al., 2012). Moreover, exogenous activation of these DANs can substitute for reinforcing stimuli to induce either appetitive or aversive memory, depending on DAN cell type (Figure 1A) (Yamagata et al., in press, Perisse et al., 2013; Schroll et al., 2006; Claridge-Chang et al., 2009; Aso et al., 2010, 2012; Liu et al., 2012; Burke et al., 2012). In sum, while the identity of the learned odor is likely encoded by the small subset of KCs activated by that odor, whether dopamine-mediated modulation assigns positive or negative valence to that odor would be determined by where in the MB lobes KC-MBON synapses are modulated and thus which MBON cell types alter their response to the learned odor.
Combining the above observations with our comprehensive anatomical characterization of MB inputs and outputs (Aso et al., 2014) lays the groundwork for testing models of how the MB functions as a whole. We suggest that each of the 15 MB compartments—regions along the MB lobes defined by the arborization patterns of MBONs and DANs (see Figure 1)—functions as an elemental valuation system that receives reward or punishment signals and translates the pattern of KC activity to a MBON output that serves to bias behavior by altering either attraction or aversion. This view implies that multiple independent valuation modules for positive or negative experiences coexist in the MB lobes, raising the question of how the outputs across all the modules are integrated to result in a coherent, adaptive biasing of behavior.
Although several MBON cell types have been shown to play a role in associative odor memory (Sejourne et al., 2011; Pai et al., 2013; Placais et al., 2013), the functions of most MBONs have not been studied. Based on our anatomical analyses (Aso et al., 2014), we believe that just 34 MBONs of 21 types provide the sole output pathways from the MB lobes. To gain mechanistic insight into how the ensemble of MBONs biases behavior, we would first like to know the nature of the information conveyed by individual MBONs and the extent to which their functions are specialized or segregated into different information channels. Then we need to discover how the activities of individual MBONs contribute to influence the behavior exerted by the complete population of MBONs. Thus, in order to understand how memory is translated into changes in behavior, we need to have experimental access to a comprehensive set of MBONs and investigate how the outputs from different MBONs bias behavior, singly and in combination.
In the accompanying paper (Aso et al., 2014), we describe the detailed anatomy of the DANs and MBONs (Figure 1) and the generation of intersectional split-GAL4 driver lines to facilitate their study. All but one of the 21 MBON cell types consists of only one or two cells per hemisphere (Table 1). Dendrites of MBONs that use the same neurotransmitter—GABA, glutamate or acetylcholine—are spatially clustered in the MB lobes. Intriguingly, this spatial clustering resembles the innervation patterns of modulatory input by two clusters of dopaminergic neurons, PPL1 and PAM. MBONs have their axonal terminals in a small number of brain regions, but their projection patterns also suggest pathways for relaying signals between compartments of the MB lobes; three MBONs send direct projections to the MB lobes and several other MBONs appear to target the dendrites of specific DANs.
Our split-GAL4 drivers give us the capability to express genetically encoded effectors in identified MBONs to modify their function. In this study, we examine the roles of specific MBONs in various learning and memory tasks as well as in the regulation of locomotion and sleep. We also studied whether direct activation of specific MBONs are sufficient to elicit approach or avoidance. Our results indicate that the ensemble of MBONs does not directly specify particular motor patterns. Instead, MBONs collectively bias behavior by conveying the valence of learned sensory stimuli, irrespective of the modality of the stimulus or the specific reward or punishment used during conditioning.
A powerful strategy to discover if a neuronal population plays a role in a particular behavior is to observe the effects of inactivating or activating those neurons. A genetic driver can be used to express an exogenous protein that either promotes or blocks neuronal activity. By repeating such manipulations with a large collection of drivers, each specific for a different set of neurons, one can in principle discover cell types required for a particular behavior. This is analogous to a screen to identify genes that when mutated disrupt a cellular function, but it is the activity of cells—rather than that of genes—being altered. This approach has been widely used in Drosophila (Reviewed in Venken et al., 2011; Griffith, 2012).
There are many challenges in carrying out such an approach. As in all biological systems, we expect extensive resiliency to perturbation. Such robustness might mask the effects of manipulating the activity of a small population of neurons, making them undetectable above the normal variation between animals. In addition to these inherent limitations, the genetic tools at our disposal have often been inadequate. In this study, we have used improved genetic tools and employed several strategies to mitigate their remaining limitations, as detailed below and in ‘Materials and methods’. By assaying many different behaviors with the same genetic reagents, we were better able to evaluate the specificity of the behavioral effects we observed. Our experimental design placed an emphasis on avoiding false positives. Extensive analysis was restricted to MBON lines showing a phenotype in the initial screens and our scoring criteria were conservative. Consequently, we are likely to have missed detecting some cell types with effects on the behavior under assay.
We sought to first determine the nature of the information conveyed by MBONs. In the most widely used olfactory conditioning assay, memory is assessed after training by allowing flies to distribute between two arms of a T-maze: one arm perfused with a control odor and the other arm with an odor that had been previously associated with punishment or reward. If the valence of the learned odor were encoded by the altered activities of specific MBONs, artificial activation of those MBONs in untrained flies in the absence of odor presentation would be expected to result in avoidance or approach behavior that mimicked the conditioned odor response. To test this hypothesis, we used a circular arena in which groups of flies expressing the red-shifted channelrhodopsin CsChrimson (Klapoetke et al., 2014) in MBONs were allowed to freely distribute between dark quadrants and quadrants with activating red light (Figure 2A; see ‘Materials and methods’ for details). Activating sensory neurons for CO2 or bitter taste in this manner induced strong avoidance (Figure 2B), consistent with previous reports (Suh et al., 2007). By testing our collection of MBON split-GAL4 drivers in this assay, we found cell types whose activation resulted in avoidance of the red light and others whose activation led to attraction (Figure 2C). Behavioral valence was highly correlated with MBON transmitter type: all MBONs eliciting aversion were glutamatergic and all the MBONs eliciting attraction were either GABAergic or cholinergic. We selected two split-GAL4 drivers for each neurotransmitter type for further analysis, choosing lines that gave robust phenotypes in the initial screening and that showed highly specific expression patterns based on direct assessment of CsChrimson expression (Figure 3A–F). The phenotypes of these lines were reproducible (Figure 3G–I) and none of the drivers showed significant preference to the red light in the absence of the CsChrimson effector (Figure 3G), confirming that phototaxis was limited at the wavelength and intensity of light used. These results demonstrate that activation of MBONs is itself sufficient to elicit either approach or avoidance, depending on the cell type.
Although activation of individual cell types (see Figure 3D,E) can result in robust phenotypes, some of the strongest effects were observed with drivers that express in combinations of MBONs. For example, MB011B and MB052B, which drive expression in groups of three or five cell types that use the same transmitter (the glutamatergic M4/M6 and cholinergic V2 clusters, respectively; Table 1), caused strong responses, while activation with drivers for smaller subsets or individual cell types within these groups had much reduced effects (Figure 2C). Although activation of the single cell types MBON-γ1pedc>α/β or MBON-γ4>γ1γ2 had a significant effect, these neurons send axonal projections to other compartments in the MB lobes, giving them the potential to directly influence the activity of additional MBON cell types. Given that multiple MBONs can independently contribute to behavioral valence as measured by attraction vs repulsion, we sought to determine how conflicting or consonant information from multiple MBONs is integrated to bias behavior. We combined two split-GAL4 drivers for cell types with either similar or opposite effects in the same fly (Figure 4, Figure 4—figure supplement 1; ‘Materials and methods’). When combining drivers eliciting similar responses, flies generally showed a stronger response than to either driver alone. Conversely, co-activation of MBON cell types with opposing effects resulted in intermediate responses. Together, these data are consistent with a simple combinatorial model of valence integration.
How do MBONs bias behavior to cause approach or avoidance? A conditioned response to a learned odor is unlikely to be achieved by eliciting a predetermined motor pattern. To carry out appropriate changes in speed and direction, a fly needs to evaluate both the valence of the learned odor and its own trajectory relative to the location of the odor source. To assess which aspects of locomotion are modified by MBON activation to generate a bias between approach and avoidance, we analyzed the behavior of flies in a 10 mm-wide graded lighting choice zone centered on the border between dark and illuminated quadrants. We tracked the trajectories of individual flies (Figure 5A–C) and calculated the fraction of flies crossing the choice zone from the ‘light-off’ to the ‘light-on’ area (and vice versa) as well as the fraction of flies changing direction in the choice zone thus returning to the area they came from.
Most control flies entering the choice zone from either the light-off or light-on side continued moving forward, crossing into the other side (Figure 5D,E; Video 1). Flies expressing CsChrimson in either GABAergic (MB112C or MB083) or cholinergic (MB077B or MB052B) MBONs behaved similarly to control flies when entering the choice zone from the light-off area. When they entered from the light-on side, these flies showed a slight tendency to turn around in the choice zone (Figure 5D) consistent with their preference for illuminated areas (Figure 2C). Flies expressing CsChrimson in a combination of GABAergic and cholinergic MBONs (MB112C plus MB077B), which displayed the strongest preference for lighted areas (Figure 4), also showed the highest rates of exiting to the light-on side when entering the choice zone from the light-off area (Figure 5D). On the other hand, flies expressing CsChrimson in glutamatergic MBONs (MB434B, MB011B or a combination of them) frequently turned around in the choice zone when entering from the light-off side while crossing into the light-off area when entering the choice zone from the illuminated side (Figure 5D,E; Video 2 and Video 3), behaviors that are consistent with these flies' avoidance of illuminated areas (Figure 2C).
We next calculated the fraction of flies exiting the choice zone into the light-on side irrespective of its entry direction and found this ‘choice probability’ to be highly correlated with preference index (Figure 5F). In contrast, we found little correlation between preference index and either mean walking speed or mean angular velocity, an indicator of turning probability, of flies in illuminated quadrants (Figure 5G,H). Flies are able to adjust these parameters to execute avoidance behaviors in other contexts; we found that activation of some chemosensory neurons and projection neuron combinations that repelled flies significantly altered both walking and angular speed (Figure 5—figure supplement 1; Video 4 and Video 5). Our results indicate that MBON activity biases the direction that flies turn in the choice zone, thereby biasing the direction in which they exit that zone. We did not observe a stereotyped turning behavior in the choice zones; more specifically, the time between entering the choice zone and making a turn as well as the precise direction of the turn varied (Figure 5E; Videos 2 and 3). Moreover, flies displayed apparently normal behavior within the uniform illumination of lighted quadrants (Video 2 and 3), showing no apparent increase in their speed and turning rates. These observations support the view that MBONs represent valence, abstract information that serves to bias—rather than direct—specific motor patterns.
We asked which MBONs were required for aversive and appetitive odor memory, using a well-established discriminative olfactory learning paradigm (Figure 6A) (Tempel et al., 1983; Tully and Quinn, 1985; Schwaerzel et al., 2003; Gerber et al., 2004; Davis, 2005). In this paradigm, flies are exposed to an odor together with an unconditioned stimulus (US) of either an electric shock punishment or a sugar reward, and then to another odor without the US. The ‘trained’ flies are tested at a later time to determine if they exhibit a differential response to the two odors, which is taken as indication of memory formation.
