In the experiments reported here we have developed a new group-training protocol for assessing long-term memory for habituation in Caenorhabditis elegans . We have replicated all of the major findings of the original single-worm protocol using the new protocol: (1) distributed training produced long-term retention of training, massed training did not; (2) distributed training at long interstimulus intervals (ISIs) produced long-term retention, short ISIs did not; and (3) long-term memory for distributed training is protein synthesis-dependent as it could be blocked by heat shock during the inter-block interval. In addition, we have shown that long-term memory for habituation is graded, depending on the number of blocks of stimuli in training. The inter-block interval must be >40 min for long-term retention of training to occur. Finally, we have tested long-term memory for habituation training in a strain of worms with a mutation in a vesicular glutamate transporter in the sensory neurons that transduce tap ( eat-4 ). The results from these eat-4 worms indicate that glutamate release from the sensory neurons has an important role in the formation of long-term memory for habituation.
Abstract Recent advances in neurogenetics have highlighted Drosophila melanogaste r as an exciting model to study neural circuit dynamics and complex behavior. Automated tracking methods have facilitated the study of complex behaviors via high throughput behavioral screening. Here we describe a newly developed low-cost assay capable of real-time monitoring and quantifying Drosophila group activity. This platform offers reliable real-time quantification with open source software and a user-friendly interface for data acquisition and analysis. We demonstrate the utility of this platform by characterizing ethanol-induced locomotor activity in a dose-dependent manner as well as the effects of thermo and optogenetic manipulation of ellipsoid body neurons important for ethanol-induced locomotor activity. As expected, low doses of ethanol induced an initial startle and slow ramping of group activity, whereas high doses of ethanol induced sustained group activity followed by sedation. Advanced offline processing revealed discrete behavioral features characteristic of intoxication. Thermogenetic inactivation of ellipsoid body ring neurons reduced group activity whereas optogenetic activation increased activity. Together, these data establish the f ly G roup A ctivity M onitor (flyGrAM) platform as a robust means of obtaining an online read out of group activity in response to manipulations to the environment or neural activity, with an opportunity for more advanced post-processing offline.
Abstract Substance use disorders are chronic relapsing disorders often impelled by enduring memories and persistent cravings. Alcohol, as well as other addictive substances, remolds neural circuits important for memory to establish obstinate preference despite aversive consequences. How pertinent circuits are selected and shaped to result in these unchanging, inflexible memories is unclear. Using neurogenetic tools available in Drosophila melanogaster we define how circuits required for alcohol associated preference shift from population level dopaminergic activation to select dopamine neurons that predict behavioral choice. During memory expression, these dopamine neurons directly, and indirectly via the mushroom body (MB), modulate the activity of interconnected glutamatergic and cholinergic output neurons. Transsynaptic tracing of these output neurons revealed at least two regions of convergence: 1) a center of memory consolidation within the MB implicated in arousal, and 2) a structure outside the MB implicated in integration of naïve and learned responses. These findings provide a circuit framework through which dopamine neuron activation shifts from reward delivery to cue onset, and provides insight into the inflexible, maladaptive nature of alcohol associated memories.
Locomotion is a behavioral readout that can be used to understand responses to specific stimuli or perturbations. The fly Group Activity Monitor (flyGrAM) provides a high-throughput and high-content readout of the acute stimulatory and sedative effects of ethanol. The flyGrAM system is adaptable and seamlessly introduces thermogenetic or optogenetic stimulation to dissect neural circuits underlying behavior and tests responses to other volatilized stimuli (humidified air, odorants, anesthetics, vaporized drugs of abuse, etc.). The automated quantification and readout of activity provide users with a real-time representation of the group activity within each chamber throughout the experiment, helping users to quickly determine proper ethanol doses and duration, run behavioral screens, and plan follow-up experiments.
