Distinct protocerebral neuropils associated with attractive and aversive female-produced odorants in the male moth brain

Abstract

The pheromone system of heliothine moths is an optimal model for studying principles underlying higher-order olfactory processing. In Helicoverpa armigera, three male-specific glomeruli receive input about three female-produced signals, the primary pheromone component, serving as an attractant, and two minor constituents, serving a dual function, i.e. attraction versus inhibition of attraction. From the antennal-lobe glomeruli, the information is conveyed to higher olfactory centers, including the lateral protocerebrum, via three main paths - of which the medial tract is the most prominent. In this study, we traced physiologically identified medial-tract projection neurons from each of the three male specific glomeruli with the aim of mapping their terminal branches in the lateral protocerebrum. Our data suggest that the neurons' wide-spread projections are organized according to behavioral significance, including a spatial separation of signals representing attraction versus inhibition - however, with a unique capacity of switching behavioral consequence based on the amount of the minor components.

Full text

1
Distinct protocerebral neuropils associated with attractive and aversive female-1
produced odorants in the male moth brain 2
3
Jonas Hansen Kymre1†, XiaoLan Liu2, 3†, Elena Ian1, Christoffer Nerland Berge1, XinCheng 4
Zhao2, GuiRong Wang3, Bente G. Berg1, Xi Chu1* 5
6
1Chemosensory lab, Department of Psychology, Norwegian University of Science and 7
Technology, Trondheim, Norway. 8
2Department of Entomology, College of Plant Protection, Henan Agricultural University, 9
Zhengzhou, China. 10
3State Key Laboratory for Biology of Plant Disease and Insect Pests, Institute of Plant 11
Protection, Chinese Academy of Agricultural Sciences, Beijing, China. 12
13
†These authors contributed equally to this work. 14
* Corresponding author: xi.chu@ntnu.no 15
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

2
Abstract 16
The pheromone system of heliothine moths is an optimal model for studying principles 17
underlying higher-order olfactory processing. In Helicoverpa armigera, three male-specific 18
glomeruli receive input about three female-produced signals, the primary pheromone 19
component, serving as an attractant, and two minor constituents, serving a dual function, i.e. 20
attraction versus inhibition of attraction. From the antennal-lobe glomeruli, the information is 21
conveyed to higher olfactory centers, including the lateral protocerebrum, via three main 22
paths – of which the medial tract is the most prominent. In this study, we traced 23
physiologically identified medial-tract projection neurons from each of the three male-specific 24
glomeruli with the aim of mapping their terminal branches in the lateral protocerebrum. Our 25
data suggest that the neurons’ wide-spread projections are organized according to behavioral 26
significance, including a spatial separation of signals representing attraction versus inhibition 27
– however, with a unique capacity of switching behavioral consequence based on the amount 28
of the minor components. 29
30
Keywords 31
Olfaction; Pheromone; Interspecific signal; Macroglomerular complex; Morphology; 32
Physiology, Projection neurons. 33
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

3
Introduction 34
Olfactory circuits serve a central role in encoding and modulating sensory input from the 35
natural surroundings. Understanding how these chemosensory circuits translate signals with 36
different hedonic valences into behavior is an essential issue in neuroscience. With a 37
relatively simple brain and a restricted number of associated odors evoking opposite innate 38
behaviors, i.e. attraction and aversion, the insect pheromone pathway is an optimal system to 39
address this question. In the noctuid moth, pheromone-evoked behaviors are linked to a 40
hardwired circuit in the lateral protocerebrum, including the lateral horn. This brain region 41
shares many neural principles with the mammalian amygdala (Miyamichi et al., 2011; 42
Sosulski et al., 2011). In contrast to the random neuron connectivity in another higher-order 43
olfactory center of the insect, the mushroom body calyx (mammalian piriform cortex analog, 44
Su et al., 2009), the neuronal wiring in the lateral protocerebrum is characterized by a form of 45
spatial clustering. Here, neurons responding to food odors versus pheromones as well as 46
attractive versus aversive odors are spatially segregated (Grabe & Sachse, 2018). Thus, at the 47
level of the lateral protocerebrum, it appears that the relevant odor cues are represented in 48
different wide-spread sub-domains that display a form of spatial pattern according to 49
behavioral significance. 50
In the moth antennal lobe (AL; mammalian olfactory bulb analog), the male-specific 51
macroglomerular complex (MGC) receives input from olfactory sensory neurons (OSNs). The 52
MGC glomeruli process input about a few female-produced signals, of which one primary 53
constituent acts as an unambiguous attractant and others often enhance attraction at low doses 54
but serves as behavioral antagonists at higher doses (Chang et al., 2017; Gothilf et al., 1978; 55
Kehat & Dunkelblum, 1990; Wu et al., 2015; Zhang et al., 2012). From the MGC, the 56
pheromone signals are conveyed to higher brain centers, including the lateral protocerebrum, 57
by male-specific projection neurons (PNs) following three main tracts, i.e. the medial, 58
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

4
mediolateral, and lateral antennal-lobe tract (mALT, mlALT and lALT, respectively, see 59
Homberg et al., 1988; Lee et al., 2019). Although a considerable number of medial-tract MGC 60
neurons have previously been reported in various moth species (Anton et al., 1997; Berg et 61
al., 1998; Christensen & Hildebrand, 1987; Christensen et al., 1991; Christensen et al., 1995; 62
Hansson et al., 1994; Hansson et al., 1991; Jarriault et al., 2009; Kanzaki et al., 1989; Kanzaki 63
et al., 2003; Nirazawa et al., 2017; Seki et al., 2005; Vickers et al., 1998; Zhao & Berg, 2010; 64
Zhao et al., 2014), the main focus has so far been odor coding within the MGC. Only a few 65
studies have paid attention to the protocerebral projections of these medial-tract PNs (Kanzaki 66
et al., 2003; Seki et al., 2005; Zhao et al., 2014). Thereby, our two main questions are where 67
in the lateral protocerebrum these neurons project to and how the pheromone neural circuit at 68
this level processes information with opposite valences. 69
The male moth used in this study, Helicoverpa armigera (Lepidoptera, Noctuidae, 70
Heliothinae), utilizes cis-11-hexadecenal (Z11-16:Al) as the primary pheromone component 71
and cis-9-hexadecenal (Z9-16:Al) as a secondary component (Kehat & Dunkelblum, 1990). 72
Notably, the secondary constituent is identical with the primary pheromone component of the 73
coresidential and closely related species, Helicoverpa assulta (Berg et al., 2014), and could 74
thus act as an aversive signal at high concentrations. Another female-produced minor 75
component, cis-9-tetradecenal (Z9-14:Al), undoubtedly plays a dual role, acting as a 76
behavioral antagonist, i.e. inhibiting the attraction elicited by the primary pheromone at higher 77
dosages, i.e. >5% (Gothilf et al., 1978; Kehat & Dunkelblum, 1990), and as an agonist at 78
lower concentrations, i.e. 0.3-5% (Wu et al., 2015; Zhang et al., 2012). This functional duality 79
of a single molecular component indicates the complexity of the neural circuits processing 80
pheromone information. The system must be capable of encoding attractive and aversive 81
signals appropriately in order to elicit coordinated responses maximizing reproductive fitness. 82
In the study presented here, we characterized the male-specific PNs passing along the 83
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

5
prominent mALT and the slightly thinner mlALT in H. armigera, with focus on their 84
projection patterns in the lateral protocerebrum. Based on the intracellular recording/staining 85
technique, combined with calcium-imaging experiments, we have mapped the projection 86
patterns of physiologically identified MGC-neurons. The current study provides solid 87
evidence for a spatial arrangement within the relevant protocerebral neuropils of the male 88
moth demonstrating distinct regions receiving input about separate or intermixed 89
female-released signals associated with attraction versus inhibition of attraction. 90
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

6
Results 91
Mapping odor-evoked responses of the MGC output neurons by means of calcium imaging 92
The projection pattern of the male-specific sensory neurons onto the MGC units was 93
previously mapped via bath application calcium-imaging studies (Kuebler et al., 1998; Wu et 94
al., 2013; Wu et al., 2015). To measure the output signals from the same three MGC 95
glomeruli, we performed calcium-imaging measurements of odor-evoked responses in a group 96
of PNs exclusively. By applying a calcium-sensitive dye (Fura 2) into the calyces (Fig. 1A), 97
we label primarily the population of medial-tract uniglomerular neurons (Ian et al., 2016; 98
Kymre et al., 2020). In the species used here, H. armigera, the MGC comprises three units 99
(Skiri et al., 2005; Zhao et al., 2016) receiving input from three OSN categories: (1) the 100
cumulus from OSNs tuned to the primary component, Z11-16:Al, (2) the dorsomedial 101
posterior (dmp) unit from OSNs tuned mainly to the secondary component, Z9-16:Al, and (3) 102
the dorsomedial anterior (dma) unit from OSNs tuned to the behavioral antagonist/enhancer, 103
Z9-14:AL (Wu et al., 2015). For simplicity, Z9-14:Al is mentioned as a behavioral antagonist 104
in the subsequent text. The imaging data of mALT PNs connected to each of the three MGC 105
units was obtained from dorsally oriented brains (Fig. 1B). One example of repeated traces 106
illustrated that the increase in intracellular Ca2+ in the MGC during antennal stimulation with 107
the two pheromone components was consistent (Fig. 1C). 108
We also checked whether the population of mALT PNs from each MGC unit showed 109
consistent responses across individual insects. The response consistency of a group of PNs 110
can be quantified as the Pearson’s correlation coefficient of the response vectors in two 111
individual insects, where each vector contains the trial-averaged responses of an individual to 112
a given set of odors (Mittal et al., 2020; Schaffer et al., 2018). The average correlation 113
between the cumulus medial-tract PNs responses (mean calcium signal during stimulation 114
windows subtracted by that during the pre-stimulation period) across individuals was 0.45. 115
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

7
The corresponding average correlations of PNs from the dma and dmp units were 0.55 and 116
0.24, respectively (Fig. 1D). These paired-individual correlations were greater than the 117
nonlinear relationship with a chance level of 0 (t-test, ps<0.002), confirming the general 118
response consistency across different insects. We profiled the pheromone responses in mALT 119
PNs from each MGC unit in 8 insects, by comparing the Fura signals evoked by each stimulus 120
during the 2s stimulation window with the control (hexane) (Fig. 1E). The average calcium 121
traces observed in these insects are presented in Figure 1 – figure supplement 1. As expected, 122
the cumulus PN population showed a pronounced activation during stimulation with the 123
primary pheromone and the pheromone blend. The dma output neuron population, in turn, 124
responded not only to the behavioral antagonist but also to the primary pheromone and the 125
pheromone blend. The dmp projection neurons responded to the secondary pheromone and 126
the behavioral antagonist, as well as to the pheromone mixture. All these responses showed a 127
phasic component that decayed over the course of the stimulation period. Taken together, 128
each stimulus evoked a unique activation pattern in the three MGC units (Fig. 1E). We also 129
analyzed the mean calcium trace of 8 individuals across 7 stimuli (see Odor stimulation 130
section in Materials and methods). The across-stimuli correlation plot illustrates that, unlike 131
the relatively defined responses of the PNs innervating the cumulus and the dmp unit, the 132
population of dma PNs evoked a broad activation pattern including responses to all female-133
produced chemicals tested (Figure 1 – figure supplement 2). In contrast to the narrowly tuned 134
male-specific OSNs, we found that the odor response profiles of the medial-tract output 135
neurons were considerably more intricate. 136
137
Morphological characteristics of individual MGC projection neurons 138
We next aimed to elucidate the functioning and morphology of individual neurons involved in 139
processing pheromone information. Intracellular recording and staining were executed from 140
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