In the first set of experiments we assayed memory 2 hr after training, using a set of 23 split-GAL4 lines to transiently block neuronal activity in different subsets of MBON cell types. Two memory processes that are thought to rely on different molecular and circuit mechanisms, anesthesia resistant memory and anesthesia sensitive memory, contribute to memory at this retention time (Dudai et al., 1976; Folkers et al., 1993; Isabel et al., 2004; Krashes and Waddell, 2008; Aso et al., 2010; Pitman et al., 2011; Knapek et al., 2011). We blocked neuronal function throughout the training, retention and test periods. Thus we expect to detect impairments in any phase of memory processing, including formation, consolidation and retrieval.
We first screened the lines using a strong Shibirets1 effector (pJFRC100-20XUAS-TTS-Shibire-ts1-p10 in VK00005). Because some lines had phenotypes at the permissive temperature, presumably due to the effector's high level of expression, we retested the nine lines that showed a reduction in memory performance with a weaker Shibirets1 effector (UAS-Shi x1; see Figure 6 legend and ‘Materials and methods’). With these parameters, only one line, MB112C, also showed significant memory impairment (Figure 6B). The experimental MB112C flies showed significantly lower memory performance than genetic control groups at the restrictive temperature (Figure 6C), but not at the permissive temperature (Figure 6D). These flies displayed normal shock and odor avoidance at the restrictive temperature (Figure 6E,F), indicating that the observed memory impairment was not due to a defect in sensory or motor pathways. MB112C drives expression in MBON-γ1pedc>α/β (Figure 6—figure supplement 1A). We confirmed the requirement for this cell type using a second driver, MB085C, not included in the original 23 lines screened (Figure 6C; Figure 6—figure supplement 1A).
One of the PPL1 cluster dopaminergic neurons, PPL1-γ1pedc (also known as MB-MP1), innervates the same MB compartments as MBON-γ1pedc>α/β. Blocking PPL1-γ1pedc activity, using the MB438B split-GAL4 driver, also impaired aversive memory (Figure 6C–F). Conversely, activation of PPL1-γ1pedc with the temperature gated cation channel dTrpA1 (Hamada et al., 2008) substituted for electric shock as the unconditioned stimulus (US), inducing robust aversive odor memory (Figure 6G). This confirms a conclusion reached using less specific enhancer trap GAL4 drivers (Aso et al., 2010, 2012). Consistent with these observations, restoring expression of the D1-like dopamine receptor specifically in the γ Kenyon cells has been shown to rescue the aversive odor memory defect of a receptor mutant (Qin et al., 2012). MBON-γ1pedc>α/β is immunoreactive to GABA (Figure 6—figure supplement 1C) and is one of only three MBON cell types with axon terminals within the MB lobes (Figure 6H; Figure 6—figure supplement 1D,E) (Aso et al., 2014). We visualized the single-cell morphologies of MBON-γ1pedc>α/β (Figure 6I) and PPL1-γ1pedc (Figure 6J) and using two-color labeling confirmed that the axon terminals of the DAN (one or two cells per brain hemisphere) precisely overlap with the dendrites of the MBON (a single cell per brain hemisphere) in the MB pedunculus and γ1 compartment (Figure 6K). This establishes two essential components of a circuit for 2-hr aversive odor memory.
A larger set of MBONs was involved in 2-hr appetitive olfactory memory (Figure 7). Inactivation of the neurons represented in 12 out of the 23 split-GAL4 lines tested showed an effect in the initial screening with the strong Shibirets1 effector; when retested with the weaker effector, eight of these lines produced a significant impairment (Figure 7, Figure 7—Figure supplement 1). All eight lines displayed normal memory at the permissive temperature (Figure 7C) and normal attraction to sugar at the restrictive temperature (Figure 7D). Five of these lines (MB082C, MB093C, MB018B, MB051B and MB077B) express in subsets of the so-called V3/V4 cluster of cholinergic MBONs (MBON-γ2α′1, MBON-α′2 and MBON-α3; Table 1), establishing a role for this group of MBONs in 2-hr appetitive memory. MB310C labels the glutamatergic MBON-α1. MB011B labels three types of M4/M6 cluster glutamatergic MBONs (MBON-γ5β′2a, MBON-β′2mp and MBON-β′2mp_bilateral; Table 1), suggesting a role for one or more of these cell types. Two lines that express in subsets of MB011B cell types, MB210B and MB002B, showed a memory defect in the primary screening but failed to reach statistical significance when retested with the weaker effector. Similarly, we observed that activating CsChrimson with MB011B, but not with MB210B or MB002B, produced a significant aversive effect (see Figure 2C). These data are consistent with an additive role of these MBONs on behavior, which is further supported by the anatomical observation that the axon terminals of some of these MBONs converge to the same areas outside the MB (Figure 7E,F) (Aso et al., 2014).
The MB has been proposed to play diverse roles in visual behaviors including context generalization and sensory preconditioning between olfactory and visual cues (Liu et al., 1999; Zhang et al., 2013, 2007; Brembs, 2009; van Swinderen et al., 2009). Using visual learning assays in which flies are trained to associate a color (blue or green) with either an electric shock punishment or a sugar reward (Figure 8A and Figure 9A), Vogt et al., 2014 demonstrated that γ-lobe KCs are required for immediate visual associative memory and that activation of specific subsets of PPL1 and PAM cluster DANs can substitute as the US for electric shock or sugar, respectively. Here we use the same assays to ask which MBONs are required for visual memory. We used the strong Shits effector (pJFRC100-20XUAS-TTS-Shibire-ts1-p10 in VK00005) to silence MBONs (Materials and methods).
For aversive visual memory (Figure 8B), we identified one driver line, MB112C, that labels MBON-γ1pedc>α/β. We confirmed the requirement for this cell type with two additional split-GAL4 lines, MB262B (Figure 8C; Figure 8—figure supplement 1) and MB085C (data not shown). This is the same MBON we found to be important for olfactory aversive memory (Figure 6) and so we asked if the same DAN as in olfactory memory was likewise required. Indeed, we found that MB438B, a driver line for PPL1-γ1pedc, showed a significant impairment (Figure 8C; Figure 8—figure supplement 1). The experimental MB112C, MB262B and MB438B lines showed normal memory at the permissive temperature (Figure 8C) and normal shock avoidance at the restrictive temperature (Figure 8D). Our finding that the same MBON and its modulatory DAN were required for both visual and olfactory aversive memory suggests that this local MB circuit is important for aversive memory in general, rather than specifically for a particular sensory modality.
For appetitive visual memory, we identified five drivers with significant impairments: MB434B, MB011B, MB210B, MB542B and MB052B (Figure 9B,C; Figure 9—figure supplement 1). These lines showed normal memory at the restrictive temperature in the absence of the effector (Figure 9C) and at the permissive temperature with the effector (Figure 9D), as well as normal sugar attraction at restrictive temperature (Figure 9E). The anatomy of the cell types in these driver lines is illustrated in Figure 9F,G. Unlike in the case of aversive memory, the cell types required for appetitive memory differed somewhat between visual and olfactory modalities. Nevertheless, these modalities both employed the M4/M6 cluster glutamatergic MBONs labeled in MB011B and MB210B (MBON-γ5β′2a, MBON-β′2mp and MBON-β′2mp_bilateral; Table 1). Moreover, the observation that multiple cholinergic and glutamatergic MBONs play a role in appetitive memory, but not aversive memory, was shared across modalities.
Repeated pairing of an odor and an electric shock, with inter-trial rest intervals, results in protein synthesis dependent aversive long-term memory (LTM) (Tully et al., 1994). The molecular and cellular mechanisms underlying aversive LTM are known to differ from those responsible for memories with shorter retention times (Yin et al., 1994; Pascual and Preat, 2001; Dubnau et al., 2003; Comas et al., 2004; Isabel et al., 2004; Yu et al., 2006; Blum et al., 2009; Akalal et al., 2010, 2011; Trannoy et al., 2011; Huang et al., 2012; Placais et al., 2012).
We assessed the requirement of MBONs in the retrieval phase of aversive LTM by training flies at permissive temperature and blocking their activity only during the memory test at 24 hr after training (Figure 10A). Inactivation of MB052B, which broadly labels cholinergic V2 cluster MBONs (Figure 3F; Table 1), resulted in nearly complete loss of aversive odor LTM recall (Figure 10C). Our results confirm a previously reported requirement for the V2 cluster MBONs in long-term aversive odor memory recall (Sejourne et al., 2011). We also assayed five lines that express in subsets of the MBONs found in MB052B. While some of them showed lower memory scores, none showed a statistically-significant memory impairment (Figure 10B). Therefore, our results suggest that V2 cluster MBONs (Figure 10D) function as a group in the retrieval of aversive odor LTM. Consistent with this implied combinatorial action of V2 cluster MBONs, activating subsets of these MBONs with CsChrimson resulted in weak attraction that only reached statistical significance when MB052B was used as the driver (Figure 2C). In calcium imaging experiments, V2 cluster MBONs were reported to reduce their response to an odor that had been learned to be aversive (Sejourne et al., 2011); this sign of plasticity is consistent with our observation that activation of these neurons elicited attraction.
Flies are able to associate an odor with the intoxicating properties of ethanol (Figure 11A). Briefly, flies are exposed to two consecutive odors; the second of which is paired with a mildly intoxicating concentration of ethanol vapor. Flies are later tested for their odor preference for the paired vs unpaired odor (Figure 11A) (Kaun et al., 2011). Flies avoid the odor they experienced at the time of intoxication when tested 30 min after training, but show a long-lasting preference for that odor when tested 24 or more hours later (control data in Figure 11B,C) (Kaun et al., 2011). Blocking KC synaptic output has been shown to interfere with this memory (Kaun et al., 2011), indicating a role for the MB.
To test the role of MBONs in appetitive ethanol memory, we blocked MBON function during training and memory retrieval and assayed for changes in performance. When tested at 24 hr, eight driver lines showed significant memory impairment compared to genetic controls (Figure 11B,D; Figure 11—figure supplement 1). These eight lines were not significantly different from controls in 30-min aversive memory, although two exhibited a trend towards decreased 30-min memory (Figure 11C); the ability to form memories at 30 min establishes that flies of these genotypes can sense the odors and learn to associate them with ethanol.
Our results indicate that MBON-γ4>γ1γ2 (MB434B and MB298B) and MBON-α′2 (MB018B), whose involvement we confirmed with a second line not part of the original screening set (MB091C; Figure 11E), are preferentially required for 24 hr memory.