Repeated alcohol experiences can produce long-lasting memories for sensory cues associated with intoxication. These memories can problematically trigger relapse in individuals recovering from alcohol use disorder (AUD). The molecular mechanisms by which ethanol changes memories to become long-lasting and inflexible remain unclear. New methods to analyze gene expression within precise neuronal cell types can provide further insight toward AUD prevention and treatment. Here, we used genetic tools in Drosophila melanogaster to investigate the lasting consequences of ethanol on transcription in memory-encoding neurons. Drosophila rely on mushroom body (MB) neurons to make associative memories, including memories of ethanol-associated sensory cues. Differential expression analyses revealed that distinct transcripts, but not genes, in the MB were associated with experiencing ethanol alone compared to forming a memory of an odor cue associated with ethanol. Adult MB-specific knockdown of spliceosome-associated proteins demonstrated the necessity of RNA-processing in ethanol memory formation. These findings highlight the dynamic, context-specific regulation of transcription in cue-encoding neurons, and the lasting effect of ethanol on transcript usage during memory formation.
In natural environments where food abundance and quality can change drastically over time, animals must continuously alter their food acquisition strategies. Although genetic variation contributes to this plasticity, the specific genes involved and their interactions with the environment are poorly understood. Here we report that natural variation in the Drosophila gene, foraging (for), which encodes a cGMP-dependent protein kinase (PKG), affects larval food acquisition in an environmentally dependent fashion. When food is plentiful, the wild-type rover (for(R)) allele confers lower food intake and higher glucose absorption than both the wild-type sitter (for(s)) allele and the mutant for(s2) allele. When food is scarce, for(R), for(s) and for(s2) larvae increase food intake to a common maximal level, but for(R) larvae retain their increased absorption efficiency. Changes in for expression can induce corrective behavioral modifications in response to food deprivation. When reared in environments with low food levels, for(R) larvae have higher survivorship and faster development than for(s) and for(s2) larvae. Together, these results show that natural variation in for has far reaching implications affecting a suite of phenotypes involved in the regulation of food acquisition.
Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The mushroom body (MB) is a well-characterized associative memory structure within the Drosophila brain. Analyzing MB connectivity using multiple approaches is critical for understanding the functional implications of this structure. Using the genetic anterograde transsynaptic tracing tool, trans-Tango, we identified divergent projections across the brain and convergent downstream targets of the MB output neurons (MBONs). Our analysis revealed at least three separate targets that receive convergent input from MBONs: other MBONs, the fan-shaped body (FSB), and the lateral accessory lobe (LAL). We describe, both anatomically and functionally, a multilayer circuit in which inhibitory and excitatory MBONs converge on the same genetic subset of FSB and LAL neurons. This circuit architecture enables the brain to update and integrate information with previous experience before executing appropriate behavioral responses. Our use of trans-Tango provides a genetically accessible anatomical framework for investigating the functional relevance of components within these complex and interconnected circuits. Introduction Neural circuits underlie all brain functions, from sensation and perception to learning, memory, and behavior. One of the greatest scientific challenges is to understand how neural circuits are structurally and functionally connected to support the extensive repertoire of behaviors animals use to interact with the world. Drosophila melanogaster is a powerful model for mapping the fundamental architecture of neural circuit organization in the context of specific behaviors due to its complex yet tractable brain. With a nervous system of approximately 100,000 neurons and a rich genetic toolkit that offers the potential to selectively manipulate subsets of neurons in