8
the thick dendrites of MGC output neurons, including PNs confined to both the mALT and 141
mlALT (Fig. 2 and Fig. 2 – figure supplement 1). We labeled 35 PNs across 32 preparations 142
(see supplementary table S1), along with four preparations with multiple co-labeled 143
heteromorphic MGC neurons. 144
145
MGC PNs in the medial tract 146
In total, 29 individual MGC PNs confined to the mALT were stained. For simplification, any 147
identical PNs that were co-labeled are referred to as a single PN, as physiological recordings 148
contained only one analyzed waveform. All mALT PNs had their somata in the medial cell 149
body cluster. Amongst these neurons, 24 had uniglomerular dendrites, of which 16 arborized 150
in the cumulus (Fig. 2B), four in the dmp (Fig. 2C), and four in the dma (Fig. 2D-E). In 151
addition, five mALT PNs had multiglomerular dendrites. One of these multiglomerular PNs 152
arborized in both the dma and posterior-complex glomeruli (Fig. 2F), while the remaining 153
four PNs innervated several MGC units, with different densities across the glomeruli. Two of 154
these multiglomerular PNs had evenly distributed dendrites across the MGC (Fig. 2G), while 155
the other two had dense innervation of either the dma or dmp, and sparse innervation in the 156
remaining MGC glomeruli (Fig. 2H-I, respectively). 157
Generally, the protocerebral projection patterns of mALT PNs innervating the 158
cumulus were homogeneous. These PNs projected sparsely to a restricted area in the inner 159
layer of the calycal cups before targeting two main regions in the lateral protocerebrum, i.e. 160
the ventrolateral protocerebrum (VLP) and superior lateral protocerebrum (SLP). Generally, 161
the SLP was innervated through two separate branches targeting anterior and intermediate 162
regions, respectively, with the most ventral parts positioned close to the superior clamp. In 163
addition, at least seven of the 16 cumulus PNs innervated the posterodorsal part of the 164
superior intermediate protocerebrum (SIP), localized adjacent to the vertical lobe. Besides the 165
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

9
numerous cumulus-neurons, eight uniglomerular PNs originated in the two smaller MGC-166
units, four in dma and four in dmp. These PNs targeted primarily the VLP and the 167
anteroventral lateral horn (LH), where they overlapped, as demonstrated from two 168
preparations where uniglomerular dma- and dmp-PNs were co-labeled (Fig. 2J). Moreover, 169
two dma-PNs had additional axon terminals in the region between the LH and SLP. The target 170
areas of the five multiglomerular mALT PNs resembled those of the uniglomerular dma- and 171
dmp-PNs, as their output terminals were confined to the VLP and anteroventral LH. All the 172
mALT PN innervations in the VLP were spatially restricted, primarily into the most 173
dorsoanterior regions of this neuropil. 174
175
MGC PNs in the mediolateral tract 176
In addition to the medial-tract PNs, we labeled three multiglomerular MGC PNs projecting 177
through the mediolateral ALT, which is considerably thinner than the medial tract and 178
projects directly to the lateral protocerebrum without innervating the calyces. Two of these 179
PNs arborized in all three MGC units with evenly distributed dendrites (Fig. 2K). In contrast 180
to most non-MGC mlALT PNs, innervating dorsal neuropils such as the SLP (Kymre et al., 181
2020), both PNs projected solely to ventral neuropils, mainly the VLP. Similar neurons were 182
previously reported in another heliothine species (Lee et al., 2019). The final mlALT MGC-183
PN had dense dendritic arborizations in the cumulus and sparse innervations in the dma and 184
dmp along with a few posterior-complex glomeruli (Fig. 2L). Its axonal projections targeted 185
the VLP, SLP, and posterodorsal SIP, like the cumulus-PNs of the medial tract. 186
187
Output regions of PNs across distinct antennal-lobe tracts 188
Previous and present data including confocal images and 3D reconstructions of MGC PNs 189
confined to the medial, mediolateral, and lateral ALTs, indicated that their output regions 190
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

10
could potentially overlap in specific protocerebral regions (Chu et al., 2020a). To explore 191
whether such PNs intersect, we next performed multi-staining experiments. One preparation 192
contained three individual MGC neurons confined to each of the three main tracts (Fig. 3A). 193
Here, one mALT PN and one lALT PN were strongly labeled, whereas the mlALT PN was 194
moderately stained. The lateral-tract neuron projected to the ipsilateral column of the SIP, as 195
previously described by Chu et al. (2020a). This region was not innervated by any other 196
MGC-PN types. PNs confined to the medial and mediolateral tract, however, overlapped in 197
the VLP (dashed circle in Fig. 3A). Two co-stained cumulus-PNs in another preparation, 198
included one lateral-tract PN targeting the anteriorly located column of the SIP and one 199
medial-tract PN targeting the VLP, SLP, and posterior SIP (Fig. 3B). These PNs had no 200
overlapping terminals. 201
202
Physiological characteristics of the MGC PNs 203
Next, we characterized the electrophysiological features of the labelled PNs. The summarized 204
mean spiking activity of all individual PNs, indicating which responses were significantly 205
different from pre-stimulation firing rates, are reported in Fig. 4 (electrophysiological traces 206
are in Fig. 4 - figure supplement 1-7). In addition, we report a heat map of every neuron`s 207
mean Z-scored instantaneous firing rate during application of each female-produced stimulus 208
(Fig. 5), as well as repeated trials of the same stimulus in individual PNs (Fig. 5 – figure 209
supplement 1-3). Each neuron was named with a unique ID. For PNs having dendritic 210
innervation only in one of the MGC units, the neuron ID is expressed as “innervated 211
glomerulus-ALT number/letter”, and for PNs innervating multiple MGC units as “MGCmain
212
innervated glomerulus-ALT number/letter”. Here the field of “number/letter” represents two different 213
concentration protocols (see Odor stimulation section in the Materials and methods). Finally, 214
we averaged the responses of all individually recorded PNs, and describe how the sampled 215
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

11
MGC PNs represent pheromone signals across distinct MGC units and protocerebral 216
neuropils. 217
218
Medial-tract PN response-profiles were only partly congruent with OSN inputs 219
The physiological profiles within each neuronal group innervating the same MGC unit varied 220
to some extent, both with respect to odor discrimination and temporal response 221
characteristics. The uniglomerular cumulus mALT PNs commonly responded with increased 222
spiking frequency to both low and high concentrations of the primary pheromone and the 223
pheromone mixture (Fig. 4 and Fig. 4 - figure supplements 1-2). In most cases, the responses 224
were characterized by a sharp phasic onset that gradually faded towards baseline activity (Fig. 225
5). The phasic onset was most prominent for PNs in the low concentration protocol (Fig. 4 - 226
figure supplement 1D); these neurons also appeared to discriminate between odorants more 227
precisely than PNs in the high concentration protocol (Fig. 4 - figure supplement 1E). This 228
coincides with previous studies reporting that OSNs display less specific response profiles 229
with increasing odor concentrations (Malnic et al., 1999; Sato et al., 1994). 230
Notably, 25 % of the cumulus-PNs did not respond to the primary pheromone, and 231
31 % were excited by the behavioral antagonist and/or the secondary component (Fig. 4 and 232
Fig. 4 – figure supplement 2). As these responses seem unlikely to arise due to OSN input, we 233
hypothesized they might be mediated by local interneurons or multiglomerular MGC-PNs, 234
which in case should imply delayed responses. Therefore, we quantified the onset and peak 235
response latency of each recorded trial for the neurons responding to the relevant stimuli (Fig. 236
4 – figure supplement 2G). There were no clear distinctions across the different stimuli with 237
respect to response onset latencies, but the peak latencies appeared to be delayed in response 238
to the secondary pheromone. 239
The four uniglomerular dma PNs included two morphological sub-groups. The first 240
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

12
one, including two PNs having fine branches approaching the SLP in addition to the terminals 241
in the LH and VLP, responded with significant excitation to the behavioral antagonist and the 242
secondary pheromone (Fig. 4 – figure supplement 3A-C). The second sub-group, consisting of 243
the two remaining dma-PNs with restricted projections in the VLP and LH, were not excited 244
by the behavioral antagonist nor the secondary compound (Fig. 4 and Fig. 4 – figure 245
supplement 3D-F). Three of the dma PNs, including the two latter neurons, were excited by 246
the pheromone mixture and/or primary pheromone. This is in agreement with our 247
calcium-imaging data, demonstrating excitation in the population of dma-PNs during 248
application of the stimuli including the primary pheromone. In addition to the uniglomerular 249
dma-PNs, it is relevant to mention one multiglomerular PN here, arborizing in the dma and 250
three posterior-complex glomeruli; this PN appeared to be excited by the behavioral 251
antagonist and inhibited by the primary pheromone (Fig. 4 – figure supplement 4), but 252
nonsignificantly (Fig. 4). 253
The four uniglomerular mALT PNs innervating the dmp were stimulated with the 254
high-concentration protocol only. Even though the dmp unit reportedly receives input about 255
the secondary pheromone and the behavioral antagonist (Wu et al., 2015), only one of the 256
dmp-PNs identified here was significantly excited by both of these components (Fig. 4 – 257
figure supplement 5). Of the three remaining dmp-PNs, one was excited by the behavioral 258
antagonist exclusively, whereas two displayed no responses during any stimulus application. 259
The AL innervations of the four multiglomerular medial-tract PNs innervating all 260
MGC units varied considerably. To investigate a putative association between MGC 261
innervation and response characteristics, we quantified the dendritic density of each PN by 262
measuring the fluorescence intensities within the separate MGC units (Fig. 4 – figure 263
supplement 6), as previously performed (see Chu et al., 2020b; KC et al., 2020). One PN, 264
arborizing primarily in the dmp, displayed phasic responses to the secondary pheromone and 265
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