Our data also suggest the involvement of MBON-γ2α′1, where one driver line (MB077B) had a strong effect, while the second driver (MB051B) had a weaker effect that did not reach statistical significance (Figure 11D). Our results raise the possibility of the involvement of MBON-α3. However, the two lines we have for this cell type (MB082C and MB093C) gave discordant results and MB082C shows significant expression in MBON-α′2, a cell type that has a large effect on appetitive ethanol memory. Finally, blocking M4/M6 cluster MBONs from γ5 and β′2 (MB011B and MB210B) significantly affected 24 hr memory, but also appeared to decrease 30 min memory. The MBONs required for 24 hr appetitive ethanol memory (Figure 11F,G) are partially overlapping with those required for other forms of appetitive memory, again with involvement of multiple glutamatergic and cholinergic MBONs.
Silencing MBON-α′2, a cell type composed of just one cell in each hemisphere (Figure 11—figure supplement 1F,G) (Aso et al., 2014), resulted in persistence of the aversive memory after 24 hr, when control flies show appetitive memory (see MB018B in Figure 11D and MB091C in Figure 12E). These results suggest that, while memories for the aversive and rewarding effects of ethanol intoxication are formed simultaneously, they are expressed at different times through independent MB circuits. Moreover, maintenance of the aversive memory upon inactivation of MBON-α′2 argues against a passive dissipation of the aversive memory with time, and implies an active process in the conversion of aversive to appetitive memory.
The involvement of the MB in regulating sleep was first established by demonstrating that blocking synaptic output from KCs can either increase or decrease sleep, depending on the GAL4 driver used (Joiner et al., 2006; Pitman et al., 2006). Sleep in Drosophila is under circadian and homeostatic regulation and defined as sustained periods of inactivity (5 min or more) that coincide with increased sensory thresholds, altered brain activity and stereotyped body posture (Hendricks et al., 2000; Shaw et al., 2000; Hendricks and Sehgal, 2004; Ganguly-Fitzgerald et al., 2006; Parisky et al., 2008; Donlea et al., 2011). Little is known about how the MB's sleep regulating functions are executed. As a first step in elucidating these mechanisms, we asked if activation of specific MBONs changed sleep pattern and amount.
Sleep was measured before, during, and after heat-gated activation of MBONs (Figure 12A) by expression of the temperature-gated dTrpA1 channel. We identified five MBON drivers that suppressed sleep and seven MBON drivers that increased sleep (Figure 12B). All five sleep-suppressing drivers express in glutamatergic MBONs (MBON-γ5β′2a, MBON-β′2mp, MBON-β′2a_bilateral and MBON-γ4>γ1γ2), while sleep-promoting drivers express either in GABAergic (MBON-γ3β′1 and MBON-γ3) or cholinergic MBONs (MBON-γ2α′1); the neurotransmitter for MBON-calyx has not been determined (Figure 12B). Despite the 48 hr period of dTrpA1 activation, the effects of activation were reversible and the temperature shift had only minor effects on sleep in genetic control groups (Figure 12C). We retested the split-GAL4 drivers that showed a significant effect in the primary screening using a different dTrpA1 effector inserted at another genomic location, allowing us to assess expression more accurately since we had access to a reporter construct inserted at that site (Figure 12—figure supplement 1); these assays confirmed the initial results for all but one of the split-GAL4 driver lines (Figure 12D). None of the lines showed significant effects on general locomotion as assessed by video tracking (see ‘Materials and methods’). We identified distinct MBONs that either decrease (Figure 12E) or increase sleep (Figure 12F). The subset of MBONs that promoted sleep was similar to the subset in which CsChrimson activation was attractive; conversely, the MBON subset that promoted wakefulness was similar to the subset in which CsChrimson activation was aversive.
The projection patterns of MBONs provide insight into how these bidirectional signals might be integrated in the fly brain (Figure 12G). For example, the axons of the sleep promoting cholinergic MBON-γ2α′1 and wake promoting glutamatergic MBON-β′2mp project their termini to the same location in the brain (Video 6). These circuit arrangements of sleep- and wake-promoting neurons may facilitate the transition between the sleep and wake behavioral states by providing opposing inputs to shared downstream targets.
In the insect brain, a sparse and non-stereotyped ensemble of Kenyon cells represents environmental cues such as odors. The behavioral response to these cues can be neutral, repulsive or attractive, influenced by the prior experiences of that individual and how dopaminergic and other modulatory inputs have changed the weight of its KC-MBON synapses. MBONs are thought to encode the predictive value associated with a stimulus. A fly would then use that information to bias its selection of behavioral responses in an ever-changing environment. In the accompanying paper (Aso et al., 2014), we describe in detail the projection patterns of the MBON and DAN cell types that comprise the MB lobes in Drosophila. In this report, we begin the process of determining the nature of the information conveyed by MBONs. We also correlate specific MBON cell types with roles in associative memory and sleep regulation (Figure 13). Our anatomical and behavioral results lay the groundwork for understanding the circuit principles for a system of memory-based valuation and action selection.
We demonstrated that optogenetic activation of MBONs in untrained flies can induce approach or avoidance. The ability of the MBONs to induce changes in behavior in the absence of odors suggests that MBONs can bias behavior directly. This observation is consistent with a recent study showing that flies are able to associate artificial activation of a random set of KCs—instead of an odor stimulus—with electric shock, and avoid reactivation of the same set of KCs in the absence of odors (Vasmer et al., 2014), a result that recapitulates a finding in the potentially analogous piriform cortex of rodents (Choi et al., 2011). We found that the sign of the response to MBON activation was highly correlated with neurotransmitter type; all the MBONs whose activation resulted in avoidance were glutamatergic, whereas all the attractive MBONs were cholinergic or GABAergic.
By tracking flies as they encounter a border between darkness and CsChrimson-activating light, we showed that activation of an MBON can bias walking direction. Although activation of glutamatergic MBONs repelled flies, the avoidance behaviors were not stereotyped; flies showed a variety of motor patterns when avoiding the red light. This observation implies that MBONs are unlikely to function as command neurons to drive a specific motor pattern, as has been observed, for example, in recently identified descending neurons that induce only stereotyped backward walking (Bath et al., 2014; Bidaye et al., 2014).
Rather, we view fly locomotion as a goal-directed system that uses changes in MBON activity as an internal guide for taxis. For example, walking in a direction that increases the relative activity of aversive-encoding MBONs, which would occur as a fly approaches an odorant it had previously learned to associate with punishment (or when a fly expressing CsChrimson in an avoidance-inducing MBON approaches the CsChrimson activating light as in Figure 5E), signals the locomotive system to turn around and walk the other direction. Detailed studies of locomotor circuitry will be required to determine the mechanisms of executing such taxic behaviors and should help elucidate how MBON inputs guide this system.
In this view, the MBON population functions as neither a purely motor nor a purely sensory signal. From the motor perspective, as described above, MBONs bias locomotive outcomes rather than dictate a stereotyped low-level motor program. From the sensory perspective, we have shown that the same MBONs can be required for experience-dependent behavioral plasticity irrespective of whether a conditioned stimulus is a color or an odor, and irrespective of the specific identity of the odor. Taken together with the fact that MBONs lie immediately downstream of the sites of memory formation, these observations support our proposal that MBONs convey that a stimulus has a particular value—but not the identity of the stimulus itself. This contrasts with sensory neurons whose activity can also induce approach or avoidance, but which do convey the stimulus per se. In mammals, neural representations of abstract variables such as ‘value’, ‘risk’ and ‘confidence’ are thought to participate in cognition leading to action selection (for example, see Kepecs et al., 2008; Kiani and Shadlen, 2009; Levy et al., 2010; Yanike and Ferrera, 2014). From the point of view of this framework, the MBON population representing the value of a learned stimulus and informing locomotion might be operationally viewed as a cognitive primitive.
Co-activating multiple MBON cell types revealed that the effects of activating different MBONs appear to be additive; that is, activating MBONs with the same sign of action increases the strength of the behavioral response, whereas activating MBONs of opposite sign reduces the behavioral response. Thus, groups of MBONs, rather than individual MBONs, likely act collectively to bias behavioral responses. Consistent with the idea of a distributed MBON population code, all 19 MBON cell types imaged so far show a calcium response to any given odor (Hige et al., unpublished, Sejourne et al., 2011; Pai et al., 2013; Placais et al., 2013).
If it is the ensemble activity of a large number of MBONs that determines memory-guided behavior, how can local modulation of only one or a few MB compartments by dopamine lead to a strong behavioral response? Activation of a single DAN such as PPL1-γ1pedc that innervates a highly localized region of the MB can induce robust aversive memory, yet the odor associated with the punishment will activate MBONs from all compartments, including MBONs that can drive approach as well as those that drive avoidance. We propose that, in response to a novel odor stimulus, the activities of MBONs representing opposing valences may initially be ‘balanced’, so that they do not impose a significant bias. Behavior would then be governed simply by any innate preference a fly might have to that odor, using neuronal circuits not involving the MB. Now suppose an outcome associated with that stimulus is learned. Such learning involves compartment-specific, dopamine-dependent plasticity of the KC-MBON synapses activated by that stimulus. If that occurs, the subsequent ensemble response of the MBONs to that stimulus would no longer be in balance and an attraction to, or avoidance of, that stimulus would result. Consistent with this idea, eliminating MB function by disrupting KCs, which are nearly 10% of neurons in the central brain, had surprisingly minor effects on odor preference (de Belle and Heisenberg, 1994, Heisenberg et al., 1985; McGuire et al., 2001). Figure 14 shows a conceptual model of how this could be implemented at the level of neuronal circuits.
Recent studies of dopamine signaling have implicated distinct sites of memory formation within the MB lobes (summarized in Figure 1A) (Schroll et al., 2006; Claridge-Chang et al., 2009; Aso et al., 2010, 2012; Burke et al., 2012; Liu et al., 2012; Perisse et al., 2013). Consistent with this large body of work, we found that one type of PPL1 cluster DAN, PPL1-γ1pedc, played a central role in formation of aversive memory in both olfactory (Figure 6; see also Aso et al., 2012; Aso et al., 2010) and visual learning (Figure 8) paradigms. This DAN also mediates aversive reinforcement of bitter taste (Das et al., 2014). For appetitive memory, PAM cluster DANs that innervate other regions of the MB lobes, in particular the compartments of glutamatergic MBONs, are sufficient to induce appetitive memory (Yamagata et al., in press) (Burke et al., 2012; Liu et al., 2012) (Perisse et al., 2013). These results strongly suggest that the synaptic plasticity underlying appetitive and aversive memory generally occurs in different compartments of the MB lobes.