behaving animals, significant effort has been devoted to establishing a detailed map of structural neural connectivity in the fly in an effort to then layer on function (Aso et al., 2014a; Bates et al., 2020; Couto et al., 2005; Deng et al., 2019; Eichler et al., 2017; Eschbach et al., 2020; Fishilevich and Vosshall, 2005; Frechter et al., 2019; Grabe et al., 2015; Kondo et al., 2020; Li et al., 2020; Marin et al., 2020; Otto et al., 2020; Peng et al., 2011; Shao et al., 2014; Takemura et al., 2017; Zheng et al., 2018). However, establishing a map of connectivity has proven to be a monumental task. Here, we bypass time and manpower by mapping mushroom body (MB) neural circuits across multiple animals using the recently developed genetic anterograde tracing tool trans-Tango (Talay et al., 2017). In trans-Tango, a synthetic signaling pathway converts the activation of a cell surface receptor into expression of a reporter gene via site-specific proteolysis. This pathway is introduced into all neurons while the starter neurons of interest express the ligand that activates the pathway and present it in their synapses. Binding of the ligand to its receptor on the postsynaptic partners activates the signaling pathway and leads to expression of a reporter that selectively labels these postsynaptic neurons (Talay et al., 2017). The insect MB is a prominent neuropil structure that integrates inputs from multiple sensory modalities (Caron et al., 2013; Ehmer and Gronenberg, 2002; Gruntman and Turner, 2013; Li and Strausfeld, 1997; Li and Strausfeld, 1999; Liu et al., 2006; Liu et al., 2016; Marin et al., 2020; Marin et al., 2002; Schildberger, 1984; Strausfeld and Li, 1999a; Strausfeld and Li, 1999b; Vogt et al., 2016; Vogt et al., 2014; Wang et al., 2016; Yagi et al., 2016; Zars, 2000) and has a well-established role in learning and memory (Davis, 1993; de Belle and Heisenberg, 1994; Heisenberg, 1998; Heisenberg, 2003; Heisenberg et al., 1985; Pascual and Préat, 2001; Zars et al., 2000). The MB comprises thousands of densely packed Kenyon cell neural fibers that are organized into three separate lobes (α/β, α′/β′, and γ). Kenyon cell neural fibers form en passant synapses along the length of their axons with efferent cells called MB output neurons (MBONs; Aso et al., 2014a; Eichler et al., 2017; Eschbach et al., 2020; Li and Strausfeld, 1997; Li and Strausfeld, 1999; Mobbs, 1982; Takemura et al., 2017). In addition to receiving processed sensory information, the MB integrates valence-related input from dopamine neurons (DANs; Aso et al., 2014a; Eichler et al., 2017; Eschbach et al., 2020; Takemura et al., 2017). This architecture positions the MB as a high-level integration center for the representations of multisensory cues and their perceived valence. Thus, the MB is an ideal neural structure for mapping structural connectivity and inferring fundamental architecture of neural circuits in the context of defined inputs and outputs across species. Early neuroanatomical and functional work in insects described distinct organization within the MB's afferent and efferent innervation patterns (Ito et al., 1998; Li and Strausfeld, 1997; Li and Strausfeld, 1999; Mao and Davis, 2009; Nässel and Elekes, 1992; Tanaka et al., 2008; Waddell, 2013). A more refined analysis of the neural circuits associated with the Drosophila MB was recently achieved through the use of split-Gal4 lines that enabled selective genetic access to specific neuronal populations (Aso et al., 2014a). These delineate a compartmentalization of the MB lobes by overlapping patterns of innervating DANs and MBONs (Aso et al., 2014a; Eichler et al., 2017; Eschbach et al., 2020; Takemura et al., 2017). Projections from the MBONs terminate within discrete neuropils, including the lateral horn (LH), crepine (CRE), superior medial (SMP), intermediate (SIP), and lateral (SLP) protocerebrum (Aso et al., 2014a; Ito et al., 2014). These neuropils have also been described as convergence sites of MBONs as different MBONs send converging outputs to similar subregions in these structures (Aso et al., 2014a; Ito et al., 2014). Within these neuropils, evidence suggests that MBON axons synapse onto dendrites of DANs and other MBONs providing opportunities for feedback to the MB (Aso et al., 2014a; Eichler et al., 2017; Scaplen et al., 2020). Evidence also suggests MBON axons synapse onto dendrites of neurons projecting to other structures, including the FSB (Aso et al., 2014a; Eichler et al., 2017; Scaplen et al., 2020). Additionally, similar to other insects and to the first instar Drosophila larva, MBONs in the adult