13
the behavioral antagonist. The second multiglomerular PN, innervating primarily the dma, 266
had no responses to any of the female-produced compounds. The remaining two 267
multiglomerular PNs, with approximately evenly distributed dendrites, had divergent response 268
profiles. One exhibited a significant tonic excitation only to the behavioral antagonist, while 269
the other showed no responses to any stimuli. 270
271
The responses of mlALT PNs corresponded with OSN inputs 272
The three mediolateral-tract MGC-PNs consisted of two morphological sub-types displaying 273
different response profiles. The singular PN constituting the first sub-type targeted the VLP, 274
SLP, and SIP, while its dendrites filled the cumulus densely and the dma, dmp, and some 275
posterior complex glomeruli sparsely. This neuron responded with a weak and early phasic 276
excitation to most stimuli. Yet, the primary pheromone induced the only significant response 277
(Fig. 4 & 5; Fig. 4 – figure supplement 7). The other sub-type, including two mlALT PNs, 278
covered the MGC units quite uniformly. These PNs were more broadly tuned, responding 279
with excitation to all stimuli, however their responses to the pheromone mixture and the 280
primary pheromone lasted substantially longer than the phasic excitations elicited by the 281
secondary component and the behavioral antagonist (Fig. 4 – figure supplement 7; Fig. 5). 282
283
Representation of output signals from MGC PNs 284
Based on the precise electrophysiological and morphological data obtained from intracellular 285
recordings, we took the mean response of MGC medial- and mediolateral-tract PNs, to 286
provide a summary of odor-stimuli representation across the separate MGC units and the 287
protocerebral output regions, respectively (Fig. 6). We first categorized the neurons into four 288
groups based on their dendritic arborization, i.e. in the cumulus, dma, dmp, or in all MGC 289
units. The mean responses showed that mALT PNs with dendrites in the cumulus deal with 290
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

14
the primary pheromone, and those in the dmp unit with the secondary pheromone and the 291
behavioral antagonist (Fig. 6A). This is consistent with the calcium-imaging results on 292
medial-tract PNs (Fig. 1), and with former reports on response properties of the corresponding 293
OSNs (Wu et al., 2015). However, unlike previous reports on OSN input, both our individual 294
PN recordings and calcium-imaging tests indicated a role for the dma PNs in processing 295
information not only about the behavioral antagonist, but also, surprisingly, about the primary 296
pheromone and pheromone mixture. However, averaging the activity across the uniglomerular 297
dma neurons, demonstrated that the most potent stimuli for these PNs were the behavioral 298
antagonist and the pheromone mixture, while the primary pheromone elicited only minor 299
increases in mean firing rates. The responses to the mixture and the primary pheromone could 300
be influenced by other multiglomerular antennal-lobe neurons. For the fourth group, which 301
consists of the multiglomerular mALT and mlALT PNs arborizing in the entire MGC, the 302
averaged data demonstrated activation by all female-produced components. 303
Next, we investigated how the activity of MGC medial- and mediolateral-tract PNs is 304
represented in the higher brain regions (Fig. 6B). For the protocerebral output areas, several 305
neuropils, such the VLP, LH, SLP, and the posterior SIP, contained terminals of 306
medial/mediolateral-tract MGC neurons (Fig. 2). We found that the SLP and SIP received 307
signals regarding the primary pheromone, whereas the LH received input mainly about the 308
behavioral antagonist. The second pheromone was also represented in the LH, but with a 309
rather low intensity. All female-produced substances were represented the VLP. 310
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

15
Discussion 311
In moths, the sex pheromone is usually produced as a blend of several components in a 312
species-specific ratio (Baker & Hansson, 2016; Christensen et al., 1995). Principally, the male 313
is attracted by the major pheromone component released by a conspecific female. While 314
minor components do not elicit upwind flight on their own; they may enhance attraction 315
(Kehat & Dunkelblum, 1990), and can also serve as behavioral antagonists, since such 316
components are often produced by heterospecific females (reviewed by Berg et al., 2014) or 317
immature conspecific females (Chang et al., 2017). Despite these innate responses, the mate-318
searching activities, including an initial surge and zig-zag casting behavior (Cardé & Willis, 319
2008; Kuenen & Carde, 1994; Vickers & Baker, 1994), are not simple olfactory reflexes. The 320
data presented here, comprising a large number of MGC medial-tract PNs, mainly originating 321
from one of three easily identifiable glomeruli, indicated that such behavioral responses are 322
related not only to spatial representation of odor valence in the lateral protocerebrum, but also 323
to the intensity of the relevant signals. 324
325
Pheromone signaling along parallel antennal-lobe tracts 326
In a recent study, we found that lateral-tract MGC PNs, which appear to originate from the 327
cumulus only, convey a robust and rapid signal about the primary pheromone into a specific 328
target area, i.e. the column in the SIP (Chu et al., 2020a). In the study presented here, we 329
explored MGC-PNs in the two remaining tracts, the medial and the mediolateral. To clarify 330
how the pheromone-information is carried along the three parallel tracts, we integrated the 331
morphological findings from the previous study with our current results, and constructed a 332
comprehensive map displaying the neural connections between MGC/AL glomeruli and the 333
protocerebral target areas for each tract (Fig. 7A-B). 334
As pheromone-signaling appears to be particularly fast in lateral-tract MGC PNs (Chu 335
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

16
et al., 2020a), we decided to illustrate how the temporal response properties of MGC PNs 336
from the three main tracts are represented in the protocerebral neuropils. Interestingly, when 337
including the data from lateral-tract MGC PNs (Chu et al., 2020a), we discovered a sequential 338
and logic response pattern in the different protocerebral areas (Fig. 7B-C). Here, the 339
excitatory response to the pheromone mixture arose first in the SIP, as a typical phasic 340
response. After 20 ms, a phasic-tonic response to the binary mixture appeared in the SLP and 341
VLP. Finally, about 120 ms later, a weak signal lasting for 10 ms could be seen in the LH. 342
The primary pheromone evoked similar temporal response patterns in the SIP, SLP, and VLP, 343
but barely any signal could be observed in the LH. The secondary pheromone elicited weak 344
and delayed excitation in the VLP and LH, corresponding in time with the tiny peak evoked 345
by the binary mixture in the LH. Finally, the behavioral antagonist induced a clear and long-346
lasting response in the LH and a considerably weaker activation in the VLP. Notably, the very 347
fast response in the SIP region, which occurred during stimulation with the pheromone blend 348
and the primary constituent exclusively involves the previously reported lateral-tract MGC-349
PNs terminating in the column (Chu et al., 2020a). Comparing the response onset in the SIP 350
with the other output regions, demonstrates that the lateral-tract PNs react prior to PNs in the 351
other main tracts (Fig. 7C). 352
353
Protocerebral output-areas of MGC medial-tract neurons are organized according to 354
MGC-unit origin 355
Despite the relatively wide-spread and partly overlapping terminal branches of the total 356
assembly of medial-tract MGC-PNs, including 16 originating in the cumulus, 4 in the dmp, 4 357
in the dma, and 5 being multiglomerular, we discovered a pattern implying a spatial 358
arrangement according to the cumulus-PNs versus the dma- and dmp-PNs. As discussed in 359
detail below, this may correspond to the ‘lateral-medial’ representation of attraction- and 360
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

17
aversion-related pheromone-signals previously reported in the protocerebrum of the closely 361
related Helicoverpa species, H. assulta, and also in corresponding areas in the silk moth, 362
Bombyx mori (Kanzaki et al., 2003; Seki et al., 2005; Zhao et al., 2014). In H. armigera, we 363
found that signal representation related to attraction and inhibition is both segregated and 364
integrated, but in distinct protocerebral neuropils. 365
366
Inputs related to sexual attraction are primarily processed in the SLP and SIP 367
All uniglomerular mALT PNs innervating the SLP and SIP had their dendrites in the 368
cumulus. These cumulus-PNs responded mainly to the primary pheromone. Although the 369
behavioral antagonist excited 19% (3 of 16) of the cumulus-PNs as well, these responses were 370
not at all comparable to those associated with the primary pheromone. All data, including our 371
calcium-imaging measurements (Fig. 1) and mean intracellular responses (Fig. 6A) confirm 372
that the cumulus is devoted to process input about the primary pheromone. Notably, the 373
multiglomerular mlALT PN innervating the SLP and SIP had its most dense innervations in 374
the cumulus as well (Fig. 2L). Thus, third-order pheromone-processing neurons in the SLP 375
and SIP should primarily be involved in computations related to attraction. 376
377
Anteroventral parts of the LH process dma and dmp input 378
In contrast to the SLP and SIP, being innervated by the medial-tract cumulus-neurons, the 379
anteroventral LH was the target for the PNs originating in one of the two smaller MGC-units, 380
the dma and dmp. Thereby, input about the primary pheromone is processed dorsomedially to 381
input about the secondary component and the behavioral antagonist. This lateral-medial 382
separation resembles the spatial pattern previously reported in the closely related species, H. 383
assulta, where the primary pheromone is represented medially to the interspecific signal 384
(Zhao et al., 2014). In the “older” species H. assulta, however, the secondary pheromone is 385
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

18
not processed together with the behavioral antagonist, but with the primary pheromone. 386
Summarizing the response patterns of the MGC PNs innervating the LH (Fig. 6B & 387
7C2), we found that the highest mean firing rate appeared during application of the behavioral 388
antagonist. The secondary pheromone (plus the pheromone blend) also induced increased 389
spike frequency, but substantially weaker. As mentioned in the introduction, the behavioral 390
antagonist, Z9-14:Al, reduces mating-associated behaviors at high concentrations (Kehat & 391
Dunkelblum, 1990; Wu et al., 2015), while it enhances such behaviors at low concentrations 392
(Wu et al., 2015; Zhang et al., 2012). Indeed, low concentrations of this component is released 393
by H. armigera females (Kehat & Dunkelblum, 1990; Zhang et al., 2012), whilst sympatric 394
Heliothis peltigera females emit higher concentrations of the signal (Hillier & Baker, 2016). 395
The secondary pheromone component, Z9-16:Al, resembles Z9-14:Al in that it may also 396
include both enhancement and inhibition of attraction, dependent on the concentration. 397
Concretely, Z9-16:Al is known to augment attraction when combined with the primary 398
pheromone, but to reduce attraction when constituting more than 10 % of the binary 399
pheromone mixture (Kehat & Dunkelblum, 1990; Kehat et al., 1980). Notably, high 400
concentrations of Z9-16:Al may serve as an interspecific signal, due to it being the primary 401
pheromone of sympatric H. assulta females (Cork et al., 1992). Altogether, in light of the 402
complex behavioral functions of Z9-16:Al and Z9-14:Al, our data indicate that the 403
anteroventral part of the LH is involved in inhibition of attraction when the moth encounters 404
high concentrations of the relevant components. 405
406
A proposed interaction between the LH and the SLP/SIP 407
How can each of the two substances, Z9-14:Al and Z9-16:Al, serve opposite functions? 408
During mate-searching behavior, the likelihood of encountering the primary pheromone is 409
much larger than detecting the minor components, due to their proportion ratios in the natural 410
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