The sign of preference we observed in response to CsChrimson activation of particular MBONs was, in general, opposite to that of the memory induced by dopaminergic input to the corresponding MB compartments. For example, activation of MBON-γ1pedc>α/β and MBON-γ2α′1 attracted flies (Figure 2), whereas DAN input to these regions induced aversive memory (Figure 6G) (Aso et al., 2010, 2012). Conversely, activation of glutamatergic MBONs repelled flies (Figure 2), while DAN input to the corresponding regions is known to induce appetitive memory (Yamagata et al., in press) (Burke et al., 2012; Liu et al., 2012) (Perisse et al., 2013). These results are most easily explained if dopamine modulation led to synaptic depression of the outputs of the KCs representing the CS + stimulus. Consistent with this mechanism, the PE1 MBONs in honeybees (Okada et al., 2007) as well as the V2 cluster MBONs in Drosophila (Sejourne et al., 2011) reduce their response to a learned odor and depression of KC-MBON synapses has been shown for octopamine modulation in the locust MB (Cassenaer and Laurent, 2012). Moreover, long-term synaptic depression is known to occur in the granular cell synapses to Purkinje cells in the vertebrate cerebellum (Ito et al., 1982), a local neuronal circuit with many analogies to the MB (Schurmann, 1974; Laurent, 2002; Farris, 2011). Other mechanisms are also possible and multiple mechanisms are likely to be used. For example, dopamine may modulate terminals of KCs to potentiate release of an inhibitory cotransmitter such as short neuropeptide F, which has been demonstrated to be functional in KCs (Knapek et al., 2013) and hyperpolarizes cells expressing the sNPF receptor (Shang et al., 2013). RNA profiling of MBONs should provide insights into the molecular composition of synapses between KCs and MBONs. It is also noteworthy that the effect of dopamine can be dependent on the activity status of Kenyon cells; activation of PPL1-γ1pedc together with odor presentation induces memory, while its activation without an odor has been reported to erase memory (Berry et al., 2012; Placais et al., 2012). In the vertebrate basal ganglia, dopamine dependent synaptic plasticity important for aversive and appetitive learning is known to result in both synaptic potentiation and synaptic depression (Shen et al., 2008).
In this study, we looked at the effects of selectively and specifically manipulating the activities of a comprehensive set of MBONs on several behaviors. As a consequence, we gained some insights into the extent to which the relative importance of particular MBONs differed between behaviors (Figure 13). Most obvious was the segregation between appetitive and aversive behaviors. For example, we found that blocking MBON-γ1pedc>α/β impaired both short-term aversive odor and visual memory, suggesting a general role in aversive memory independent of modality. Conversely, a subset of glutamatergic MBONs was required in all three appetitive memory assays. It still remains to be demonstrated that the outputs of these MBONs are required transiently during memory retrieval. Nevertheless, CsChrimson activation experiments demonstrate that activation of these MBONs can directly and transiently induce attraction and avoidance behaviors.
In the cases described above, the DANs and MBONs mediating a particular behavior innervate the same regions of MB lobes. We also found cases where the DANs and MBONs required for a behavior do not innervate the same compartments of the MB lobes. For example, even though several cholinergic MBONs are required for appetitive memory (Placais et al., 2013), the compartments with cholinergic MBONs do not receive inputs from reward-mediating PAM cluster DANs, but instead from PPL1 cluster DANs that have been shown to be dispensable for odor-sugar memory (Schwaerzel et al., 2003) (Figure 1A). What accounts for this mismatch? Perhaps these cholinergic MBONs' primarily function is in memory consolidation rather than retrieval. But the fact that CsChrimson activation of the cholinergic MBON-γ2α′1 and V2 cluster MBONs resulted in attraction, strongly suggests that at least some of the cholinergic MBONs have a role in directly mediating the conditioned response. Indeed, previous studies found a requirement for cholinergic MBONs (the V2 cluster and MBON-α3) during memory retrieval (Sejourne et al., 2011; Pai et al., 2013; Placais et al., 2013). One attractive model is that requirement of cholinergic MBONs originates from the transfer of information between disparate regions of the MB lobes through the inter-compartmental MBONs connections within the lobes or by way of connections outside the MB, like those described in the next two sections.
The multilayered arrangement of MBONs (see Figure 17 of the accompanying manuscript) (Aso et al., 2014) provides a circuit mechanism that enables local modulation in one compartment to affect the response of MBONs in other compartments. Once local modulation breaks the balance between MBONs, these inter-compartmental connections could amplify the differential level of activity of MBONs for opposing effects (Figure 14). For example, the avoidance-mediating MBON-γ4>γ1γ2 targets the compartments of attraction-mediating MBON-γ2α′1 and MBON-γ1pedc>α/β (Figure 14).
This network topology might also provide a fly with the ability to modify its sensory associations in response to a changing environment (see Discussion in the accompanying manuscript) (Aso et al., 2014). Consider the α lobe. Previous studies and our results indicate that circuits in the α lobe play key roles in long-term aversive and appetitive memory (Figure 10) (Pascual and Preat, 2001; Isabel et al., 2004; Yu et al., 2006; Blum et al., 2009; Sejourne et al., 2011; Cervantes-Sandoval et al., 2013; Pai et al., 2013; Placais et al., 2013). The α lobe is targeted by MBONs from other compartments and comprises the last layer in the layered output model of the MB (Figure 1B; see also Figure 17 of the accompanying manuscript) (Aso et al., 2014). The GABAergic MBON-γ1pedc>α/β and the glutamatergic MBON-β1>α both project to α2 and α3, where their axonal termini lie in close apposition (see Figure 17 of the accompanying manuscript) (Aso et al., 2014). DAN input to the compartments housing the dendrites of these feedforward MBONs induces aversive and appetitive memory, respectively (this work) (Perisse et al., 2013; Yamagata et al., in press). As pointed out in (Aso et al., 2014), this circuit structure is well-suited to deal with conflicts between long-lasting memory traces and the need to adapt to survive in a dynamic environment where the meaning of a given sensory input may change. To test the proposed role of the layered arrangement of MBONs in resolving conflicts between old memories and new sensory inputs, we will also need behavioral paradigms that, unlike the simple associative learning tasks used in our current study, assess the neuronal requirements for memory extinction and reversal learning.
The neuronal circuits that are downstream of the MBONs and that might read the ensemble of MBON activity remain to be discovered. However, the anatomy of the MBONs suggests that, at least in some cases, summation and canceling effects may result from convergence of MBON terminals on common targets (Aso et al., 2014). For example, the terminals of the sleep-promoting cholinergic MBON-γ2α′1 overlap with terminals of wake-promoting glutamatergic MBONs (γ5β′2a, β′2mp and β′2mp bilateral) in a confined area in CRE and SMP. In addition, some MBONs appear to terminate on the dendrites of DANs innervating other compartments, forming feedback loops. Using these mechanisms, local modulation in a specific compartment could broadly impact the ensemble of MBON activity and how it is interpreted.
Testing these and other models for the roles of the MBON network, both within the MB lobes and in the surrounding neuropils, will be facilitated by an EM-level connectome to confirm the synaptic connections we have inferred based on light microscopy. We will also need physiological assays to confirm the sign of synaptic connections and to measure plasticity. For example, we do not know the sign of action of glutamate in the targets of glutamatergic MBONs, as this depends on the receptor expressed by the target cells (Xia et al., 2005; Jan and Jan, 1976; Liu and Wilson, 2013). In this regard, we note that previous studies demonstrated a role for NMDA receptors in olfactory memory (Wu et al., 2007; Miyashita et al., 2012).
Neurons that are thought to mediate innate response to odors—a subset of projection neurons from the antennal lobes and output neurons from the lateral horn—also project to these same convergence zones (Figure 15; Figure 15—figure supplement 1). We propose that these convergence zones serve as network nodes where behavioral output is selected in the light of both the innate and learned valences of stimuli. What are the neurons downstream to these convergence zones? One obvious possibility is neurons that project to the fan-shaped body of the central complex whose dendrites are known to widely arborize in these same areas (Hanesch et al., 1989; Young and Armstrong, 2010a; Ito et al., 2013; Yu et al., 2013). It would make sense for the MB to provide input to the central complex, a brain region involved in coordinating motor patterns (Strauss, 2002). Figure 15 provides a diagrammatic summary of these proposed circuits.
Using inactivation to uncover the roles of specific cell types is inherently limited by redundancy and resiliency within the underlying neural circuits. For example, consider the MBONs from the α/β lobes. The output of the α/β Kenyon cells is known to be required for retrieval of aversive memory (Isabel et al., 2004; Krashes et al., 2007; McGuire et al., 2001; Cervantes-Sandoval et al., 2013). Our anatomical and behavioral results show that MBON-γ1pedc>α/β, a cell type we found to be critical for aversive memory, has terminals largely confined inside the α/β lobes, well-positioned to regulate a total 6 types of MBONs from the α/β lobes (Figure 1B). Yet we did not detect a requirement for any of these MBONs in short term aversive memory when tested individually. The ability to detect phenotypes also depends on the strength of the effector; for example, four glutamatergic MBON drivers showed aversive memory impairment in initial screening assays with a strong inhibitor of synaptic function, but we were not able to confirm these effects using a weaker effector (Figure 6B).
The failure to see effects when inactivating individual cell types is most easily explained by combinatorial roles and redundancy between MBONs. We note that this high level of resiliency is very reminiscent of observations made with genetic networks, where less than half of gene knockouts of evolutionarily conserved Drosophila genes result in a detectable phenotype (Ashburner et al., 1999). Whether or not we detect a requirement for a particular MBON in a particular learning paradigm is likely to depend on which DANs are recruited by the US used in that paradigm as well as the degree of redundancy in the MBON representation of valence. It will be informative to test systematically whether blocking combinations of MBONs, which did not show significant behavioral effect when blocked separately, results in significant memory impairment. It will also be important in future experiments to employ imaging and electrophysiological methods, in which the activities of individual neurons, and the consequences of plasticity, can be observed without being obscured by redundancy.
The MBs are implicated in functions beyond processing of associative memory (Martin et al., 1998; Liu et al., 1999; Joiner et al., 2006; Pitman et al., 2006; Zhang et al., 2007; Hong et al., 2008). MBONs that influence approach to, or avoidance of, a learned stimulus may also have roles in innate preference behaviors for temperature and hunger-dependent CO2 avoidance (Hong et al., 2008; Bang et al., 2011; Bracker et al., 2013). Moreover, we expect the behavioral repertoire that MBONs govern to go beyond simple approach and avoidance; the MB is known to play a role in experience-dependent regulation of proboscis extension (Masek and Scott, 2010) and courtship (McBride et al., 1999) as well as regulation of sleep (Joiner et al., 2006; Pitman et al., 2006) and post-mating behaviors such as oviposition (Fleischmann et al., 2001; Azanchi et al., 2013). Intriguingly, we found that MBONs whose activation was repulsive promoted wakefulness, whereas MBONs whose activation was attractive promoted sleep; it would make sense for flies to be awake and attentive in an adverse environment. Other internal states, in addition to sleep, are likely to affect the decision to carry out a particular memory-guided behavior; for example, the state of satiety has been shown to regulate memory expression (Krashes et al., 2009). The diverse influences of MBONs on behavior can be most easily explained if we assume that the activity of the ensemble of MBON conveys an abstract representation of both valence and internal state. In this view, the ensemble of MBONs may represent internal states along axes such as pleasant-unpleasant or aroused-not aroused. It is upon these axes that primitive forms of emotion are thought to have evolved (Anderson and Adolphs, 2014).