brain are hypothesized to synapse on local interneurons whose processes are confined to the limits of the target neuropil but play a role in modulating input and output signals (Eichler et al., 2017; Phillips-Portillo and Strausfeld, 2012). These convergent neuropils, however, are characterized by highly complex arborizations of dendrites and axons. Therefore, identifying the specific neural components that receive synaptic input from various MBONs is challenging. Postsynaptic partners of specific neurons were initially identified by mapping the movement of cobalt ions from one neuron into another (Strausfeld and Obermayer, 1976). Later, candidate synaptic partners were identified either through the use of computational approaches to reveal overlapping arborization patterns or using molecular techniques such as fluorescent protein reconstitution across neurons (Chiang et al., 2011; Feinberg et al., 2008; Jefferis et al., 2007; Li et al., 2016; Lin et al., 2013; Macpherson et al., 2015; Shearin et al., 2018; Wolff et al., 2015). Recently, much effort has been devoted to map synaptic connections across the fly brain using whole brain serial electron microscopy (EM; Li et al., 2020; Ohyama et al., 2015; Schneider-Mizell et al., 2016; Xu et al., 2020; Zheng et al., 2017; Zheng et al., 2018). Although EM reconstruction offers synaptic structural resolution, it is labor intensive and it does not account for the synaptic strength nor the potential variability in synaptic connectivity across animals. We sought to test previous predictions regarding MBON connectivity (Aso et al., 2014a) and complement the EM anatomic data by mapping the postsynaptic partners of all MBONs using the genetic anterograde transsynaptic tracing tool, trans-Tango (Talay et al., 2017). We found that MBONs have a broad reach in their spread of postsynaptic connections. We observed abundant interconnectivity as previously predicted, with MBONs synapsing on DANs, and several MBONs converging on other MBONs. Further, we confirmed direct connections between the MBONs and two additional regions, the fan-shaped body (FSB) and the lateral accessory lobe (LAL). We identified, both anatomically and functionally, a multilayer circuit that includes GABAergic and cholinergic MBONs that converge on the same subset of FSB and LAL postsynaptic neurons. This circuit architecture provides an opportunity to integrate information processing before executing behavior, and we propose that multilevel integration across brain regions is critical for updating information processing and memory. Results Divergence and convergence of the MBONs circuits Circuit convergence, divergence, and re-convergence can be found throughout the nervous systems of both invertebrates and vertebrates and play a pivotal role in providing behavioral flexibility (Eschbach et al., 2020; Jeanne and Wilson, 2015; Man et al., 2013; Miroschnikow et al., 2018; Mišić et al., 2014; Ohyama et al., 2015). Given the importance of the MBONs in driving behavioral choice, we first sought to reveal patterns of divergence and convergence by identifying the postsynaptic connections of the MBONs innervating each of the 15 MB compartments using trans-Tango (Talay et al., 2017). Since trans-Tango signal depends on the strength and specificity of the GAL4 driver being used, we selected 28 previously published MBON split-GAL4 lines specific to individual MBONs, or sparse but overlapping subsets of MBONs (Aso et al., 2014a). We combined trans-Tango with chemogenetic active zone marker using the brp-SNAP knock-in to increase uniformity of neuropil labeling (Kohl et al., 2014). We successfully identified the postsynaptic connections of 25 split-GAL4 lines (Figure 1, Figure 1—figure supplements 1–23, open access raw data video files are available at https://doi.org/10.26300/mttr-r782). trans-Tango signals from MB112C (MBON γ1pedc>α/β) and G0239 (MBON α3) were too weak and were excluded from further analysis. In contrast, signals from MB242A (MBON calyx) proved to be too noisy to confidently identify postsynaptic connections. We also employed three new split-GAL4 lines that had more specific expression for γ5β′2a, β′2mp, and α2sc MBONs. Postsynaptic connections of glutamatergic, GABAergic, and cholinergic MBONs vary with regard to the divergence and breadth of their postsynaptic connections (Figure 1, Figure 1—figure supplements 1–23, external open access raw data video files are available at https://doi.org/10.26300/mttr-r782). For instance, MB011B, which includes glutamatergic MBONs γ5β′2a, β′2mp, and β′2mp-bilateral has extensive connections across