19
female-released pheromone mixture. As such, the SLP and SIP, which receive information 411
about the primary pheromone, is likely to be activated prior to the anteroventral LH. Notably, 412
a prominent neural link between the LH and SLP in moths (Namiki & Kanzaki, 2019) appears 413
to be a strong candidate for governing the “correct” mate-searching behavior. A study in the 414
fruit fly has demonstrated that the LH output projects primarily to SLP and secondly to the 415
SIP, and about one third of these neurons are GABAergic/glutamatergic (Dolan et al., 2019). 416
Since both neurotransmitters are inhibitory (Liu & Wilson, 2013), this indicates that LH input 417
may block the activation of SLP and SIP output neurons and their downstream circuits. The 418
interaction between the LH and SLP/SIP could act as a Boolean logic gate implementing the 419
AND-NOT function, only passing information downstream when aversive signaling is 420
minimally present in the anteroventral LH. 421
We propose that this form of Boolean logic represents a flexible interaction between 422
the LH and SLP/SIP. The LH output is mainly modulated by the input from the AL PNs, and 423
the activity of these neurons increases with increasing odor concentration (Gupta & Stopfer, 424
2012; Lerner et al., 2020; Sachse & Galizia, 2003). When a moth encounters the primary 425
pheromone and a very low concentration of the secondary pheromone (released by 426
conspecific female), the LH input is weak and the inhibition from the LH to SLP/SIP 427
minimal, therefore, the output from the SLP/SIP to downstream targets remains strong. 428
Should a big amount of the secondary pheromone or the behavioral antagonist (released by a 429
heterospecific female) recruit vigorous MGC PN output onto the LH, the 430
GABAergic/glutamatergic LH neurons will inhibit the activity in SLP/SIP and the 431
downstream signaling. This might be perceived as a loss of contact with the primary 432
pheromone, leading the moth to quickly engage in a casting flight, i.e. crosswind flight with 433
no net upwind movement (Kuenen & Carde, 1994), and thereby promote an appropriate mate-434
searching strategy. 435
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

20
Given that the SLP/SIP is involved in processing pheromone information solely about 436
attraction, while the LH represents concentration-dependent ambiguous signals, one question 437
appearing is how the presence of low-concentrations of the minor components enhances 438
attraction when added to the primary pheromone. Concerning Z9-16:Al, for example, our 439
findings indicate that integration of the secondary and primary pheromone might occur at the 440
level of MGC output PNs. Here, we first analyzed the physiological data of 11 uniglomerular 441
medial-tract PNs with projections to SLP/SIP having excitatory responses to the primary 442
pheromone and pheromone mixture. Their response amplitudes as well as response peaks 443
were compared during stimulation with the primary pheromone alone and the binary mixture 444
(secondary pheromone < 5%). In both electrophysiological recording and calcium-imaging 445
experiments, the pheromone mixture showed a tendency of evoking a stronger response (Fig. 446
13E). We also found at least three cumulus-PNs with projections in the SLP that had a tonic 447
and delayed excitation to the secondary pheromone (Fig. 4 and Fig. 12). 448
449
The pheromone-processing role of the VLP is complex 450
Unlike the neuropils discussed above, the VLP was innervated by all labelled MGC PNs. 451
Specifically, they sent terminal projections into a compact sub-domain of the dorsoanterior 452
VLP, which thereby gets intermingled signals about all female-relevant compounds. This 453
suggests that the VLP is involved in combinatorial coding, possibly recognizing the optimal 454
species-specific signal. In addition, combinatorial coding of pheromone signals seems to 455
occur at an earlier level as well. The dendritic organization of the multiglomerular MGC PNs 456
prime these neurons for combinatorial coding. This was, perhaps, most clearly demonstrated 457
by the broad response profiles of the mediolateral-tract PNs with evenly distributed dendrites 458
across the MGC (Fig. 4 and Fig. 4 - figure supplement 7). These PNs responded to all 459
female-produced components, but with distinct temporal patterns (Fig. 5), and may provide 460
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

21
excitatory or inhibitory input to the medial-tract PNs in the VLP since about half of the axons 461
forming this tract is GABAergic (Berg et al., 2009). Furthermore, we demonstrated that these 462
mlALT PNs had overlapping terminals with the medial-tract MGC PNs in this neuropil (Fig. 463
3A). As the VLP is the only known region with overlapping terminals of pheromone PNs 464
across different tracts, this makes it a particularly interesting neuropil for future studies 465
investigating parallel processing in the pheromone system. 466
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

22
Materials and methods 467
Insects 468
Male moths (2-3 days) of H. armigera were used in this study. Pupae were purchased from 469
Keyun Bio-pesticides (Henan, China). After emergence, the moths were kept at 24 °C and 470
70% humidity on a 14:10 h light/dark cycle, with 10% sucrose solution available ad libitum. 471
According to Norwegian law of animal welfare, there are no restrictions regarding 472
experimental use of Lepidoptera. 473
474
Calcium imaging 475
Totally, 8 males (2-3 days) were used in calcium-imaging experiments. Retrograde selective 476
staining of MGC PNs has been reported elsewhere (Chu et al., 2020a; Ian et al., 2017; Sachse 477
& Galizia, 2002; Sachse & Galizia, 2003). A glass electrode tip coated with Fura-2 dextran 478
(potassium salt, 10,000 MW, Molecular Probes) was inserted into the calyces to selectively 479
label the medial-tract PNs. The insects were then kept for 12 h at 4 °C before the experiment, 480
to facilitate retrograde transportation. 481
In vivo calcium imaging recordings were obtained with an epifluorescent microscope 482
(Olympus BX51WI) equipped with a 20x/1.00 water immersion objective 483
(OlympusXLUMPlanFLN). Images were acquired by a 1344x1224 pixel CMOS camera 484
(Hamamatsu ORCA-Flash4.0 V2 C11440-22CU). The preparation was excited with 340 nm 485
and 380 nm monochromatic light, respectively (TILL Photonics Polychrome V), and data 486
were acquired ratiometrically. A dichroic mirror (420nm) and an emission filter (490-530nm) 487
were used to separate the excitation and emission light. Each recording consisted of 100 488
double frames at a sampling frequency of 10 Hz with 35 ms and 10 ms exposure times for the 489
340 nm and 380 nm lights, respectively. The duration of one recording trial was 10 s, 490
including 4 s with spontaneous activity, 2 s odor stimulation, and a 4 s post-stimulus period. 491
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

23
The odor stimulation was carried out by a stimulus controller (SYNTECH CS-55), via which 492
humidified charcoal filtered air was delivered through a 150 mm glass Pasteur-pipette with a 493
piece of filter paper containing the stimulus. Each odor stimulus was applied twice. The 494
interval between trials was 60 s to avoid possible adaptation. 495
496
Intracellular recording and staining 497
Preparation of the insect has been described in detail elsewhere (KC et al., 2020; Kymre et al., 498
2020). Briefly, the moth was restrained inside a plastic tube with the head exposed and then 499
immobilized with dental wax (Kerr Corporation, Romulus, MI, USA). The brain was exposed 500
by opening the head capsule and removing the muscle tissue. The exposed brain was 501
continuously supplied with Ringer's solution (in mM): 150 NaCl, 3 CaCl2, 3 KCl, 25 sucrose, 502
and 10 N-tris (hydroxymethyl)-methyl-2-amino-ethanesulfonic acid, pH 6.9). 503
The procedure of intracellular recording/staining of neurons was performed as 504
previously described (Chu et al., 2020a; Ian et al., 2016; Zhao et al., 2014). Sharp glass 505
electrodes were made by pulling borosilicate glass capillaries (OD: 1 mm, ID: 0.5 mm, with 506
hollow filament 0.13 mm; Hilgenberg GmbH, Germany) on a horizontal puller (P97; Sutter 507
Instruments, Novarto, CA, USA). The tip of the micro-pipette was filled with a fluorescent 508
dye, i.e. 4% biotinylated dextran-conjugated tetramethylrhodamine (3000 mw, micro-ruby, 509
Molecular Probes; Invitrogen, Eugene, OR, USA) in 0.2 M KAc. The glass capillary was 510
back-filled with 0.2 M KAc. A chlorinated silver wire inserted into the muscle in the 511
mouthpart served as the reference electrode. The recording electrode, having a resistance of 512
70–150 MΩ, was carefully inserted into the dorsolateral region of the AL via a 513
micromanipulator (Leica). Neuronal spike activity was amplified (AxoClamp 1A, Axon 514
Instruments, Union, CA, USA) and monitored continuously by oscilloscope and loudspeaker. 515
Spike2 6.02 (Cambridge Electronic Design, Cambridge, England) was used as acquisition 516
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

24
software. During recording, the moth was ventilated constantly with a steady stream of fresh 517
air. During odor stimulation, a pulse of air from the continuous airstream was diverted via a 518
solenoid-activated valve (General Valve Corp., Fairfield, NJ, USA) through a glass cartridge 519
bearing the odorant on a piece of filter paper. Up to six odors were tested in each recording 520
experiment, while the number of trials were dependent on neuronal contact. The stimulation 521
period was 400 ms. After testing all odor stimuli, the neuron was iontophoretically stained by 522
applying 2–3 nA pulses with 200 ms duration at 1 Hz for about 5–10 min. In order to allow 523
neuronal transportation of the dye, the preparation was kept overnight at 4 °C. The brain was 524
then dissected from the head capsule and fixed in 4% paraformaldehyde for 1-2 h at room 525
temperature, before it was dehydrated in an ascending ethanol series (50%, 70%, 90%, 96%, 2 526
× 100%; 10 min each). Finally, the brain was cleared and mounted in methylsalicylate. 527
528
Odor stimulation 529
During intracellular recordings, the following stimuli were tested: (i) the primary sex 530
pheromone of H. armigera, Z11-16:Al, (ii) the secondary sex pheromone, Z9-16:Al, (iii) the 531
binary mixture of Z11-16:Al and Z9-16:Al, (iv) the behavioral antagonist of H. armigera, 532
Z9-14:Al, (v) the head space of a host plant (sunflower leaves), and (vi) hexane as a vehicle 533
control. The three insect-produced components were obtained from Pherobank (Wijk bij 534
Duurstede, Netherlands). Stimuli i-iv were used in two distinct stimulations protocols, i.e. 535
either with low or high concentrations of the respective stimuli. In the low concentration 536
protocol, the mixture of Z11-16:Al and Z9-16:Al were in a 95:5 proportion, while the high 537
concentration protocol used a 97:3 proportion, both resembling the natural blend emitted by 538
conspecific females (Hillier & Baker, 2016; Kehat et al., 1980; Piccardi et al., 1977; Wu et al., 539
1997). The insect-produced stimuli were diluted in hexane (99%, Sigma), and applied to a 540
filter paper that was placed inside a 120 mm glass cartridge. For the low concentration 541
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