The construction and characterization of the split-GAL4 lines are described in detail in (Aso et al., 2014). The MBON cell types are listed in Table 1 and diagrammed in Figure 1B. The split-GAL4 lines used in this study are described in Table 2; expression patterns, using the most relevant UAS-reporter, are shown in the Figures. pBDPGAL4U (attP2), an enhancerless GAL4 construct (Pfeiffer et al., 2010), was used as a control driver line in behavioral assays.
To combine the expression patterns observed in two split-GAL4 lines, flies were generated by standard genetic crosses that contained the two DNA-binding (DBD) halves and the two activation-domain (AD) halves found in the parent split-GAL4 lines; the AD and DBD components of all split-GAL4 lines are given in Table 1 of the accompanying manuscript (Aso et al., 2014). In general, these lines contained more off-target cells than the parent split-GAL4 lines, due to the interactions of the AD and DBD combinations not present in the parent lines. To identify lines for behavioral experiments, we directly assessed the expression patterns; the majority of combinations produced useful reagents (Figure 4—figure supplement 1 shows the expression patterns of those combinations used in this work).
The following constructs were used for activating or silencing neuronal function: 5XUAS-CsChrimson-mVenus (attP18), 10XUAS-CsChrimson-mVenus (attP18), 20XUAS-CsChrimson-mVenus (attP18) (Klapoetke et al., 2014); pJFRC124-20XUAS-IVS-dTrpA1 (attP18); 10XUAS-dTrpA1 (attP16) (Hamada et al., 2008); pJFRC100-20XUAS-TTS-Shibirets1-p10 (VK00005) (Pfeiffer et al., 2012); UAS-Shi x1 was generated in Thomas Preat's lab by segregating one of the multiple insertions found in the lines described by (Kitamoto, 2001).
To compare the expression levels driven by split-GAL4 drivers in specific cell types, 3–7 days post-eclosion female brains were dissected, antibody-stained, mounted and imaged at 20× under identical conditions (see the accompanying manuscript for details) (Aso et al., 2014). The relative expression levels in individual cell types are presented as a 0–5 unit gray scale based on the intensity of the signals in the dendrites obtained for each cell type in each split-GAL4 line. The signal intensity depends on the morphology of individual cell types as well as how many cells of the same cell type innervate the same compartment; thus comparing intensities across cell types is less accurate than the comparisons between lines for the same cell type.
Since the purpose of obtaining these data was to estimate the expression levels of the UAS-effectors used to manipulate cell function, and expression levels are known to vary with genomic insertion site (Pfeiffer et al., 2010), we sought to collect expression data using UAS-indicator lines inserted into the same genomic location as the effectors. We believe that this practice addresses a potential weakness in many prior studies where expression patterns have been determined with an indicator construct inserted in one chromosomal site, while perturbing function with an effector residing at another site. The lack of precise correlation between the expression pattern of the indicator and effector introduces significant uncertainty. The best practice would be to directly measure the expression of the effector protein itself, something we were able to do for the red-shifted channel rhodopsin CsChrimson-mVenus (Klapoetke et al., 2014) by staining for mVenus. The next best approach is to have the indicator of expression and the effector inserted at the same chromosomal location, which we were able to achieve for all cases except the assays of the 2 hr aversive and appetitive odor memory which used a weaker UAS-Shibire effector based on a P-element insertion. We used the following UAS-indicators for the matrices shown in Figures: Figure 2B, 20XUAS-CsChrimson-mVenus (attP18) reared at 22°C; Figures 7B, 9B, 10B, 11B, pJFRC2-10XUAS-IVS-mCD8::GFP (VK00005) reared at 18°C [similar expression was observed with pJFRC225-5xUAS-IVS-myr::smGFP-FLAG (VK00005) reared at 25°C]; Figure 12B, pJFRC200-10xUAS-IVS-myr::smGFP-HA (attP18) reared at 22°C. Full confocal stacks of these images are available at www.janelia.org/split-gal4.
We used a set of highly specific GAL4 drivers, made using the split-GAL4 intersectional approach (Aso et al., 2014). These drivers have much more restricted expression patterns than those previously used, allowing greater certainty in assigning the effects of perturbations to specific cells. Even with these improved GAL4 drivers, there can be significant variation in expression levels between drivers or in different cell types within the expression pattern of a single driver, as well as off-target expression and variations in genetic background. To assign a function to a cell population, we therefore required that the effects of a manipulation be observed using two different GAL4 drivers for that cell population. In cases where we only had one split-GAL4 driver for a cell type, we believe it is only appropriate to interpret an observed phenotype as suggestive, except in cases where we were confirming a previously published result. Finally, we interpret some results as simply raising the possibility of a role for a cell type. For example, where one GAL4 line resulted in a significant effect, but a second line with a very similar expression pattern did not. There were also cases where we saw a consistent tendency in multiple lines, but where none of the individual lines themselves reached statistical significance.
Detailed methods for immunohistochemistry and image analysis are described in the accompanying manuscript (Aso et al., 2014). For the data in Figure 6—figure supplement 1B, rabbit anti-GABA (1:500; A2052, Sigma-Aldrich, St Louis, MO 63103) was used.
We asked if using the split-GAL4 lines to activate or inactivate neurons would perturb general innate behaviors, such as locomotion and visual perception that might interfere with our assays of memory, locomotion and sleep. Given the results of these tests, we selected 23 split-GAL4 lines for use in the primary behavioral screening of the MBONs and additional lines to confirm the results of primary screening (see Table 1). We also performed behavioral assays to verify that the animals carrying the driver line and effector were able to perceive odors, electric shocks and sugar rewards (see Results). Thus, the behavioral phenotypes we observed in these lines are unlikely to result from general defects in innate behaviors.
We first screened 27 MBON split-GAL4 driver lines, crossed to a multi-insert UAS-Shibirets1 effector line (UAS-Shits1 on the third chromosome) (Kitamoto, 2001) at 34°C. 3–7 days old adult males of each genotype were wet starved for 1–4 hr and then were assayed for 22 parameters of basic locomotion in response to startle, optomotor and phototaxis stimuli using an apparatus (Fly Behavioral Olympiad, unpublished) inspired by published assays (Benzer, 1967; Zhu et al., 2009). Fourteen split-GAL4 lines showed a difference in one or more parameters from the pBDPGAL4U and these lines were re-screened with pJFRC100-20XUAS-TTS-Shibire-ts1-p10 (VK00005) DL (Pfeiffer et al., 2012).
Although some lines had phenotypes in some behavioral categories, none of the output lines showed consistent phenotypes across the two Shibire effectors. All of the lines screened were able to appropriately respond to visual stimuli, showed positive phototaxis towards green and UV light, and were able to walk when MBONs were inactivated with Shibire. Only one line, MB549C, showed a significant reduction in walking speed, though subsequent analysis suggests this reduction in speed was due to the genetic background of that line rather than silencing of MBONs. Thus flies can move and orient themselves when MBONs, and any neurons with off-target expression in these split-GAL4 drivers, are inactivated and the small differences from wild-type would not be expected to significantly affect behavior in the assays we performed that used Shibire as an effector.
We also assayed the behavior of flies from each split-GAL4 driver line in a high-throughput open-field arena described in (Kabra et al., 2013) for 15 min during dTrpA1 activation using 10X UAS-dTRPA1 (attP16) (Hamada et al., 2008) at 30°C. We first tracked the body and wing position of the flies (Branson et al., 2009; Kabra et al., 2013), and then automatically annotated 14 social and locomotor behaviors of flies such as walking, chasing, grooming, etc (Supplementary file 1) (Kabra et al., 2013). Although we observed variation of locomotion levels between drivers and their GAL4/+ controls, only one driver, MB052B/dTrpA1, showed an obvious phenotype. This phenotype of MB052B was limited to male flies and presumably attributed to male specific expression in this line; we therefore did not use males of this driver in experiments involving activation.
The choice assay was performed in a 10 cm diameter and 3 mm high circular arena as previously described (Klapoetke et al., 2014). Flies expressing CsChrimson were allowed to distribute between two dark quadrants and two quadrants illuminated with 617 nm LEDs (Red-Orange LUXEON Rebel LED—122 lm; Luxeon Star LEDs, Brantford, Ontario, Canada). This wavelength efficiently activates neurons expressing CsChrimson (Klapoetke et al., 2014), but was distant enough from the peak absorption spectrum of endogenous rhodopsins that at the light intensity used (34 µW/mm2) negligible phototaxis of the control genotype was observed. To maintain a constant temperature, the LED board was placed on a heat sink and air (150 ml/min) was exchanged through holes at the center and four corners of the arena in a similar way as in the previously described olfactometer (Vet et al., 1983). The four quadrants were separated by 1 mm dividers. The bottom of arena consisted of a 3 mm thick diffuser with an IR absorption film (YAG, Laser PVC Film; Edmund optics, Barrington, NJ 08007-1380). The intensity of red light decreased from 34 to 3 µW/mm2 over a 10 mm gradient extending from the border between light on and off quadrants, as shown in Figure 5A.
Crosses were kept on standard cornmeal food supplemented with retinal (0.2 mM all-trans-retinal prior to eclosion and then 0.4 mM) at 22°C at 60% relative humidity in the dark. Groups of approximately twenty 4–10 days post-eclosion females were tested at 25°C at 50% relative humidity in a dark chamber. Videography was performed under reflected IR light using a camera (ROHS 1.3 MP B&W Flea3 USB 3.0 Camera; POINT GREY, Richmond, BC, Canada) with an 800-nm long pass filter (B&W filter; Schneider Optics) at 30 frames per sec,1024 × 1024 pixel resolution and analyzed using Fiji (Schindelin et al., 2012) or Ctrax (Branson et al., 2009). Statistical comparisons were performed using Prism (Graphpad Inc, La Jolla, CA 92037); Kruskal Wallis One way ANOVA followed by Dunn's post-test for comparison between control and experimental genotype in Figure 2C and Figure 5—figure supplement 1; One way ANOVA followed by Bonferroni's multiple comparison test for Figures 3G and 4; In Figure 5D, p-values for the exit direction were computed using the test of equal proportions from R (http://stat.ethz.ch/R-manual/R-patched/library/stats/html/prop.test.html) followed by multiple comparisons with the Dunn-Sidak correction. Only data obtained with 20XUAS-CsChrimson-mVenus (attP18) are shown in this study. Our preliminary results with 5XUAS-CsChrimson-mVenus (attP18) or 10XUAS-CsChrimson-mVenus (attP18) indicate that either too weak or too strong expression may result in a failure to observe a phenotype.