the superior protocerebrum (Figure 1A), whereas MB542B, which includes cholinergic MBONs α′1, α2p3p, α′3 m has limited connections within the LH (Figure 1N). The innervation patterns did not seem to correlate with neurotransmitter type or number of MBONs expressing each split-GAL4. Figure 1 with 23 supplements see all Download asset Open asset MBONs have divergent connections across the brain. Exemplar max-stacks of glutamatergic MBONs (A) MB011B, (B) MB002B, (C) MB399B, (D) MB310C, (E) MB434B, (F) MB298B, GABAergic MBONs (G) MB110C and (H) MB057B, and cholinergic MBONs (I) MB077B, (J) MB018B, (K) MB026B, (L) MB080C, (M) MB082C, (N) MB542B, (O) MB050B, (P) MB051C, (Q) MB549C and (R) MB027B, trans-Tango identified postsynaptic connections. For max-stacks: green, presynaptic MBONs, magenta, postsynaptic trans-Tango signal, blue, brp-SNAP neuropil. A map of the MBONs that are included in the expression pattern in each driver line accompanies each exemplar with the relative expression pattern (grayscale, 1–5) accordingly to FlyLight (https://splitgal4.janelia.org/cgi-bin/splitgal4.cgi). MBON maps are organized by neurotransmitter type: green=glutamatergic, blue=GABAergic, red=cholinergic. Scale bar = 50 μm. However, it was clear that some of the data were confounded by split-GAL4 lines that had off-target expression. We excluded extraneous signals by segmenting trans-Tango signals that were continuous with MBON terminals (Figure 2A–B) and then quantified the distribution of postsynaptic signals across brain regions in the standard brain (Ito et al., 2014). Nearly all MBONs have divergent connections across the dorsal brain regions, CRE, SMP, SIP, SLP, LH, as well as FSB, and LAL (Figure 2C–E, Figure 2—figure supplement 1). Figure 2 with 1 supplement see all Download asset Open asset Whole brain distribution of MBON postsynaptic connections overlap. (A) Example of presynaptic MBON γ5β′2a (SS01308) and postsynaptic trans-Tango signal in a registered brain. For max-stacks: green, presynaptic MBONs, magenta, postsynaptic trans-Tango signal. (B) Example of segmented trans-tango signals that was continuous to MBON γ5β′2a terminals. For max-stack: gray, postsynaptic trans-Tango signal. (C) Heatmap displaying the overlap in segmented MBON postsynaptic signal by brain region. Postsynaptic signal for each MBON was normalized within each brain to capture respective expression levels. SS01308 was used to target MBON γ5β′2a, MB399B was used to target MBON β2β′2a, MB002B was used to target MBONs γ5β′2a, β′2mp, SS01143 was used to target MBON β′2mp, MB011B was used to target MBONs γ5β′2a, β′2mp, β′2mp_bi, MB057B was used to target MBON β′1, and MB110C was used to target MBONs γ3, γ3β′1. MB433B was used to target MBONs β1>α, γ4>γ1γ2, MB298B was used to target MBON γ4>γ1γ2, MB077C was used to target MBON γ2α′1 and MB50B was used to target MBONs α′1, α2sc. MB018B was used to target MBON α′2, MB027B was used to target MBON α′3ap, α′3 m, and SS01194 was used to target MBON α2sc. For raw postsynaptic signal see Figure 2—figure supplement 1. (D) Schematic of fly brain highlighting the most anterior brain regions included in mask analysis starting at AL and ending with SLP. (E) Schematic of fly brain highlighting the most posterior brain regions included in mask analysis starting at NO and ending with PB. AL: antennal lobe, AMMC: antennal mechanosensory and motor center, ATL: antler, AVLP: anterior ventrolateral protocerebrum, CRE: crepine, EB: ellipsoid body, EPA: epaulette, FSB: fan-shaped body, FLA: flange, GA: shoulder of lateral accessory lobe, GOR: gorget of ventral complex, IB: interior bridge, ICL: inferior clamp, IPS: inferior posterior slope, IVLP: inferior ventrolateral protocerebrum, LAL: lateral accessory lobe, LB: bulb of lateral complex, LH: lateral horn, MB: mushroom body, NO: noduli, OTU: optic tubercle, PB: protocerebral bridge, PLP: posterior lateral protocerebrum, PRW: prow, PVLP: posterior ventrolateral protocerebrum, SAD: saddle, SCL: superior clamp, SEG: subesophageal ganglion, SIP: superior intermediate protocerebrum, SLP: superior lateral protocerebrum, SMP: superior medial protocerebrum, SPS: superior posterior plate, VES: vest of ventral complex, WED: wedge. Scale bar = 50 μm. DANs are postsynaptic to MBONs Of the DANs innervating the MB, 90% have dendritic arborizations that are localized to four of the five proposed MBON convergent regions, including CRE, SMP, SIP, and SLP (Aso et al., 2014a). Subsets of MBON axons overlapping with DAN dendritic arborizations provide feedback opportunities for MBONs to modulate DAN input thereby indirectly modulating