25
protocol, the final amount per filter paper was 10 ng of the relevant stimulus, whereas the 542
high concentration protocol included 100 ng. Note that the ID of PNs stimulated with the low 543
concentration protocol were numbered (e.g. Cu-mALT1), while the PNs stimulated with the 544
high stimulation protocol were named with letters (e.g. Cu-mALTa). To avoid adaptation, 545
PNs in the high concentration protocol did not receive repeated trials, as large dosages of 546
pheromones are associated with prolonged adaptation of the relevant OSNs (Dolzer et al., 547
2003). 548
The same odor stimuli as listed above were used during the calcium imaging 549
experiment, but at a higher concentration (required to evoke a response in this technique), i.e. 550
10 µg at the filter paper. To avoid adaptation, the interstimulus interval was 1 min. An 551
additional stimulus containing a 50:50 mixture of host plant (20µl) and pheromone mix (20µl 552
95:5 mixture) was added in the calcium imaging measurements. 553
554
Confocal microscopy 555
Whole brains were imaged by using a confocal laser-scanning microscope (LSM 800 Zeiss, 556
Jena, Germany) equipped with a Plan-Neofluar 20x/0.5 objective and/or a 10x/0.45 water 557
objective (C-achroplan). Micro-ruby staining was excited with a HeNe laser at 553 nm and 558
the fluorescent emission passed through a 560 nm long-pass filter. In addition to the 559
fluorescent dyes, the auto-fluorescence of endogenous fluorophores in the neural tissue was 560
imaged to visualize relevant structures in the brain containing the stained neurons. Since 561
many fluorophore molecules were excited at 493 nm, auto-fluorescent images were obtained 562
by using the argon laser in combination with a 505-550 nm band pass filter. Serial optical 563
sections with resolution of 1024 x 1024 pixels were obtained at 2-8 µm intervals. Confocal 564
images were edited in ZEN 2 (blue edition, Carl Zeiss Microscopy GmbH, Jana, Germany) 565
and MATLAB R2018b. 566
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

26
567
Reconstruction of individual neurons and nomenclature 568
For visualization purposes, each successfully stained neuron was reconstructed in AMIRA 5.3 569
(Visualization Science Group) by using the SkeletonTree plugin (Evers et al., 2005; Schmitt et 570
al., 2004) and by this the morphology of the neuronal filaments and their thickness were 571
digitally reproduced. Further, the reconstructed neurons were manually transformed into the 572
3D representative brain, as used previously (Chu et al., 2020a; Chu et al., 2020b), for 573
illustration. Based on studies in the silk moth, B. mori, the “delta region of the inferior lateral 574
protocerebrum” (∆ILPC), a pyramid-shaped area formed by medial and mediolateral tract 575
PNs, was previously described as a main MGC-output area (e.g. Seki et al., 2005). The 576
current standard insect brain nomenclature, established by Ito et al. (2014), does not include 577
the ∆ILPC, which seems to be part of several protocerebral neuropils (Ito et al., 2014; Lee et 578
al., 2019; Lei et al., 2013). The identification and naming of neuropil structures in the 579
representative brain has been adapted from the standard nomenclature established by Ito et al. 580
(2014). The orientation of all brain structures is indicated relative to the body axis of the 581
insect, as in Homberg et al. (1988). 582
583
Calcium imaging data analysis 584
In this study, responses of medial-tract PNs innervating each MGC unit were analyzed. 585
Recordings were acquired with Live Acquisition V2.3.0.18 (TILL Photonics GmbH, 586
Kaufbeuren, Germany) and imported in KNIME Analytics Platform 2.12.2 (KNIME GmbH, 587
Konstanz, Germany). Here, ImageBee neuro plugin (Strauch et al., 2013) was used to 588
construct AL maps and glomerular time series. To determine an average baseline activity, the 589
Fura signal representing the ratio between 340 and 380 nm excitation light (F340/F380) from 0.5 590
to 2.5 s (frames 5-25, within 4 s spontaneous activity) was selected and set to zero. Neuronal 591
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

27
activity traces were thus represented as changes in fluorescent level, specified as ΔF340/F380. 592
We defined a Threshold based upon the control (hexane) traces across 8 individuals at a 5% 593
significance level:  = (
 















+ 1.96
 

). Responses 594
were defined when the mean peak of ΔF340/F380 trace (n=8) for distinct stimuli was higher 595
than Threshold. For displaying the response amplitude for each stimulus, the averaged 596
ΔF340/F380 within 2s stimulation window was used. 597
598
Spike data analysis 599
The electrophysiological data was spike-sorted and analyzed in Spike 2.8. Each odor 600
application trial comprised a total period of 2.4 s, including a 1 s pre-stimulation window for 601
baseline activity prior to the stimulus onset, 0.4 s stimulation period, and 1 s post-stimulation 602
period. For describing the temporal neural activity, the Z-scored instantaneous firing rates 603
(ZIFR) of every 10 ms for each trial were registered. To measure stimulus-specific responses 604
of individual PNs, the odor-evoked response properties were analyzed in the mean ZIFR 605
(MZIFR) across repetitive trials with the same stimulus. Significant responses were 606
determined by the upper threshold (TU) and lower threshold (TL) calculated according to the 607
mean MZIFR in the 1 s pre-stimulation window (MZIFRPS) at a 5% significance level: 608

    =













+ 1.96 609

    =













1.96 610
611
If there was an individual ZIFR in the stimulation window higher than the value of TU or 612
lower than TL, the stimulation was determined as evoking excitatory or inhibitory response, 613
respectively. 614
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

28
To illustrate the response patterns, the stimulation window was divided into two sub-615
windows (SW), the SW(i) included the first 100 ms of the stimulus application and SW(ii) 616
included the remaining 300 ms of the stimulus application. If one of the mean ZIFR during 617
the stimulation sub-windows was higher than the value of ‘Upper Threshold’ or lower than 618
‘Lower Threshold’, the change in activation was determined as an excitatory or inhibitory 619
response, respectively. 620
For displaying the response amplitude for each stimulus, we first standardized the 621
baseline activity by setting the MZIFR before stimulation onset to zero. The response 622
amplitude was then quantified as the ΔMZIFR averaged within two stimulation sub-windows, 623
respectively. The ΔMZIFR in a 200ms post-stimulation window was also computed. For each 624
trial, the onset of an excitatory response was determined at the time point when the Z-scored 625
instantaneous firing rate (binned every 1ms) exceeded the corresponding response threshold 626
(same formula as TU). The mean responses of PNs within groups categorized by the dendritic 627
or axonal innervations were also computed. 628
629
Funding 630
This project was funded by the Norwegian Research Council, Project No. 287052, to BG 631
Berg, and the National Natural Science Foundation of China, Project No. 31861133019, to 632
GR Wang. 633
634
Acknowledgments 635
We thank Stanley Heinze for his contribution to the representative brain atlas, Baiwei Ma for 636
assistance with data collection, and Tom Knudsen for technical support. 637
638
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

29
Competing interests 639
The authors declare no competing interests. 640
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

30
References 641
Anton, S., Löfstedt, C., & Hansson, B. (1997). Central nervous processing of sex pheromones 642
in two strains of the European corn borer Ostrinia nubilalis (Lepidoptera: Pyralidae). 643
Journal of Experimental Biology, 200(7), 1073-1087. 644
Baker, T. C., & Hansson, B. S. (2016). Moth Sex Pheromone Olfaction: Flux and Flexibility 645
in the Coordinated Confl uences of Visual and Olfactory Pathways. In J. D. Allison & 646
R. T. Cardé (Eds.), Pheromone Communication in Moths: Evolution, Behavior, and 647
Application. Oakland: University of California Press. 648
Berg, B. G., Almaas, T. J., Bjaalie, J. G., & Mustaparta, H. (1998). The macroglomerular 649
complex of the antennal lobe in the tobacco budworm moth Heliothis virescens: 650
specified subdivision in four compartments according to information about biologically 651
significant compounds. Journal of comparative Physiology A, 183(6), 669-682. 652
https://doi.org/10.1007/s003590050290 653
Berg, B. G., Schachtner, J., & Homberg, U. (2009). Gamma-aminobutyric acid 654
immunostaining in the antennal lobe of the moth Heliothis virescens and its 655
colocalization with neuropeptides. Cell and Tissue Research, 335(3), 593-605. 656
https://doi.org/10.1007/s00441-008-0744-z 657
Berg, B. G., Zhao, X.-C., & Wang, G. (2014). Processing of pheromone information in related 658
species of heliothine moths. Insects, 5(4), 742-761. 659
https://doi.org/10.3390/insects5040742 660
Cardé, R. T., & Willis, M. A. (2008). Navigational strategies used by insects to find distant, 661
wind-borne sources of odor. Journal of Chemical Ecology, 34(7), 854-866. 662
https://doi.org/10.1007/s10886-008-9484-5 663
Chang, H., Liu, Y., Ai, D., Jiang, X., Dong, S., & Wang, G. (2017). A pheromone antagonist 664
regulates optimal mating time in the moth Helicoverpa armigera. Current Biology, 665
27(11), 1610-1615.e1613. https://doi.org/10.1016/j.cub.2017.04.035 666
Christensen, T. A., & Hildebrand, J. G. (1987). Male-specific, sex pheromone-selective 667
projection neurons in the antennal lobes of the moth Manduca sexta. Journal of 668
Comparative Physiology A, 160(5), 553-569. https://doi.org/10.1007/BF00611929 669
Christensen, T. A., Mustaparta, H., & Hildebrand, J. G. (1991). Chemical communication in 670
heliothine moths. Journal of Comparative Physiology A, 169(3), 259-274. 671
https://doi.org/10.1007/BF00206990 672
Christensen, T. A., Mustaparta, H., & Hildebrand, J. G. (1995). Chemical communication in 673
heliothine moths. Journal of Comparative Physiology A, 177(5), 545-557. 674
https://doi.org/10.1007/BF00207184 675
Chu, X., Heinze, S., Ian, E., & Berg, B. G. (2020a). A novel major output target for 676
pheromone-sensitive projection neurons in male moths. Frontiers in cellular 677
neuroscience, 14(147). https://doi.org/10.3389/fncel.2020.00147 678
Chu, X., Kc, P., Ian, E., Kvello, P., Liu, Y., Wang, G. R., & Berg, B. G. (2020b). Neuronal 679
architecture of the second-order CO2 pathway in the brain of a noctuid moth. Scientific 680
Reports, 10(1), 19838. https://doi.org/10.1038/s41598-020-76918-1 681
Cork, A., Boo, K., Dunkelblum, E., Hall, D., Jee-Rajunga, K., Kehat, M., . . . Xun, L. (1992). 682
Female sex pheromone of oriental tobacco budworm, Helicoverpa assulta 683
(Guenee)(Lepidoptera: Noctuidae): Identification and field testing. Journal of chemical 684
ecology, 18(3), 403-418. https://doi.org/10.1007/BF00994240 685
Dolan, M.-J., Frechter, S., Bates, A. S., Dan, C., Huoviala, P., Roberts, R. J., . . . Jefferis, G. 686
S. (2019). Neurogenetic dissection of the Drosophila lateral horn reveals major outputs, 687
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