Behavioral experiments were performed at 60% humidity in dim red light for training and in complete darkness for test. The odors, 3-octanol (OCT; Merck) and 4-methylcyclohexanol (MCH; Sigma–Aldrich) were diluted to 1% and 2%, respectively, in paraffin oil (Sigma–Aldrich). Flies were placed in the apparatus and shifted to the restrictive temperature (32°C) from 30 min prior to the commencement of training until the end the experiment. During the 2-hr period between training and testing, trained flies were kept in a vial with moistened filter paper. The trained flies were then allowed to choose between MCH and OCT for 2 min in a modified transparent T-maze. Odors were placed in cups with 30 mm diameter and delivered at a flow rate of 0.6 l/min per tube. The distribution of the flies was monitored by videography and the preference index was calculated by taking the mean indices of the last 10 s in the 2-min choice period. Half of the trained groups received reinforcement together with the first presented odor and the other half with the second odor to cancel the effect of the order of reinforcement.
For aversive memory, a group of ∼50 flies in a training tube alternately received OCT and MCH for 1 min in a constant air stream; twelve 1.5 s 90 V electric shocks spaced over 60 s were paired with one of the odors. In the primary screening using pJFRC100-20XUAS-TTS-Shibire-ts1-p10 (VK00005) as the effector, flies were raised at 25°C. In the secondary screening using the UAS-Shi x1 effector, flies were raised at 18°C. Odor avoidance was measured by asking flies to choose between air and either MCH or OCT at the same concentrations used in the memory assay; these odors are aversive to naïve flies. For shock avoidance, flies were asked to choose between two tubes, both with copper grids, but only one electrically active. For appetitive memory, the conditioning protocol was as described previously (Liu et al., 2012). Flies were starved prior to the experiments until ∼10% mortality was reached. For sugar attraction, flies are asked to choose between two tubes, one with plain filter paper and one with sugar-embedded paper.
Statistical analyses were performed with Prism5 software (GraphPad). The tested groups that did not violate the assumption of normal distribution (D'Agostino-Pearson test) or homogeneity of variance (Bartlett's test) were analyzed with parametric statistics: one-sample t-test or one-way analysis of variance followed by planned pairwise multiple comparisons (Bonferroni). The significance level for statistical tests was set to 0.05. As some of the data points in Figure 7B violated the assumption, non-parametric statistics were applied to the dataset (Kruskal–Wallis test followed by Dunn's multiple test).
The effector-line pJFRC100-20XUAS-TTS-Shibire-ts1-p10 (VK00005) was used, and all experimental flies were heterozygous (w+/w-) or wild-type (w+/Y) for white. Flies were sorted by genotype under CO2 anesthesia at least 2 days prior to experiments; each measurement used 30–40 mixed males and females. For appetitive conditioning experiments, 2–4 days post-eclosion flies were starved on moistened filter paper to approximately 20% mortality (Schnaitmann et al., 2010); for aversive conditioning experiments, flies were not starved. Control responses to sugar and shock were measured as described previously (Schnaitmann et al., 2010; Vogt et al., 2014).
Appetitive and aversive conditioning paradigms and behavioral tests were as previously described (Schnaitmann et al., 2013; Vogt et al., 2014). Briefly, conditioned stimuli were presented from below using LEDs with peak wavelengths of 452 nm and 520 nm (Seoul Z-Power RGB LED) or 456 nm and 520 nm (H-HP803NB, and H-HP803PG, 3 W Hexagon Power LEDs, Roithner Lasertechnik, Vienna, Austria), adjusted to 14.1 Cd m−2 s−1 (blue) and 70.7 Cd m−2 s−1 (green). Each quadrant of the arena was also equipped with an IR-LED (850 nm) to provide background illumination for videography. For appetitive conditioning, filter paper soaked in 2 M sucrose and subsequently dried was presented as reward (Schnaitmann et al., 2010). For aversive conditioning, a 1 s electric shock (AC 60 V) was applied 12 times in 60 s during CS + presentation using a transparent shock grid made of laser-structured ITO on a glass plate. In both appetitive and aversive conditioning assays, differential training was followed by a binary choice without reinforcement. During the 90 s test period, blue and green light were presented in two diagonally opposite quadrants of the arena and the color choice of flies was recorded from above at 1 frame per second with a CMOS camera (Firefly MV, Point Grey). A preference index for each frame was calculated by subtracting the number of flies on the green quadrants from the number on the blue quadrants, divided by the total number of flies. The difference in average visual stimulus preference between the two groups was used to calculate a performance index. Sugar preference and shock avoidance tests were performed as described previously (Schnaitmann et al., 2010; Vogt et al., 2014).
Statistical analyses were performed with Prism5 software (GraphPad). The groups that did not violate the assumption of normal distribution (Shapiro–Wilk test) or homogeneity of variance (Bartlett's test) were analyzed with parametric statistics: one-way analysis of variance followed by the planned pairwise multiple comparisons (Bonferroni). For data that significantly differed from the normal distribution or did not show homogeneity of variance (Bartlett's test), non-parametric statistics were applied (Kruskal–Wallis test followed by Dunn's multiple test). The significance level of statistical tests was set to 0.05. Only the most conservative statistical result of multiple pairwise comparisons is indicated.
For behavioral experiments concerning long-term aversive olfactory memory (LTM), wild type (Canton S) or Split-GAL4 female flies were crossed to either pJFRC100-20XUAS-TTS-Shibire-ts1-p10 (VK00005) (outcrossed to a Canton S genetic background) or wild type males. Flies were raised on standard medium containing yeast, cornmeal and agar at 18°C and 60% relative humidity under a 12 hr:12 hr light–dark cycle.
The day before the experiment, 0–2 days post-eclosion flies were transferred to fresh food vials. Flies were trained with five cycles of aversive conditioning spaced by 15 min inter-trial intervals (spaced conditioning) at 25°C. The time course of one cycle of aversive conditioning was as follows: flies were exposed to the first odorant for 1 min while twelve 1.5 s, 60-V electric shocks, separated by 3.5 s, were delivered; after a 45 s rest, flies were exposed to the second odorant for 1 min. Two odorants, 3-octanol and 4-methyl-cyclohexanol, were used alternatively as the conditioned stimulus. For all assays (training, memory test and olfactory acuity), odorants were diluted in mineral oil at a final concentration of 0.36 mM for octanol and 0.325 mM for methyl-cyclohexanol, and were delivered by 0.4 l/min airflow bubbled through odor-containing bottles. Except during conditioning, flies were kept on food and were maintained at 18°C between training and test. The memory test was performed as described in (Trannoy et al., 2011). Flies were allowed to acclimatize to the restrictive (32°C) or permissive (25°C) temperature for 30 min prior to the test. Memory scores are displayed as means ± SEM.
In a primary screen, each Split-Gal4 line was tested for aversive LTM (n = 7-10, except MB093C, n = 20). The scores obtained for each line were compared by a two-tailed unpaired t-test to the pool of +/UAS-Shits scores (n > 150). Due to multiple comparisons, a Benjamini-Hochberg procedure (Benjamini and Hochberg, 1995) was applied to control the false discovery rate with a significance level of 0.05. Putative hits after the primary screen were then re-assayed in comparison with +/UAS-Shits and Split-GAL4/+ at both the restrictive and permissive temperatures. In this second set of experiments, the scores from the three genotypes were compared using one-way ANOVA followed by pairwise comparisons by Newman–Keuls posthoc tests. MB052B/Shi showed memory impairment compared to MB052B/+ and +/Shi at the restrictive temperature (ANOVA, F2,33 = 14.84, p < 0.0001; ***: p < 0.001 by Newman–Keuls pairwise comparison; n ≥ 9 for all genotypes) but not at the permissive temperature (ANOVA, F2,24 = 0.95, p = 0.40, n ≥ 8 for all genotypes).
Flies were reared at room temperature (22°C–23°C) under ambient light with no constrained light/dark cycle. Split-GAL4 males were crossed to pJFRC100-20XUAS-TTS-Shibire-ts1-p10 (VK00005) females. A daily control (pJFRC100 x pBDPGAL4U) was run alongside all experimental crosses. Split-GAL4/+ crosses were performed at a different time than the original split-GAL4/Shits screen and run alongside UAS-Shits/+ flies. Thermoactivation of Shits1 was carried out at 31°C during both the training and test period. For 24-hr memory, flies were kept at 22°C–23°C under ambient room light between training and test.
Odors used were 1:36 (vol:vol) ethyl acetate in mineral oil and 1:36 (vol:vol) iso-amyl alcohol in mineral oil. Choice tests for groups of 30 flies were performed in a Y-maze (each arm 2.5 cm in length and 1.5 cm in diameter). Odors were actively streamed individually through the top arms of the Y at 0.3 l/min. Vials of flies were placed at the lower Y arm and flies climbed up and chose between opposing arms of the Y into 14 ml culture tubes; one arm contained one of the odors and the other arm contained air streamed through mineral oil. The preference index was calculated by the formula: (number of flies in odor vial–number of flies in air vial) / total number of flies. The conditioned preference index was the average of the preference indices in reciprocal trials.
Ethanol conditioning was performed essentially as described in (Kaun et al., 2011). Groups of 30 males were trained in perforated 14 ml culture vials filled with 1 ml of 1% agar and covered with mesh lids. 96 vials of flies were trained simultaneously in two 30 × 15 × 15 cm training boxes. Training consisted of a 10 min habituation to the training chamber with air, a 10 min presentation of odor 1 (1:36 odor:mineral oil actively streamed at 2 l/min), then 10 min of odor 2 (1:36 odor:mineral oil actively streamed at 2 l/min), with 60% ethanol. Air flow was matched for CS+ and CS− experiments, and ethanol was delivered by mixing pure ethanol vapor (1.5 l/min) with humidified air (1.1 l/min) at a specified ratio (Wolf et al., 2002). Reciprocal training was performed to ensure that an inherent preference for either odor did not affect the results. Vials of flies from Group 1 and Group 2 were paired according to placement in the training chamber and tested simultaneously. Flies were tested in the Y-maze described above either 30 min or 24 hr after training. Reciprocal groups were averaged for each n = 1.
Statistical analyses were performed using the statistical software JMP 10.0.0 (SAS Institute, Inc., Cary, NC 27513-2414). Each split-GAL4/UAS-Shits1 cross was run on two separate days and pooled for a total with n = 12/group. Statistical significance for any split-GAL4 line was determined by performing a Wilcoxon test for each split-GAL4/UAS-Shits cross against a pooled control. The pooled control included 12 randomly sampled means from the pBPDG4U/UAS-Shits daily control. A Benjamini-Hochberg False discovery rate (FDR) test (Benjamini and Hochberg, 1995) was performed on the p-values for each Wilcoxon comparison. Lines showing p < 0.05 following the FDR test were considered statistically significant. GAL4/+ controls were performed for these significant hits, and Kruskal–Wallis comparisons were made comparing each split-GAL4/UAS-shits, split-GAL4/+ and +/UAS-Shits. Lines were considered significant hits if they passed the FDR correction, splitGAL4/+ Kruskal–Wallis test and showed no significantly decreased sensitivity for either odor used in the assay.
