Skip to main content
. 2020 Dec 14;9:e62576. doi: 10.7554/eLife.62576

Figure 11. Dorsal accessory calyx (dACA).

The dendrites of the 60 α/βp KCs define the dACA. The pie chart shows a breakdown of the inputs to these KCs. The majority convey visual information, either directly from visual projection neurons (VPN; 9.8%) or through intermediate local visual interneurons (LVIN; 57.8%) that receive input from VPNs (see Figure 11—figure supplement 1). VPNs can be subdivided based on the location of their dendrites in either ME or LO, as indicated in the outer circle. More than two-thirds of the indirect input is mediated by the LVIN cell types SLP360, SLP362, and SLP371, shown in Figure 11—figure supplement 2C; this SLP360/361/371 cluster of 13 neurons contributes about 30% of total input to the 60 α/βp KCs in the dACA. Neurons of similar morphology have also been observed to be presynaptic to KCα/βp in the dACA in a recent study (Li et al., 2020). Another LVIN, MB-CP2 (LHPV3c1) (479935033), provides 12.6% of the input to KCs in the dACA; however, only a small percentage of its inputs are visual (see Figure 11—figure supplement 1A and 2E). The total visual information presented to KCs by VPNs and LVINs is indicated by the purple arc around the outer layer; it reflects the direct input from the VPNs plus the fraction of the LVIN input that represents visual input. The next most prominent inputs are KC-to-KC synapses in the dACA (13.3%), from APL (10.9%), from two octopaminergic neurons (2.8%; OA-VPM3, see Figure 3—figure supplement 1D, and OA-VUMa2, see Figure 3—figure supplement 1F); and local interneurons (n = 23; 2.3%). Remaining input, ‘others’, are input from 102 different neurons that are all weakly connected; and numbered sector 1 are mPNs (0.7%). The dendrites of KCα/βp neurons in the dACA (Tanaka et al., 2008; Zhu et al., 2003) are reportedly activated by bitter or sweet tastants (Kirkhart and Scott, 2015). However, the KCα/βp are not required for taste conditioning, which instead appears to depend on γ KCs (Kirkhart and Scott, 2015) and we were unable to identify strong candidates for delivering gustatory sensory information to the dACA. The PN VP5+Z adPN (5813063239) connects to two α/βp KCs has dendrites in the SEZ (Figure 9—figure supplement 3). But this is the only gustatory PN we can associate with the dACA, and it primarily projects to KCγm neurons through which it might participate in conditioned taste aversion (Kirkhart and Scott, 2015). (B) Color-coded synaptic connections from visual projection neurons (VPN; yellow) and local visual interneurons (LVIN; orange) onto α/βp KCs (gray). Note that, unlike in the vACA, there are more connections from LVINs than VPNs in the dACA.

Figure 11.

Figure 11—figure supplement 1. VPN and LVIN inputs to the dACA.

Figure 11—figure supplement 1.

(A) Plot of interneurons carrying visual information to the α/βp KCs ranked by the strength of their effective contribution of visual information. For a given interneuron, this quantity is computed by multiplying the number of synapses that interneuron makes onto α/βp KCs by the fraction of its input synapses that come from visual projection neurons. The green dots show this effective visual information quantity. Values are normalized to the most strongly connected LVIN, which was assigned a value of 1.0 (green scale on the right side of the plot). The blue line shows the cumulative amount of effective visual information conveyed by LVINs; over 80% of the input is delivered by the top 10 LVINs. The color-coded bars indicate the percentage of that neuron’s input that comes from VPNs, LVINs, and other neurons. The black number at the base of each bar shows the number of synapses made by that LVIN to KCs in the dACA. The LVINs whose ID numbers are shown in orange also provide input to the γd KCs in the vACA. Links to these 10 LVINs in neuPrint are as follows: SLP371, SLP360, CL357, SLP362, and MB-CP (LHPV3c1). Note that many of lower ranked LVINs that receive high levels of VPN input do not make strong connections to KCs and are therefore presumably performing some other role in integrating visual information. (B) Multiple sensory pathways contribute to the LVINs that connect to α/βp KCs. Effective input delivered by local neurons that each of the 19 LVINs that make at least a total of 20 synapses to α/βp KCs are shown, all of which are LVINs except one OA-neuron (329566174) which makes weak connections to most KCα/βp (50 of 60 total α/βp KCs). We did not find other interneurons strongly connected to the vACA KCs that were not LVINs; that is, interneurons that did not also receive VPN input. Effective input is calculated by multiplying the indicated neuronal population’s synaptic input to each of the 19 LVINs times the fraction of the α/βp KCs’ total input coming from that LVIN, and then summing across all 19 LVINs. Effective input is expressed as the percentage of the total input from these 19 LVINs to α/βp KCs that originated with the indicated neuronal population. The number of individual cells in each population is also indicated. Seventy percent of the most strongly connected upstream neurons (total of 350) to the 19 LVINs have been classified. The strongest inputs are VPNs and LVINs contributing combined 40.8% effective input. Local (those confined to the neuropil that is adjacent to dACA) and broad (those that expand through multiple neuropils) interneurons contribute a total of 16.0% of the effective input. Other prominent inputs are LHONs, 5-HT neurons (n = 2, 2.9%), SEZONs (n = 7, 2.0%), OA-neurons (n = 3, 1.5%), APL and DMP (1.4%) and mPNs (n = 5, 1.2%).
Figure 11—figure supplement 2. LVINs that conveys visual input onto α/βp KCs.