31
diverse behavioural functions, and interactions with the mushroom body. eLife, 8, 688
e43079. https://doi.org/10.7554/eLife.43079 689
Dolzer, J., Fischer, K., & Stengl, M. (2003). Adaptation in pheromone-sensitive trichoid 690
sensilla of the hawkmoth Manduca sexta. Journal of Experimental Biology, 206(9), 691
1575-1588. https://doi.org/10.1242/jeb.00302 692
Evers, J. F., Schmitt, S., Sibila, M., & Duch, C. (2005). Progress in functional neuroanatomy: 693
precise automatic geometric reconstruction of neuronal morphology from confocal 694
image stacks. Journal of Neurophysioly, 93(4), 2331-2342. 695
https://doi.org/10.1152/jn.00761.2004 696
Kuebler, L. S., Schubert, M., Kárpáti, Z., Hansson, B. S., & Olsson, S. B. (2012). Antennal 697
lobe processing correlates to moth olfactory behavior. Journal of neuroscience, 32(17), 698
5772–5782. https://doi.org/10.1523/JNEUROSCI.6225-11.2012 699
https://doi.org/10.1523/JNEUROSCI.6225-11.2012 Gothilf, S., Kehat, M., Jacobson, 700
M., & Galun, R. J. E. (1978). Sex attractants for male Heliothis armigera (Hbn.). 701
Experientia, 34(7), 853-854. https://doi.org/10.1007/BF01939662 702
Grabe, V., & Sachse, S. (2018). Fundamental principles of the olfactory code. Biosystems, 703
164, 94-101. https://doi.org/10.1016/j.biosystems.2017.10.010 704
Gupta, N., & Stopfer, M. (2012). Functional analysis of a higher olfactory center, the lateral 705
horn. Journal of Neuroscience, 32(24), 8138-8148. 706
https://doi.org/10.1523/JNEUROSCI.1066-12.2012 707
Hansson, B., Anton, S., & Christensen, T. (1994). Structure and function of antennal lobe 708
neurons in the male turnip moth, Agrotis segetum (Lepidoptera: Noctuidae). Journal of 709
comparative Physiology A, 175(5), 547-562. https://doi.org/10.1007/BF00199476 710
Hansson, B. S., Christensen, T. A., & Hildebrand, J. G. (1991). Functionally distinct 711
subdivisions of the macroglomerular complex in the antennal lobe of the male sphinx 712
moth Manduca sexta. Journal of Comparative Neurology, 312(2), 264-278. 713
https://doi.org/10.1002/cne.903120209 714
Hillier, N. K., & Baker, T. C. (2016). Pheromones of heliothine moths. In J. D. Allison & R. 715
T. Cardé (Eds.), Pheromone Communication in Moths: Evolution, Behavior, and 716
Application. Oakland: University of California Press. 717
Homberg, U., Montague, R., & Hildebrand, J. G. (1988). Anatomy of antenno-cerebral 718
pathways in the brain of the sphinx moth Manduca sexta. Cell and Tissue Research, 719
254(2), 255-281. https://doi.org/10.1007/BF00225800 720
Ian, E., Kirkerud, N. H., Galizia, C. G., & Berg, B. G. (2017). Coincidence of pheromone and 721
plant odor leads to sensory plasticity in the heliothine olfactory system. PLoS One, 722
12(5), e0175513. https://doi.org/10.1371/journal.pone.0175513 723
Ian, E., Zhao, X. C., Lande, A., & Berg, B. G. (2016). Individual neurons confined to distinct 724
antennal-lobe tracts in the heliothine moth: Morphological characteristics and global 725
projection patterns. Frontiers in neuroanatomy, 10(101), 101. 726
https://doi.org/10.3389/fnana.2016.00101 727
Ito, K., Shinomiya, K., Ito, M., Armstrong, J. D., Boyan, G., Hartenstein, V., . . . Vosshall, L. 728
B. (2014). A systematic nomenclature for the insect brain. Neuron, 81(4), 755-765. 729
https://doi.org/10.1016/j.neuron.2013.12.017 730
Jarriault, D., Gadenne, C., Rospars, J.-P., & Anton, S. (2009). Quantitative analysis of sex-731
pheromone coding in the antennal lobe of the moth Agrotis ipsilon: a tool to study 732
network plasticity. Journal of Experimental Biology, 212(8), 1191-1201. 733
https://doi.org/10.1242/jeb.024166 734
Kanzaki, R., Arbas, E. A., Strausfeld, N. J., & Hildebrand, J. G. (1989). Physiology and 735
morphology of projection neurons in the antennal lobe of the male moth Manduca sexta. 736
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

32
Journal of Comparative Physiology A, 165(4), 427-453. 737
https://doi.org/10.1007/bf00611233 738
Kanzaki, R., Soo, K., Seki, Y., & Wada, S. (2003). Projections to higher olfactory centers 739
from subdivisions of the antennal lobe macroglomerular complex of the male silkmoth. 740
Chemical Senses, 28(2), 113-130. https://doi.org/10.1093/chemse/28.2.113 741
KC, P., Chu, X., Kvello, P., Zhao, X.-C., Wang, G.-R., & Berg, B. G. (2020). Revisiting the 742
Labial Pit Organ Pathway in the Noctuid Moth, Helicoverpa armigera. Frontiers in 743
physiology, 11, 202. https://doi.org/10.3389/fphys.2020.00202 744
Kehat, M., & Dunkelblum, E. (1990). Behavioral responses of male Heliothis armigera 745
(Lepidoptera: Noctuidae) moths in a flight tunnel to combinations of components 746
identified from female sex pheromone glands. Journal of insect behavior, 3(1), 75-83. 747
https://doi.org/10.1007/BF01049196 748
Kehat, M., Gothilf, S., Dunkelblum, E., & Greenberg, S. (1980). Field evaluation of female 749
sex pheromone components of the cotton bollworm, Heliothis armigera. Entomologia 750
Experimentalis et Applicata, 27(2), 188-193. DOI: 10.1111/j.1570-7458.1980.tb02963.x 751
Kuenen, L., & Cardé, R. T. (1994). Strategies for recontacting a lost pheromone plume: 752
casting and upwind flight in the male gypsy moth. Physiological Entomology, 19(1), 15-753
29. https://doi.org/10.1111/j.1365-3032.1994.tb01069.x 754
Kymre, J. H., Berge, C. N., Chu, X., Ian, E., & Berg, B. G. (2020). Antennal-lobe neurons in 755
the moth Helicoverpa armigera – morphological features of projection neurons, local 756
interneurons, and centrifugal neurons. Journal of Comparative Neurology, in press. 757
https://doi.org/10.1002/cne.25034 758
Lee, S. G., Celestino, C. F., Stagg, J., Kleineidam, C., & Vickers, N. J. (2019). Moth 759
pheromone-selective projection neurons with cell bodies in the antennal lobe lateral 760
cluster exhibit diverse morphological and neurophysiological characteristics. Journal of 761
Comparative Neurology, 527(9), 1443-1460. https://doi.org/10.1002/cne.24611 762
Lei, H., Chiu, H.-Y., & Hildebrand, J. G. (2013). Responses of protocerebral neurons in 763
Manduca sexta to sex-pheromone mixtures. Journal of comparative Physiology A, 764
199(11), 997-1014. https://doi.org/10.1007/s00359-013-0844-4 765
Lerner, H., Rozenfeld, E., Rozenman, B., Huetteroth, W., & Parnas, M. (2020). Differential 766
Role for a Defined Lateral Horn Neuron Subset in Naïve Odor Valence in Drosophila. 767
Scientific reports, 10(1), 1-17. https://doi.org/10.1038/s41598-020-63169-3 768
Liu, W. W., & Wilson, R. I. (2013). Glutamate is an inhibitory neurotransmitter in the 769
Drosophila olfactory system. Proceedings of the National Academy of Sciences, 770
110(25), 10294-10299. https://doi.org/10.1073/pnas.1220560110 771
Malnic, B., Hirono, J., Sato, T., & Buck, L. B. (1999). Combinatorial receptor codes for 772
odors. Cell, 96(5), 713-723. https://doi.org/10.1016/S0092-8674(00)80581-4 773
Mittal, A. M., Gupta, D., Singh, A., Lin, A. C., & Gupta, N. (2020). Multiple network 774
properties overcome random connectivity to enable stereotypic sensory responses. 775
Nature Communications, 11(1), 1023. https://doi.org/10.1038/s41467-020-14836-6 776
Miyamichi, K., Amat, F., Moussavi, F., Wang, C., Wickersham, I., Wall, N. R., . . . Luo, L. 777
(2011). Cortical representations of olfactory input by trans-synaptic tracing. Nature, 778
472(7342), 191-196. https://doi.org/10.1038/nature09714 779
Namiki, S., & Kanzaki, R. (2019). Morphology and physiology of olfactory neurons in the 780
lateral protocerebrum of the silkmoth Bombyx mori. Scientific reports, 9(1), 1-14. 781
https://doi.org/10.1038/s41598-019-53318-8 782
Nirazawa, T., Fujii, T., Seki, Y., Namiki, S., Kazawa, T., Kanzaki, R., & Ishikawa, Y. (2017). 783
Morphology and physiology of antennal lobe projection neurons in the hawkmoth 784
Agrius convolvuli. Journal of insect physiology, 98, 214-222. 785
https://doi.org/10.1016/j.jinsphys.2017.01.010 786
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