Split-GAL4 flies were crossed to either 10X UAS-dTrpA1 (attP16) (Hamada et al., 2008) or pJFRC124-20XUAS-IVS-dTrpA1 (attP18) and maintained at 21–22°C. Virgin female progeny, 3–7 days post-eclosion, (n = 24–35) were placed in 65 mm × 5 mm transparent plastic tubes with standard cornmeal dextrose agar media, placed in a Drosophila Activity Monitoring system (Trikinetics) and locomotor activity data were collected in 1-min bins for 7 days. Activity monitors were maintained with a 12 hr:12 hr light–dark cycle at 65% relative humidity. Total 24-hr sleep amounts (daytime plus nighttime sleep) were extracted from the locomotor data as described by (Donelson et al., 2012); sleep was defined as 5 min or more of inactivity (Hendricks et al., 2000; Shaw et al., 2000). Sleep profiles were generated representing average (n = 24–32) sleep (min/30 min) for day 3 (baseline), days 4 and 5 (activation), and day 6 (recovery). In addition to permissive temperature controls, pBDPGAL4U /dTrpA1 and split-GAL4/+ were used as genotypic controls for hit detection. For all screen hits, waking activity was calculated as the number of beam crossings/min when the fly was awake; consistent with the assays performed in the Flybowl, none of the lines had discernable locomotor defects. Statistical comparisons between experimental and control genotypes were performed using Prism (Graphpad Inc) by Kruskal Wallis One way ANOVA followed by Dunn's post-test.
Structure and response patterns of olfactory interneurons in the honeybee, Apis melliferaThe Journal of Comparative Neurology 437:363–383.https://doi.org/10.1002/cne.1289
An exploration of the sequence of a 2.9-Mb region of the genome of Drosophila melanogaster: the Adh regionGenetics 153:179–219.
The mushroom body of adult Drosophila characterized by GAL4 driversJournal of Neurogenetics 23:156–172.https://doi.org/10.1080/01677060802471718
Specific dopaminergic neurons for the formation of labile aversive memoryCurrent Biology 20:1445–1451.https://doi.org/10.1016/j.cub.2010.06.048
Competing dopamine neurons drive oviposition choice for ethanol in DrosophilaProceedings of the National Academy of Sciences of USA 110:21153–21158.https://doi.org/10.1073/pnas.1320208110
Controlling the false discovery rate - a Practical and powerful approach to multiple testingJournal of the Royal Statistical Society Series B-Methodological 57:289–300.
BEHAVIORAL MUTANTS OF Drosophila ISOLATED BY COUNTERCURRENT DISTRIBUTIONProceedings of the National Academy of Sciences of USA 58:1112–1119.https://doi.org/10.1073/pnas.58.3.1112
High-throughput ethomics in large groups of DrosophilaNature Methods 6:451–457.https://doi.org/10.1038/nmeth.1328
Mushroom bodies regulate habit formation in DrosophilaCurrent Biology 19:1351–1355.https://doi.org/10.1016/j.cub.2009.06.014
System-like consolidation of olfactory memories in DrosophilaThe Journal of Neuroscience 33:9846–9854.https://doi.org/10.1523/JNEUROSCI.0451-13.2013
Tripartite mushroom body architecture revealed by antigenic markersLearning & Memory 5:38–51.
Drosophila learn opposing components of a compound food stimulusCurrent Biology 24:1723–1730.https://doi.org/10.1016/j.cub.2014.05.078
Olfactory memory formation in Drosophila: from molecular to systems neuroscienceAnnual Review of Neuroscience 28:275–302.https://doi.org/10.1146/annurev.neuro.28.061604.135651
dunce, a mutant of Drosophila deficient in learningProceedings of the National Academy of Sciences of USA 73:1684–1688.https://doi.org/10.1073/pnas.73.5.1684
Genetic and functional subdivision of the Drosophila antennal lobeCurrent Biology 15:1548–1553.https://doi.org/10.1016/j.cub.2005.07.066
Mushroom bodies and post-mating behaviors of Drosophila melanogaster femalesJournal of Neurogenetics 15:117–144.https://doi.org/10.3109/01677060109066198
Radish, a Drosophila mutant deficient in consolidated memoryProceedings of the National Academy of Sciences of USA 90:8123–8127.https://doi.org/10.1073/pnas.90.17.8123
An engram found? Evaluating the evidence from fruit fliesCurrent Opinion in Neurobiology 14:737–744.https://doi.org/10.1016/j.conb.2004.10.014
Identifying behavioral circuits in Drosophila melanogaster: moving targets in a flying insectCurrent Opinion in Neurobiology 22:609–614.https://doi.org/10.1016/j.conb.2012.01.002
Neuronal architecture of the Central complex in Drosophila melanogasterCell and Tissue Research 257:343–366.https://doi.org/10.1007/BF00261838
Drosophila mushroom body mutants are deficient in olfactory learningJournal of Neurogenetics 2:1–30.https://doi.org/10.3109/01677068509100140
Why a fly? Using Drosophila to understand the genetics of circadian rhythms and sleepSleep 27:334–342.
Hunger and memory; CRTC coordinates long-term memory with the physiological state, hungerCommunicative & Integrative Biology 6:e25152.https://doi.org/10.4161/cib.25152
Climbing fibre induced depression of both mossy fibre responsiveness and glutamate sensitivity of cerebellar Purkinje cellsThe Journal of Physiology 324:113–134.
L-glutamate as an excitatory transmitter at Drosophila Larval Neuromuscular-JunctionJournal of Physiology 262:215–236.
Independent optical excitation of distinct neural populationsNature Methods 11:338–346.https://doi.org/10.1038/nmeth.2836
Bruchpilot, a synaptic active zone protein for anesthesia-resistant memoryThe Journal of Neuroscience 31:3453–3458.https://doi.org/10.1523/JNEUROSCI.2585-10.2011
Olfactory network dynamics and the coding of multidimensional signalsNature Reviews Neuroscience 3:884–895.https://doi.org/10.1038/nrn964
Odorant-induced oscillations in the mushroom bodies of the locustThe Journal of Neuroscience 14:2993–3004.
Neural representation of subjective value under risk and ambiguityJournal of Neurophysiology 103:1036–1047.https://doi.org/10.1152/jn.00853.2009
Glutamate is an inhibitory neurotransmitter in the Drosophila olfactory systemProceedings of the National Academy of Sciences of USA 110:10294–10299.https://doi.org/10.1073/pnas.1220560110
Mushroom bodies suppress locomotor activity in Drosophila melanogasterLearning & Memory 5:179–191.
Learning-related plasticity in PE1 and other mushroom body-extrinsic neurons in the honeybee brainThe Journal of Neuroscience 27:11736–11747.https://doi.org/10.1523/JNEUROSCI.2216-07.2007
Drosophila ORB protein in two mushroom body output neurons is necessary for long-term memory formationProceedings of the National Academy of Sciences of USA 110:7898–7903.https://doi.org/10.1073/pnas.1216336110
Using translational enhancers to increase transgene expression in DrosophilaProceedings of the National Academy of Sciences of USA 109:6626–6631.https://doi.org/10.1073/pnas.1204520109
Fiji: an open-source platform for biological-image analysisNature Methods 9:676–682.https://doi.org/10.1038/nmeth.2019
Color discrimination with broadband photoreceptorsCurrent Biology 23:2375–2382.https://doi.org/10.1016/j.cub.2013.10.037
Appetitive and aversive visual learning in freely moving DrosophilaFrontiers in Behavioral Neuroscience 4:10.https://doi.org/10.3389/fnbeh.2010.00010
[On the functional anatomy of the corpora pedunculata in insects (author's transl)]Experimental Brain Research 19:406–432.https://doi.org/10.1007/BF00234464
Dopamine and octopamine differentiate between aversive and appetitive olfactory memories in DrosophilaJournal of Neuroscience 23:10495–10502.
Complementary function and integrated wiring of the evolutionarily distinct Drosophila olfactory subsystemsThe Journal of Neuroscience 31:13357–13375.https://doi.org/10.1523/JNEUROSCI.2360-11.2011
The central complex and the genetic dissection of locomotor behaviourCurrent Opinion in Neurobiology 12:633–638.https://doi.org/10.1016/S0959-4388(02)00385-9
Organization of antennal lobe-associated neurons in adult Drosophila melanogaster brainThe Journal of Comparative Neurology 520:4067–4130.https://doi.org/10.1002/cne.23142
Neuronal assemblies of the Drosophila mushroom bodyThe Journal of Comparative Neurology 508:711–755.https://doi.org/10.1002/cne.21692
Classical-conditioning and retention in Normal and mutant Drosophila-MelanogasterJournal of Comparative Physiology A-Sensory Neural and Behavioral Physiology 157:263–277.https://doi.org/10.1007/BF01350033
Olfactory representations by Drosophila mushroom body neuronsJournal of Neurophysiology 99:734–746.https://doi.org/10.1152/jn.01283.2007
Induction of aversive learning through thermogenetic activation of Kenyon cell ensembles in DrosophilaFrontiers in Behavioral Neuroscience 8:174.https://doi.org/10.3389/fnbeh.2014.00174
Molecular architecture of smell and taste in DrosophilaAnnual Review of Neuroscience 30:505–533.https://doi.org/10.1146/annurev.neuro.30.051606.094306
High-resolution analysis of ethanol-induced locomotor stimulation in DrosophilaThe Journal of Neuroscience 22:11035–11044.
Distinct dopamine neurons mediate reward signals for short- and long-term memoriesProceedings of the National Academy of Sciences of USA In press.https://doi.org/10.1073/pnas.1421930112
Representation of outcome risk and action in the anterior caudate nucleusThe Journal of Neuroscience 34:3279–3290.https://doi.org/10.1523/JNEUROSCI.3818-13.2014
Building the central complex in Drosophila: the generation and development of distinct neural subsetsThe Journal of Comparative Neurology 518:1525–1541.https://doi.org/10.1002/cne.22285
Structure of the adult central complex in Drosophila: organization of distinct neuronal subsetsThe Journal of Comparative Neurology 518:1500–1524.https://doi.org/10.1002/cne.22284
Parallel pathways for cross-modal memory retrieval in DrosophilaThe Journal of Neuroscience 33:8784–8793.https://doi.org/10.1523/JNEUROSCI.4631-12.2013
Liqun LuoReviewing Editor; Howard Hughes Medical Institute, Stanford University, United States
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
Thank you for sending your work entitled ”Mushroom body output neurons encode valence and guide memory-based action selection in Drosophila“ for consideration at eLife. Your article has been favorably evaluated by K VijayRaghavan (Senior editor) and 3 reviewers, two of whom is a member of our Board of Reviewing Editors.
The following individuals responsible for the peer review of your submission have agreed to reveal their identity: Liqun Luo (Reviewing editor); Leslie Griffith, Yi Zhong, and Josh Dubnau (peer reviewers).
The Reviewing editor and the other reviewers discussed their comments before we reached this decision. The Reviewing editor has included below the comments from three reviewers. The reviewers are impressed by the study, and most comments are meant to help further improve/clarify the presentation and can be address by textual changes. We hope that your revised manuscript will address these points.