MB circuits. Thus, we selected a subset of MBONs that were reported to co-localize with protocerebral anterior medial (PAM) DANs and co-stained with antibodies against tyrosine hydroxylase (TH) to identify overlap with trans-Tango signal (Aso et al., 2014a). As expected, some of the neurons postsynaptic to MBONs were TH positive; however, due to the complexity of trans-Tango-labeled neurons, we were unable to identify the DANs postsynaptic to a particular MBON unequivocally. Most overlap between TH and trans-Tango signals was observed with γ5β′2a (MB011B, 25 ± 0.7; n = 4; Figure 3A) and β′2mp (MB002B, 10.25 ± 1.3 n = 4 and MB074C, 4.75 ± 1.1, n = 4; Figure 3B and C) MBONs. These MBONs were predicted to co-localize with PAM DANs β′2p, β′2m and PAM DANs γ5 and β′2a, respectively (Aso et al., 2014a). Similarly, the γ3, γ3β′1 MBON was predicted to overlap with PAM γ3 and β′1m, and MB083C had an average of nine cells (9 ± 2.0, n = 10) with co-expression of TH and trans-Tango signals (Figure 3D). Likewise, the cholinergic γ2α′1 MBON (MB077C) was predicted to overlap with PAM γ4>γ1γ2, and indeed, MB077C brains averaged five cells (5 ± 1.5, n = 8) with co-expression of TH and trans-Tango signals per hemibrain (Figure 3E). There were a number of MBONs that had very few or no TH-positive postsynaptic neurons (Figure 3—figure supplement 1). The majority of these MBONs innervate the vertical lobe, including MBON α1 (MB310C; Figure 3—figure supplement 1A), MBON α′3ap, α′3 m (MB027B; Figure 3—figure supplement 1E), MBON α2sc (MB080C; Figure 3—figure supplement 1F) and MBON α′1, α2p3p, α′3m (MB542B; Figure 3—figure supplement 1G). MBON β′1 also had limited TH-positive postsynaptic neurons (MB057B; Figure 3—figure supplement 1D). Despite predictions that γ4>γ1γ2 MBON (MB298B) would co-localize with PAM γ4>γ1γ2, we found minimal co-expression of TH and trans-Tango signals (Figure 3—figure supplement 1C). This is likely a false negative due to the strength of the driver as annotations of the EM data has revealed postsynaptic connections with PAM γ4>γ1γ2 (Clements et al., 2020; Li et al., 2020). It is possible that the number of co-localized TH+ cells in our analysis here is an underestimation since some of the brains had fewer than expected TH+ neurons (Figure 3—figure supplement 2). Figure 3 with 2 supplements see all Download asset Open asset DANs postsynaptic to MBONs. Exemplar max-stacks of MBON lines in which TH+ cells overlapped with postsynaptic signal of glutamatergic (A) MBON γ5β′2a, β′2mp, β′2mp_bilateral (MB011B), (B) MBON γ5β′2a, β′2mp (MB002B), (C) MBON γ5β′2a, β′2mp, β2β′2a (MB074C), (D) GABAergic MBONs γ3, γ3β′1 (MB083C) and (E) cholinergic MBONs γ2α′1 (MB077C). Overlapping TH+ and trans-Tango cell bodies are highlighted in insets, scale bar = 10 μm. Max stacks of MB are included (Column I), scale bar = 50 μm. Column II-IV depict single optical planes from anterior to posterior outlining MB compartments. Bar graphs indicate the average number of co-localized cells per hemibrain (mean +/- standard error). Green, TH-positive cells; magenta, postsynaptic trans-Tango signal. MBON maps are organized by neurotransmitter type: green=glutamatergic, blue=GABAergic, red=cholinergic. (F) Schematic depicting the MB innervation by PAM DANs. PAM DANs extend dendrites to SMP, CRE, SIP, and SLP. (G) Schematic depicting the MBONs that synapse on TH+ cells. Convergent MBONs Whole brain overlap analysis identified the MB itself as a site of rich convergence for most MBON lines (Figure 2C). MBON postsynaptic signals in MB were not surprising given that many MBONs provide feedforward connections between MB compartments (Aso et al., 2014a). For instance, MBON γ4>γ1γ2 has dendritic arbors in γ4 and axonal projections in γ1γ2, MBON γ1pedc>α/β have dendritic arbors in γ1 and axonal projections in α/β lobes, and MBON β1>α has dendritic arbors in β1 and axon projections to the entire alpha lobe. However, further analysis revealed that in addition to providing connections between MB compartments, MBONs converge directly on other MBONs presumably through axo-axonal connections. Two different MBONs are frequently targeted: MBON β′2mp (Figure 4A) and MBON γ3β′1 (Figure 4B). Interestingly, MBON β′2mp receives convergent glutamatergic, GABAergic, and cholinergic input from MBON γ5β′2a (MB011B and MB210B), MBON γ3β′1 (MB110C and MB83C), MBON α′2 (MB018B and MB082C), and MBON γ2α′1 (MB077B and MB051C) (Figure 4A, Figure 4—figure supplement 1). MBON γ3β′1 receives convergent input from glutamatergic MBON β′2mp as revealed with split-GAL4 lines MB002B (Figure 4B) and MB074C (Figure 