Figure 11—figure supplement 2.

(A) A plot showing the cosine similarity of inputs to each of the 19 LVINs that most strongly connected (more than 20 synapses) to α/βp KCs. (B) An additional view of the distribution of VPN and LVIN inputs onto α/βp KCs. Note that LVIN input locations tend to be more posterior than those of VPNs. (C) Of the top 10 LVINs based on visual input delivered to α/βp KCs, eight are neurons with similar morphologies that fall into three cell types SLP360, SLP362 and SLP371, shown here. See Figure 11—figure supplement 3 for more morphological details of SLP371. (D) The LVIN conveying the fourth highest visual input to the dACA, CL375. (E) MB-CP2, the LVIN conveying the eighth highest visual input to the vACA; MB-CP2 (LHPV3c1) also provides 1% of input to KCs in the CA (see Figure 9). (F) A plot showing the similarity of VPN inputs for each of the top eight LVINs (using the ranking from panel A). Each vertical blue bar represents a different VPN and the intensity of the color reflects the percentage of the corresponding LVIN’s total VPN input that is contributed by that VPN. The order of the VPNs and LVINs was determined using spectral clustering.
Figure 11—figure supplement 3. Detailed morphology of a dACA LVIN.

Figure 11—figure supplement 3.

(A) One of the two SLP371 neurons (5813011738; green) is shown with the positions of its synaptic input from three color-coded classes of VPNs. (B) Synapses from this LVIN (green) onto the subset of α/βp KCs (gray) that it contacts are shown as orange dots. (C) Enlarged and rotated view of the box in (B). Note that connections from the LVIN to α/βp KCs lack the claw-like structure typical of the synapses between olfactory projection neurons and KCs in the CA.
Figure 11—figure supplement 4. Similarity of VPN inputs to individual LVINs that innervate the vACA or the dACA.

Figure 11—figure supplement 4.

A heatmap showing the similarity of LVIN inputs from VPNs. It reveals the degree of diversity of the VPN inputs received by LVINs, as well as the existence of clusters of LVINs that receive similar inputs. Each square in the heat map represents a cosine similarity of the VPN inputs to a pair of LVINs. The ordering of the LVINs shown was determined by spectral clustering based on their inputs. Colors on the diagonals indicate whether the given LVIN is among the top 20 inputs to the vACA (green), to the dACA (pink), to both the vACA and dACA (orange), or neither (dark blue).
Figure 11—figure supplement 5. Similarity of non-visual inputs to individual LVINs.

Figure 11—figure supplement 5.

A heatmap in the same format as Figure 11—figure supplement 4 but showing the similarity of LVIN inputs from non-visual (i.e. non-VPN, non-LVIN) neurons. LVINs are shown in the same order as in Figure 11—figure supplement 4. The fact that substantial structure is preserved between these plots indicates that clusters of LVINs with similar visual inputs also tend to receive similar inputs from neurons that do not (directly) relay visual information. Colors on the diagonals indicate whether the given LVIN is among the top 20 inputs to the vACA (green), to the dACA (pink), to both the vACA and dACA (orange), or neither (dark blue).