33
Piccardi, P., Capizzi, A., Cassani, G., Spinelli, P., Arsura, E., & Massardo, P. (1977). A sex 787
pheromone component of the Old World bollworm Heliothis armigera. Journal of 788
Insect Physiology, 23(11), 1443-1445. https://doi.org/10.1016/0022-1910(77)90170-6 789
Sachse, S., & Galizia, C. G. (2002). Role of inhibition for temporal and spatial odor 790
representation in olfactory output neurons: A calcium imaging study. Journal of 791
Neurophysioly, 87(2), 1106-1117. https://doi.org/10.1152/jn.00325.2001 792
Sachse, S., & Galizia, C. G. (2003). The coding of odour-intensity in the honeybee antennal 793
lobe: local computation optimizes odour representation. European Journal of 794
Neuroscience, 18(8), 2119-2132. https://doi.org/10.1046/j.1460-9568.2003.02931.x 795
Sato, T., Hirono, J., Tonoike, M., & Takebayashi, M. (1994). Tuning specificities to aliphatic 796
odorants in mouse olfactory receptor neurons and their local distribution. Journal of 797
Neurophysiology, 72(6), 2980-2989. https://doi.org/10.1152/jn.1994.72.6.2980 798
Schaffer, E. S., Stettler, D. D., Kato, D., Choi, G. B., Axel, R., & Abbott, L. F. (2018). Odor 799
perception on the two sides of the brain: Consistency despite randomness. Neuron, 800
98(4), 736-742.e733. https://doi.org/10.1016/j.neuron.2018.04.004 801
Schmitt, S., Evers, J. F., Duch, C., Scholz, M., & Obermayer, K. (2004). New methods for the 802
computer-assisted 3-D reconstruction of neurons from confocal image stacks. 803
Neuroimage, 23(4), 1283-1298. https://doi.org/10.1016/j.neuroimage.2004.06.047 804
Seki, Y., Aonuma, H., & Kanzaki, R. (2005). Pheromone processing center in the 805
protocerebrum of Bombyx mori revealed by nitric oxide-induced anti-cGMP 806
immunocytochemistry. Journal of Comparative Neurology, 481(4), 340-351. 807
https://doi.org/10.1002/cne.20392 808
Skiri, H. T., Rø, H., Berg, B. G., & Mustaparta, H. (2005). Consistent organization of 809
glomeruli in the antennal lobes of related species of heliothine moths. Journal of 810
Comparative Neurology, 491(4), 367-380. https://doi.org/10.1002/cne.20692 811
Sosulski, D. L., Bloom, M. L., Cutforth, T., Axel, R., & Datta, S. R. (2011). Distinct 812
representations of olfactory information in different cortical centres. Nature, 472(7342), 813
213-216. https://doi.org/10.1038/nature09868 814
Strauch, M., Rein, J., Lutz, C., & Galizia, C. G. (2013). Signal extraction from movies of 815
honeybee brain activity: the ImageBee plugin for KNIME. BMC Bioinformatics, 816
14(Suppl 18), S4-S4. https://doi.org/10.1186/1471-2105-14-S18-S4 817
Su, C.-Y., Menuz, K., & Carlson, J. R. (2009). Olfactory perception: receptors, cells, and 818
circuits. Cell, 139(1), 45-59. https://doi.org/10.1016/j.cell.2009.09.015 819
Vickers, N. J., & Baker, T. C. (1994). Reiterative responses to single strands of odor promote 820
sustained upwind flight and odor source location by moths. Proceedings of the National 821
Academy of Sciences, 91(13), 5756-5760. https://doi.org/10.1073/pnas.91.13.5756 822
Vickers, N. J., Christensen, T. A., & Hildebrand, J. G. (1998). Combinatorial odor 823
discrimination in the brain: attractive and antagonist odor blends are represented in 824
distinct combinations of uniquely identifiable glomeruli. Journal of Comparative 825
Neurology, 400(1), 35-56. https://doi.org/10.1002/(SICI)1096-826
9861(19981012)400:1<35::AID-CNE3>3.0.CO;2-U 827
Wu, D., Yan, Y., & Cui, J. (1997). Sex pheromone components of Helicoverpa 828
armigera:chemical analysis and field tests. Insect Science, 4(4), 350-356. 829
https://doi.org/10.1111/j.1744-7917.1997.tb00109.x 830
Wu, H., Hou, C., Huang, L.-Q., Yan, F.-S., & Wang, C.-Z. (2013). Peripheral coding of sex 831
pheromone blends with reverse ratios in two Helicoverpa species. PLoS One, 8(7), 832
e70078. https://doi.org/10.1371/journal.pone.0070078 833
Wu, H., Xu, M., Hou, C., Huang, L.-Q., Dong, J.-F., & Wang, C.-Z. (2015). Specific 834
olfactory neurons and glomeruli are associated to differences in behavioral responses to 835
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

34
pheromone components between two Helicoverpa species. Frontiers in Behavioral 836
Neuroscience, 9, 206. https://doi.org/10.3389/fnbeh.2015.00206 837
Zhang, J.-P., Salcedo, C., Fang, Y.-L., Zhang, R.-J., & Zhang, Z.-N. (2012). An overlooked 838
component: (Z)-9-tetradecenal as a sex pheromone in Helicoverpa armigera. Journal of 839
Insect Physiology, 58(9), 1209-1216. https://doi.org/10.1016/j.jinsphys.2012.05.018 840
Zhao, X.-C., & Berg, B. G. (2010). Arrangement of output information from the 3 841
macroglomerular units in the heliothine moth Helicoverpa assulta: Morphological and 842
physiological features of male-specific projection neurons. Chemical Senses, 35(6), 843
511-521. https://doi.org/10.1093/chemse/bjq043 844
Zhao, X. C., Chen, Q. Y., Guo, P., Xie, G. Y., Tang, Q. B., Guo, X. R., & Berg, B. G. (2016). 845
Glomerular identification in the antennal lobe of the male moth Helicoverpa armigera. 846
Journal of Comparative Neurology, 524(15), 2993-3013. 847
https://doi.org/10.1002/cne.24003 848
Zhao, X. C., Kvello, P., Lofaldli, B. B., Lillevoll, S. C., Mustaparta, H., & Berg, B. G. (2014). 849
Representation of pheromones, interspecific signals, and plant odors in higher olfactory 850
centers; mapping physiologically identified antennal-lobe projection neurons in the 851
male heliothine moth. Frontiers in Systems Neuroscience, 8, 186. 852
https://doi.org/10.3389/fnsys.2014.00186 853
854
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

35
Figures 855
856
Figure 1. MGC neurons confined to the media tract (mALT) and their odor responses during calcium 857
imaging. (A) Illustration of the retrograde staining from calyx (Ca) labeling the MGC mALT neurons. (B) 858
Pictorial material representing calcium imaging data: top-left, raw image of an AL stained with Fura from the 859
Ca; top-right, a processed image showing a map of recognized glomeruli; down-left and down-right, Heat maps 860
of responses to the control and primary pheromone. White dash border circumvents the area of AL, red border 861
shows the MGC region. (C) Example of calcium imaging traces showing response to the primary pheromone 862
from neurons innervating distinct MGC units. The standardized traces quantify the neuronal activity of two 863
repeated stimulations with 100 ms sampling frequency. The interval between stimulations is 1 min. Gray bar, the 864
duration of the stimulus (2 s). (D) Violin plot of consistent tests across 8 individuals. (E) Response amplitudes of 865
a population of PNs innervating the same MGC units to all presented stimuli, where * indicates a significant 866
response compared with control. 867
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

36
868
Figure 2. Morphological features of PNs in the medial and mediolateral tracts - reconstruction and 869
confocal images. (A) Diagram of the brain neuropils targeted by the MGC output neurons in a dorsal view. 870
Color codes are in correspondence with all other figure panels. AL, antennal lobe; Ca, calyces; CB, central body; 871
LH, lateral horn; SIP, superior intermediate protocerebrum; SLP, superior lateral protocerebrum; VLP, 872
ventrolateral protocerebrum. (B-E) Examples of uniglomerular PNs in the mALT, originating from each of the 873
MGC units. One of two morphological types of dma mALT PNs in (D), with terminals in the Ca, anteroventral 874
LH, VLP and approaching SLP. The other dma PN type (E) is more restricted to LH and VLP. (F) A 875
multiglomerular PN in mALT with dendritic innervation in dma and three posterior complex (PCx) glomeruli. 876
(G-I) Multiglomerular MGC PNs, with dendritic branches innervating the MGC units homogeneously (G), 877
predominantly in the dma (H), or mainly in the dmp (I). (J) Two co-labeled mALT PNs, one innervating the 878
dma (blue) and another arborizing in the dmp (magenta). Both PNs sent their axon terminals to overlapping 879
regions of the LH and VLP. (K) A mlALT PN with homogeneously distributed dendrites across all MGC units. 880
(L) Another type of mlALT PN, with dendrites in all MGC units, but dense arborizations only in the cumulus. 881
Unique neuron IDs are presented. All 3D reconstructions were manually registered into the representative brain. 882
D, dorsal; L, lateral; M, medial; P, posterior. Red asterisks indicate weakly co-labeled neurons. Scale bars: 50 883
µm. 884
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

37
885
Figure 3. Co-labeled PNs in distinct ALTs. (A) Application of dye in the MGC visualized three antennal-lobe 886
projection neurons confined to the mALT, mlALT and lALT, respectively (A1). The axonal terminals of the 887
mlALT PN overlapped with the mALT PN in the VLP (red dashed lines in A2). (B) Two co-labeled cumulus 888
PNs, confined to the mALT and lALT, respectively. The mALT PN has no overlap with the lALT PN. L, lateral; 889
P, posterior. Scale bars: 50 µm. 890
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

38
891
Figure 4. Physiological properties of individual MGC PNs. Mean response amplitudes (ΔMZIFR) were 892
registered in three adjacent episodes: Sub-window SW(i) including the first 100 ms of the stimulation window, 893
sub-window SW(ii) including the remaining 300 ms of the stimulus duration; and a 200 ms post-stimulation 894
window (PW). *indicates significant response, determined according to the threshold of baseline activity of 895
individual neurons (with 95% confidence level). 896
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

39
897
Figure 5. Temporal resolution index of individual MGC PNs. This plot displays the across-trials mean 898
instantaneous firing rates (MZIFR) for each of the reported PNs, in response to the four presented female-899
produced stimuli. 900
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

40
901
Figure 6. Summary of odor representation across MGC units and protocerebral neuropils. (A) Mean 902
responses (ΔMZIFR; delta mean Z-scored instantaneous firing rate, for details see Spike data analysis in 903
Materials and methods) of MGC medial/mediolateral tract PNs in response to distinct female-produced odorants, 904
sorted according to the dendritic arborization. Data is presented in mean+se. Note that the PNs listed as 905
innervating the cumulus (Cu), dma, and dmp were uniglomerular, while the PNs in the MGC category were 906
multiglomerular. (B) Mean responses sorted by the protocerebral projections of the relevant PNs. The SIP and 907
SLP predominantly receive input regarding attraction, while the aversive signals are strongly overrepresented in 908
the anteroventral LH. The VLP receives signals regarding all tested pheromone-related stimuli. (C) Overview of 909
all PNs included in A & B. As the PNs commonly innervated more than one protocerebral neuropil, individual 910
PNs are represented in several neuropils in B. (D) Mean responses of pheromone mixture and primary 911
pheromone in individual PNs (left) and in calcium imaging tests (right). Data is presented in mean+se. PM, 912
pheromone mixture; PP, primary pheromone, SP, secondary pheromone; BA, behavioral antagonist. 913
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

41
914
Figure 7. Summary of morphological and physiological features of MGC PNs across the three main tracts. 915
(A) A graphic representation of the olfactory pathways in male moth brain. The morphological features of MGC 916
PNs is displayed on the left hemisphere. Multiglomerular medial-tract MGC PNs (grey) and uniglomerular PNs 917
arborizing in the dmp (magenta) or dma (blue) project to the calyces (Ca), anteroventral LH, and VLP, while the 918
uniglomerular PNs with dendrites in the cumulus (Cu; black) target the Ca, VLP, SLP, and posterior SIP. The 919
multiglomerular mediolateral-tract PNs (dashed green lines) target primarily the VLP, whereas the uniglomerular 920
lateral-tract PNs innervating the cumulus runs directly to the column in the anterior SIP (Chu et al., 2020a). In 921
the right hemisphere, a plethora of different PN-types arborizing in the ordinary glomeruli (OG) innervate 922
several lateral protocerebral neuropils (Ian et al., 2016; Kymre et al., 2020). (B) Projection scheme of MGC PNs 923
versus OG PNs in the higher brain regions. Three main antennal lobe tracts (ALTs) are illustrated: Solid line, 924
medial ALT (mALT); Dash line, mediolateral ALT (mlALT); Double line, lateral ALT (lALT). (C) Summary of 925
recorded and labelled PNs’ morphologies across the three main tracts, with the 10 lALT PNs being from our 926
previous study (Chu et al., 2020a). All output neurons in (C1) were used to create an overview of temporal 927
response properties in (C2), indicating mean responses (MZIFR; mean Z-scored instantaneous firing rate) to 928
female-produced odorants. Note that these PNs were sampled from the antennal lobe, while the length of axons 929
and action potential transmission rates have not been included, the actual timing of synaptic output onto the 930
protocerebral neuropils is thereby not represented. 931
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