4—figure supplement 1) and glutamatergic MBON γ4>γ1γ2 (MB298B, Figure 4B). We hypothesize that similar to MBONs that project to other regions of the MB, MBON γ3β′1, and MBON β′2mp create opportunities for multilevel feedforward networks to update information to drive behavioral response (Figure 4C). Figure 4 with 1 supplement see all Download asset Open asset Subsets of MBONs converge on other MBONs. (A) MBON β′2mp receives convergent input from glutamatergic MBON γ5β′2a (MB011B), GABAergic MBONs γ3, γ3β′1 (MB110C) and cholinergic MBON γ2α′1 (MB077B) and MBON α′2 (MB018B). (B) MBON γ3β′1 receives convergent input from glutamatergic MBON β′2mp (MB002B) and MBON γ4>γ1γ2 (MB298B). β′2mp, γ3 and β′1 are outlined in representative stacks. (C) Schematics summarizing identified convergent MBONs (β′2mp and γ3β′1) and their respective convergent input. Solid lines represent the convergent MBON and dotted lines represent convergent input. For max-stacks: green, presynaptic MBONs, magenta, postsynaptic trans-Tango signal, blue, brp-SNAP neuropil, scale bar=50 μm. Convergence outside the MB Another site of convergence of the MBON network was the FSB (Figure 5). MBON postsynaptic connections display a laminar organization primarily across the dorsal region of the FSB. Nearly all the glutamatergic and GABAergic MBONs converge onto FSB layers 4 and 5, and to a lesser extent, layer 6 (Figure 5A–B). MBON α1 is the only type of MBON that had broad trans-Tango signals in the FSB (Figure 5A). To rule out sexual dimorphism in the postsynaptic connections of MBON α1, we compared trans-Tango signal in the FSB in male and female brains and found similar innervation patterns (Figure 5—figure supplement 1). Cholinergic MBONs also had trans-Tango signals in the dorsal FSB but with more variability across MBON lines and within each line (Figure 5C). For instance, trans-Tango with MBON γ2α′1 consistently visualized projections to FSB layers 4 and 5 in all the brains analyzed, whereas more variability was observed in FSB innervation pattern across MBON α′2 brains (Figure 5—figure supplement 2). MBON α′1 and α2sc both project exclusively to FSB layer 6 (Figure 5C). Together, FSB layers 4 and 5 receive convergent input from combinations of glutamatergic, GABAergic and cholinergic MBONs (Figure 5D; Figure 5E). Figure 5 with 2 supplements see all Download asset Open asset MBONs converge on different layers of the FSB. Exemplar max-stacks of glutamatergic (A), GABAergic (B), and cholinergic (C) MBONs whose postsynaptic neurons innervate the FSB. Max-stacks are approximately 50 μm thick. Slices were selected based on the relative position of the FSB. For FSB stacks: magenta, postsynaptic trans-Tango signal, blue, brp-SNAP neuropil. Map of MBONs accompany each exemplar with the relative expression pattern (grayscale, 1–5) accordingly to FlyLight. For each map, green=glutamatergic, blue=GABAergic, red=cholinergic. Scale bar = 50 μm. (D) Map summarizing the percentage of trans-Tango-positive signal in each FSB layer across brains for each MBON. (E) Schematic depicting MBONs that converge onto different layers of the FSB. MB compartments are colorized based on the neurotransmitter expressed by the MBON that innervates it. Lines thickness corresponds to the percentage of trans-Tango-positive signal in each FSB layer across brains for each MBON. Both visual and computational analyses confirmed the CRE, SMP, SIP, and SLP, as well as the MB and FSB as obvious postsynaptic targets of the MBON network. Visual inspection also confirmed the LAL as postsynaptic to multiple MBON lines. Its identification was less obvious in computational analysis largely because the neurites innervating the LAL were not as extensive as the LAL itself and were often difficult to segment. Although not extensive, LAL innervation was consistent across glutamatergic, GABAergic, and cholinergic MBONs (Figure 6). Specifically, glutamatergic γ5β′2a, β′2mp, and β′2mp_bilateral had postsynaptic neurites within the LAL in all of the brains analyzed (Figure 6A). Similarly, GABAergic MBON γ3, γ3β′1, and β′1 (Figure 6B) and cholinergic MBON γ2α′1 (Figure 6C) consistently had postsynaptic neurites within the LAL. Thus, like the FSB, neurons innervating the LAL receives convergent input from combinations of glutamatergic, GABAergic and cholinergic MBONs (Figure 6D; Figure 6E). Figure 6 Download asset Open asset MBONs converge onto LAL neurons. Exemplar max-stacks of glutamatergic (A), GABAergic (B), and cholinergic (C) MBONs innervating the LAL. Max-stacks are approximately 50 μm thick. Slices were selected based on the relative