This paper provides the first behavioral look at the function of newly identified MBONs. The authors (and they are multitude, reflecting the need for expertise for each behavior examined) suggest the major function of MBONs is to encode valence. In this view MB outputs are not themselves driving motor patterns, but rather they are setting the bias to favor some subset of the possible motor programs. The support for this idea is necessarily at this point somewhat circumstantial, but the upshot of the behavioral experiments is that many different behaviors require the exact same MBON or set of MBONs. The commonality between the behavioral situations does not have to do with the qualities or sensory mode of the stimulus, but rather with whether or not the animal likes or dislikes it. Importantly, the isolated activation of these neurons can be shown, in some cases, to appear to drive the animal to behave as if it has an aversion or attraction. The small number of MBONs coupled with the large number of MB-influenced behaviors demand this type of model, since one-to-one connections with motor systems would simply be anatomically untenable.
I have no substantive negatives. Overall, I think that this is an incredibly important study for several reasons. First, it is a catalog of the functional anatomy of MBONs, laying out the first pass results of what happens in several behavioral contexts when you manipulate their activity. This will be an important resource for many in the community. Second, and critically, this paper goes beyond cataloging to give us a credible model of what this piece of brain is built to do. This is something that will be important for anyone working in any nervous system who is interested in how behavioral decisions are made.
The way the paper was written also deserves comment. It is very dense and very broad ranging due to the sheer number of experiments done. In spite of this, the paper manages to very clearly deliver its bottom lines, keeping the reader's eye on the message during the first read through. The details are all there when you are ready to go back to look at them, but they don't distract. This will allow a wide range of readers to access this paper and actually get something out of it. I also like the ”Rosetta Stone“ in Table 1 that provides orientation necessary to integrate this study with the previous literature.
This is a beautiful work that attacks most fundamental but highly difficult problem in the field of neuroscience: how valence is encoded within a neural network and how memory is associated with such organization. It fully takes advantage of accessibility to powerful genetic tools available to Drosophila. The proposed model is insightful and supported by a large body of data.
First, I want to say that this is a magnum opus that will become an instant classic in this field. This is the sort of paper that we have all expected to see come out of the Janelia effort in flies. It is the sort of study where the impact is even greater than would be expected by addition of the parts.
The authors took a systematic approach, a systems biology approach, to attack what is really the next big question in the ”mushroom body field“: what do mushroom bodies actually do. A huge body of literature coming from several different insect species (not just Drosophila; maybe check to see that some of the other insects are cited early on?) has already established that mushroom bodies are a multi-modal integration center. In Drosophila, the bulk of the work on the question “what do MBs do” has come from dissection of olfactory associative memory. In this paper, and accompanying manuscripts, the authors take a big step towards answering this question by systematically cataloging and then manipulating the MB output neurons. After generating reagents to manipulate each of the individual MBONs, the effects of these manipulations were then measured in the context of a series of well-studied MB dependent behaviors. These include innate responses to odor (including attraction vs avoidance), aversive and appetitive odor memory, aversive and appetitive visual learning, aversive and appetitive ethanol learning, and sleep.
There is of course an impressive amount of effort that underlies this study. But what really makes this manuscript compelling for a high impact journal is that the findings are synthetic. Conceptual themes emerge that would not be apparent if one studied a subset of MBONs or only one behavior. What emerges is an over-arching hypothesis in which MBs integrate information about the environment (including internal state and past experience) and where activity in individual groups of MBONs bias responses towards outputs that are adaptive given the valance of the stimuli. And these findings hang together for the most part in terms of how MBONs group together in terms of valence in the innate responses, impact on memory, and neurotransmitter types.
On one level, this model is basically how we might have all imagined this could work. But to see the model emerge elegantly from an exhaustive data set is startling nevertheless. Moreover, this study provides starting points for countless follow up studies. Specific hypotheses need to be tested, but this study provides a roadmap and a toolbox to do that. So this will have immense impact on the field.
Overall, I am really enthusiastic about this manuscript. All of my critiques are subordinate to the above positive comments.
Specific critiques (combining all reviewers):
1) Blocking MB output has an all or none impact on memory retrieval. But blocking individual MBONs has smaller impact. This is expected, and is discussed. But there are cases where the added impact of blocking individual MBONs doesn’t add up to the whole behavior. The most striking example is aversive olfactory memory, where only one MBON (gamma-ped>alpha/beta) has an effect, and the effect is partial. This can be explained of course by the potential that combinations of other MBONs can impact (i.e. redundancy), or conceptually at least by there being unknown MBONs. This should be discussed.
2) There is a significant literature on the roles of each of the major MB KC subsets. This consists of two types of experiment: genetic rescue experiments in which individual mutations are rescued with spatially restricted transgene expression, and experiments in which individual KC classes are silenced/activated. The findings with MBONs here are not really integrated into in the Discussion section. The most clear example of this is the MBON-gamma-ped>alpha/beta neuron. In aversive olfactory conditioning, all relevant DA input to DopR comes in through gamma lobes during learning, and retrieval and consolidation involve alpha/beta. So this MBON is ideally positioned to play an important role, but this wasn’t discussed in the text. On the other hand, in appetitive learning, there is evidence (actually from one of the authors) that early and later memories are parallel in gamma and alpha/beta. Again, this is not really discussed. I realize that the manuscript is long, has many findings, and it is not possible to place every finding in context. But this MBON is one of the few that is front and center in this study, so I thought these findings should be integrated better conceptually into the literature.
3) Related to above, in general, the role of individual lobes, the sequential use of MB lobes, the role for ongoing neural activity (prime lobes, APL, DPM) are not really incorporated into the framework provided in the discussion. Since some of the MBONs provide feedback (which is mentioned), i thought this should be better/more explicitly framed.
4) Although the MBON anatomy is documented in a different study, there are many cases in this manuscript where MBON labeling in individual panels (mostly in supplement) are almost impossible to see. I think the authors should provide insets with higher magnification b/c many of the neurons are too small to see when the entire CNS is shown.
5) In the description of Figure 9E findings, it is stated that naive responses to sugar are normal at restrictive temperature for all the lines. When I look at the figure, it appears that it is true that none are different from control, but some lines are higher and some lower than control, and my guess is that the higher lines are significantly different from the lower lines. This is a blemish in the data. Of course all papers have some blemishes, but I think the authors should state the statistical difference if there is one and just make the case that it is not a major concern.
6) In the sleep section, there is no rebound effect after 2 nights of sleep loss (in the lines that reduce sleep). Why?
7) This study crosses many sub-fields, so citation is really not trivial to organize. But I see a few omissions where previous work was not emphasized enough (IMHO). Including these could improve the Discussion section, and will avoid inadvertently annoying a few colleagues:
In my opinion, Marin et al. Cell 2002 and Lin et al., Cell 2007 should be cited alongside Vosshall and stocker, 2007; Yu et al., 2010; Wong et al., 2002; Jefferis et al., 2007. Another place where I see missing citations is where the authors first introduce the aversive and appetitive olfactory assays, but they only cite aversive (Tully Quinn, and some reviews). The appropriate appetitive citations should be included (Tempel original paper, and the modern versions from Waddell and Preat labs).
Previous work not emphasized enough:
I see two corners of the literature that IMHO deserve more attention: First, the findings on glutamatergic MBONs could be discussed in the context of the published role of NMDA receptors in MB dependent learning. There are a series of studies (Saito lab, Chiang/Tully collaboration) on NMDAr that I always found hard to incorporate into a model. But now one might view these in a new light.
The other literature that I think is given short shrift is a bulk of data on the effects of internal state (satiety) on reward learning via monoaminergic modulation of MB. These papers, mostly from Waddell lab, have a major impact on how we frame and think about the current datasets and I imagine they impacted experimental design quite a bit. It might be nice to shed a bit more textual light on this.
8) Concentrations of odors for learning and for task relevant sensorimotor responses not always mentioned. Please add this info and state whether the concentrations for olfactory acuity and learning are the same.https://doi.org/10.7554/eLife.04580.038
- Yoshinori Aso
- Karla R Kaun
- Alice A Robie
- William J Rowell
- Rebecca M Johnston
- Teri-T B Ngo
- Nan Chen
- Wyatt Korff
- Ulrike Heberlein
- Kristin M Branson
- Gerald M Rubin
- Yoshinori Aso
- Divya Sitaraman
- Katrin Vogt
- Michael N Nitabach
- Hiromu Tanimoto
- Pierre-Yves Plaçais
- Thomas Preat
- Ghislain Belliart-Guérin
- Hiromu Tanimoto
- Hiromu Tanimoto
- Hiromu Tanimoto
- Hiromu Tanimoto
- Hiromu Tanimoto
- Michael N Nitabach
- Michael N Nitabach
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank Sung Soo Kim, TJ Florence, Michael Reiser, Igor Negrashov, Steven Sawtelle, Jinyang Liu for help in establishing the optogenetics apparatus. The Janelia Scientific Computing Software group generated software tools, Brandi Sharp and the Janelia Fly Facility helped in fly husbandry, Susana Tae, Rebecca Vorimo and the FlyLight Project Team performed brain dissections, histological preparations and confocal imaging and Heather Dionne made several molecular constructs. We thank Fabian Stamp and Megan Atkins for assistance in visual learning and optogenetics assays, respectively. We thank Brett Mensh, Joshua T Dudman, Vivek Jayaraman, Larry F Abbott, Daisuke Hattori, Toshide Hige and Glenn Turner for stimulating discussions and for comments on earlier drafts of the manuscript.
- Liqun Luo, Reviewing Editor, Howard Hughes Medical Institute, Stanford University, United States
- Received: September 3, 2014
- Accepted: November 7, 2014
- Version of Record published: December 23, 2014 (version 1)
© 2014, Aso et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Downloads (link to download the article as PDF)
Download citations (links to download the citations from this article in formats compatible with various reference manager tools)
Open citations (links to open the citations from this article in various online reference manager services)
We identified the neurons comprising the Drosophila mushroom body (MB), an associative center in invertebrate brains, and provide a comprehensive map describing their potential connections. Each of the 21 MB output neuron (MBON) types elaborates segregated dendritic arbors along the parallel axons of ∼2000 Kenyon cells, forming 15 compartments that collectively tile the MB lobes. MBON axons project to five discrete neuropils outside of the MB and three MBON types form a feedforward network in the lobes. Each of the 20 dopaminergic neuron (DAN) types projects axons to one, or at most two, of the MBON compartments. Convergence of DAN axons on compartmentalized Kenyon cell–MBON synapses creates a highly ordered unit that can support learning to impose valence on sensory representations. The elucidation of the complement of neurons of the MB provides a comprehensive anatomical substrate from which one can infer a functional logic of associative olfactory learning and memory.
A detailed map of the neurons that carry information away from the mushroom bodies in the brains of fruit flies has improved our understanding of the ways in which experiences can modify behaviour.