42
Figure supplements 932
933
Figure 1 – figure supplement 1. Mean response traces of the MGC units during stimulation with all presented 934
stimuli, where * represents a statistically significant deviation from the Fura signal evoked by control. 935
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

43
936
Figure 1 – figure supplement 2. Cross-stimuli correlation plot of mean response traces of the MGC units during 937
stimulation with all presented stimuli. 938
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

44
939
940
Figure 2 – figure supplement 1. Morphology of all individually recorded PNs. Red asterisks indicate weakly 941
co-labelled neurons. Scale bars: 50µm 942
943
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

45
944
Figure 4 – figure supplement 1. The predominant response pattern of mALT PNs originating in the 945
cumulus. (A) Simplified scheme of a mALT PN innervating the cumulus. (B1) Spiking activity of one of the 946
mALT cumulus neurons during application of odor stimuli at low concentration (10 ng). (B2-B3) Mean traces of 947
Z-scored instantaneous firing rates (ZIFR) across repeated trials to the low concentration pheromone mixture 948
(B2) and the primary pheromone (B3) in 5 cumulus neurons, Cu-mALT1-5. (C1) Spiking activity of one of the 949
mALT cumulus neurons during application of odor stimuli at high concentration (100 ng). (B2-B3) Single trial 950
traces of Z-scored instantaneous firing rates (ZIFR) to the high concentration pheromone mixture (C2) and the 951
primary pheromone (C3) in 5 cumulus neurons, Cu-mALTa-e. (D) Comparison of mean ZIFR traces between 952
low and high concentration protocols. (E) Box plots of response amplitude (ΔMZIFR) of the entire 400ms 953
stimulation window in two concentration protocols. The response amplitudes across stimuli appeared more 954
pronounced in the low concentration protocol, suggesting that impaired odor discrimination may occur at high 955
odor concentrations. 956
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

46
957
Figure 4 – figure supplement 2. Heterogeneous response profiles of cumulus-innervating mALT PNs with 958
homogenous morphologies. (A-D) Spiking activity and traces of instantaneous firing rates (ZIFR) in PNs with 959
ID CU-mALT6-9 during application of odor stimuli at low concentration (10 ng). (E-F) Spiking activity and 960
traces of instantaneous firing rates (ZIFR) in PNs CU-mALTf-g during application of odor stimuli at high 961
concentration (100 ng). (G) Onset-Peak scatter plot including latency data from individual trials across the 6 962
presented PNs in (A-F). Data is presented as mean±se. The delayed peak was only evoked by the secondary 963
pheromone. CH, hexane; PM, pheromone mixture; PP, primary pheromone; SP, secondary pheromone; and BA, 964
behavioral antagonist. 965
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

47
966
Figure 4 – figure supplement 3. Heterogeneous response profiles of two types of dma-innervating mALT 967
PNs. (A-C) Physiological profile of dma-PN type with extended terminals approaching to the SLP. The spiking 968
activity and Z-scored instantaneous firing rates (ZIFR) trace showed that both PNs of this type were excited by 969
the behavioral antagonist (BA) and the secondary pheromone (SP). PN dma-mALT1 in (B) was widely tuned, 970
responding with phasic excitation to all stimuli except the control hexane (CH). PN dma-mALTa appeared to be 971
suppressed by the pheromone mixture (PM) and the primary pheromone (PP). (D-F) Physiological profile of the 972
other dma-PN type with restricted terminal regions. The spiking activity and Z-scored instantaneous firing rates 973
(ZIFR) trace of the two relevant neurons illustrated that neither of them responded to the BA at low 974
concentration (E) or high concentration (F). 975
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

48
976
Figure 4 – figure supplement 4. Dendritic morphology and response profile of the dmaPCx-mALT PN. (A-B) 977
Confocal images demonstrate that this PN innervated the dma (blue dashed circle in A), but not the cumulus or 978
dmp (black and pink dashed circles, respectively). Three posterior complex (PCx) glomeruli also received some 979
dendritic processes (B). (C) Reconstruction demonstrating the AL innervation in sagittal view). (D-E) Spiking 980
activity and Z-scored instantaneous firing rates (ZIFR) trace showed no responses that were significantly 981
different from baseline firing rates. However, the firing rate of this PN was slightly enhanced when stimulated 982
with the behavioral antagonist (BA), and moderately inhibited by the primary pheromone (PP). Scale-bars = 50 983
µm. 984
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

49
985
Figure 4 – figure supplement 5. Response profiles of mALT PNs with uniglomerular dendrites in the dmp. 986
(A) Schematic display of a mALT PN innervating the dmp. (B-E) Spike traces and standardized instantaneous 987
firing rates (ZIFR) during stimulation with high concentrations of the female-produced stimuli indicated that the 988
most prominent excitatory responses were elicited by the behavioral antagonist (BA) and/or the secondary 989
pheromone (SP). 990
991
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

50
992
993
Figure 4 – figure supplement 6. Dendritic MGC innervation and physiological features of multiglomerular 994
mALT PNs. (A) The PN MGCdmp-mALT1 had the majority of its dendrites in the dmp, along with minor 995
innervations in the dma and cumulus (Cu; A1-A2). This neuron responded with phasic excitations to both the 996
secondary pheromone (SP) and the behavioral antagonist (BA), and was inhibited by the pheromone mixture 997
(PM) and the primary pheromone (PP; A3-A4). (B) The dendrites of PN MGC-mALTa was quite evenly 998
distributed across the MGC, yet this PN was strongly excited only by the BA stimulus. (C) PN MGC-mALTb 999
also had evenly distributed dendrites, but had no strong responses to any of the presented stimuli. (D) The 1000
dendrites of PN MGCdma-mALTc innervated the dma densely, while the cumulus and dmp were sparsely 1001
innervated. This PN displayed no changes in firing rates upon odor stimulation. 1002
1003
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

51
1004
Figure 4 – figure supplement 7. Response profiles of mlALT PNs innervating the MGC. (A) Schematic 1005
display of a MGC mlALT PN innervating mainly the cumulus. (B) PN MGCCu-mlALT arborized densely in the 1006
cumulus, and had sparse dendrites in the dma, dmp, and one posterior complex glomerulus. This neuron had 1007
minor early-onset excitations followed by inhibition during stimulation with the primary pheromone (PP) and the 1008
pheromone mixture (PM). (C) Schematic display of a mlALT PN innervating the MGC. (D-E) Two 1009
multiglomerular mlALT PNs innervated all MGC units evenly, both PNs were widely tuned and displayed early-1010
onset excitatory responses to all female-produced stimuli. Most notably, the pheromone mixture (PM) and 1011
primary pheromone (PP) elicited responses lasting throughout the entire 400 ms stimulation. In addition to the 1012
pheromone responses, MGC-mlALT2 was phasically excited by the control stimulus hexane (CH). 1013
1014
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

52
1015
Figure 5 – figure supplement 1. Z-scored instantaneous firing rate for each trial in cumulus PNs (Cu-mALT1-1016
5). 1017
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

53
1018
Figure 5 – figure supplement 2. Z-scored instantaneous firing rate for each trial in cumulus PNs (Cu-mALT6-1019
9). 1020
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

54
1021
Figure 5 – figure supplement 3. Z-scored instantaneous firing rate for each trial in dma PNs (dma-mALT1-2) 1022
and three multiglomerular MGC PNs. 1023
1024
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint

55
SUPPLEMENTARY TABLE S1 Overview of individual projection neuron morphology
Type
ID
N
Soma
AL innervations
Protocerebral innervations
Figure
mALT
Pm_a
Cu-mALT1
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
Cu-mALT2
1
MC
Cu
Ca, VLP, SLP, SIP
Fig. 2B
Pm_a
Cu-mALT3
1
MC
Cu
Ca, VLP, SLP §
Pm_a
Cu-mALT4
2
MC
Cu
Ca, VLP §
Pm_a
Cu-mALT5
1
MC
Cu
Ca, VLP, SLP
Pm_a
Cu-mALT6
1
MC
Cu
Ca, VLP, SLP §
Pm_a
Cu-mALT7
1
MC
Cu
Ca, VLP, SLP
Pm_a
Cu-mALT8
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
Cu-mALT9
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
Cu-mALTa
1
MC
Cu
Ca, VLP, SLP §
Pm_a
Cu-mALTb
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
Cu-mALTc
1
MC
Cu
Ca, VLP, SLP
Pm_a
Cu-mALTd
1
MC
Cu
Ca, VLP, LH, SLP
Pm_a
Cu-mALTe
1
MC
Cu
Ca, VLP, SLP §
Pm_a
Cu-mALTf
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
Cu-mALTg
1
MC
Cu
Ca, VLP, SLP, SIP
Pm_a
dma-mALT1
1
MC
dm-a
Ca, VLP
Pm_a
dma-mALT2
3
MC
dm-a
Ca, LH, VLP
Pm_a
dma-mALTa
1
MC
dm-a
Ca, LH, VLP
Fig. 2D
Pm_a
dma-mALTb
1
MC
dm-a
Ca, LH, VLP
Fig. 2E
Pm_a
dmaPCx-mALTa
1
MC
dm-a, PCx Gs
Ca, LH, VLP
Fig. 2F
Pm_a
dmp-mALTa
1
MC
dm-p
Ca, LH, VLP
Pm_a
dmp-mALTb
1
MC
dm-p
Ca, LH, VLP
Pm_a
dmp-mALTc
1
MC
dm-p
Ca, LH, VLP
Fig. 2C
Pm_a
dmp-mALTd
1
MC
dm-p
Ca, LH, VLP
Pm_a
MGC
dmp
-mALT1
1
MC
Cu, dm-p, dm-a
Ca, LH, VLP
Fig. 2I
Pm_a
MGC-mALTa
1
MC
Cu, dm-p, dm-a
Ca, LH, VLP
Fig. 2G
Pm_a
MGC-mALTb
1
MC
Cu, dm-p, dm-a
Ca, LH, VLP
Pm_a
MGC
dma
-mALTc
1
MC
Cu, dm-a, dm-p
Ca, VLP
Fig. 2H
mlALT
Pml_a
MGC-mlALT1
1
LC
Cu, dm-a, dm-p
VLP
Fig. 2K
Pml_a
MGC-mlALT2
1
LC
Cu, dm-a, dm-p
VLP, PLP
Pml_b
MGC
Cu
-mlALTa
1
LC
Cu, dm-p, dm-a, PCx G
VLP, SLP, SIP
Fig. 2L
1025
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.11.421289doi: bioRxiv preprint