position of the LAL. Magenta, postsynaptic trans-Tango signal, blue, brp-SNAP neuropil. Map of MBONs accompany each exemplar with the relative expression pattern (grayscale, 1–5) accordingly to FlyLight. For each map green=glutamatergic, blue=GABAergic, red=cholinergic. Scale bar = 50 μm. Scale bar for insets = 10 μm (D) Map summarizing the percentage of trans-Tango-positive signal in LAL across brains for each MBON. (E) Schematic depicting MBONs that converge onto neurons innervating the LAL. MB compartments are colorized based on the neurotransmitter expressed by the MBON that innervates it. Lines thickness corresponds to the percentage of trans-Tango-positive signal in LAL across brains for each MBON. Thus far, we have confirmed two postsynaptic targets of the MBON network that reside outside of the MB: the FSB and LAL. However, the identities of the postsynaptic neurons within FSB and LAL as well as their functions remain unknown. Our strategy for identifying FSB and LAL neurons and interrogating their functional connectivity with MBONs was to selectively label neurons in FSB and LAL using specific drivers and to examine whether they are co-localized with postsynaptic signal when we initiate trans-Tango from MBONs. To achieve this, we identified candidate FSB and LAL LexA lines by performing a mask search of the LexA lines that have overlapping expression within the convergent region and brought them together with MBON lines: MB051C and MB077C were used to target MBON γ2α′1, MB083C and MB110C were used to target γ3β′1, and MB074C was used to target MBON β′2mp. We identified three candidate LexA lines: one to target FSB layer four neurons - R47H09 (Jenett et al., 2012; Pfeiffer et al., 2013; Pfeiffer et al., 2010), and two to target LAL neurons - VT055139 and VT018476 (Tirian and Dickson, 2017). Finally, we generated trans-Tango reporter flies where the UAS-myrGFP was replaced with UAS-CD2, and LexAOp-mCD8::GFP was included in order to visualize the starter MBONs, the postsynaptic trans-Tango signal, and the LexA lines simultaneously. We successfully combined the majority of the targeted MBON split-Gal4 lines with FSB and LAL LexA lines (we were unable to combine MB074C with LexA line 47H09). Interestingly, for the cholinergic MBON γ2α′1 (MB077C), we identified at least two postsynaptic FSB neurons (labeled in the 47H09 LexA line; Figure 7A) and at least five postsynaptic LAL neurons (labeled in the VT055139 LexA line; Figure 7B). We next sought to interrogate functional connectivity between MBON γ2α′1 and 47H09 FSB neurons and VT055139 LAL neurons by combining optogenetic stimulation of MBON γ2α′1 using UAS-Chrimson and functional calcium imaging in FSB and LAL using LexAop-GCaMP6s. Stimulation of cholinergic MB077C with 400–500 ms of red light (627 nm) resulted in an increase in calcium signal in the FSB and LAL (Figure 7C). Similar activation of other cholinergic MBONs (MB080C), which do not innervate the LAL or layer 4 of the FSB, did not result in signal (Figure 7—figure supplement 1), supporting the specificity of this interaction and suggesting that the MBON γ2α′1 is both anatomically and functionally connected to the FSB and LAL. Strikingly, GABAergic MBON γ3β′1 (MB083C) also had at least one identified postsynaptic FSB neuron that was included in the expression of FSB 47H09 LexA line (Figure 7D) and at least two identified postsynaptic LAL neurons that were included in the expression of LAL VT055139 LexA line (Figure 7E). Thus, the genetically identified subsets of LAL and FSB neurons receive convergent input from GABAergic and cholinergic MBONs (Figure 7F). We hypothesize that the convergence of excitatory and inhibitory input onto both the LAL and FSB is critical for guiding behavior. Figure 7 with 3 supplements see all Download asset Open asset MBONs γ3β′1 and γ2α′1 converge onto the same subset of LAL and FSB neurons. Exemplar max-stacks of cholinergic MBON γ2α′1 (MB077C) postsynaptic connections and identified overlap with respective (A) FSB (47H09) and (B) LAL (VT015539). (C) Confirmation of functional connection with optogenetic activation of MB077C and calcium imaging of FSB neurons in SMP and FSB (47H09), and calcium imaging of LAL neurons in SMP (VT015539). The red bar indicates when the LED was on and the shutter was closed to protect the PMTs during LED stimulation. Exemplar max-stacks of GABAergic MBON γ3β′1 (MB083C) postsynaptic connections and identified overlap with respective (D) FSB (47H09) and (E) LAL (VT015539). Max-stacks are approximately 50 μm thick. Slices were selected based on the relative position of the LAL
