1

Structured odorant response patterns across a complete olfactory 1

receptor neuron population 2

3

Guangwei Si1, 2, 4, Jessleen K. Kanwal2, 3, 4, Yu Hu2, Christopher J. Tabone1, 2, Jacob Baron1, 2, 4

Matthew Berck1, 2, Gaetan Vignoud1, 2, Aravinthan D.T. Samuel1, 2, 5, * 5

6

1Department of Physics, Harvard University, Cambridge, MA 02138, USA 7

2Center for Brain Science, Harvard University, Cambridge, MA 02138, USA 8

3Program in Neuroscience, Harvard University, Cambridge, MA 02138, USA 9

4These authors contributed equally 10

5Lead Contact 11

*Correspondence: samuel@physics.harvard.edu 12

2

Graphical Abstract 13

14

15

Highlights 16

• All ORNs share a common dose-response function with variable sensitivity across odors 17

• Sensitivities across odorants and ORNs follow a power law distribution 18

• Correlation in sensitivities corresponds to a geometric molecular property 19

• ORN-odorant responses share similar temporal filters 20

21

In Brief 22

The combinatorial olfactory code conveys odor stimulus features in an entangled manner. Si et 23

al. uncover structure and statistical properties in ORN responses, allowing separated 24

representations of odor features and shining light on the molecular recognition mechanism by 25

olfactory receptors. 26

3

Summary 27

Odor perception allows animals to distinguish odors, recognize the same odor across 28

concentrations, and determine concentration changes. How the activity patterns of 29

primary olfactory receptor neurons (ORNs), at the individual and population levels, 30

facilitate distinguishing these functions remains poorly understood. Here, we interrogate 31

the complete ORN population of the Drosophila larva across a broadly sampled panel of 32

odorants at varying concentrations. We find that the activity of each ORN scales with the 33

concentration of any odorant via a fixed dose-response function with a variable 34

sensitivity. Sensitivities across odorants and ORNs follow a power law distribution. 35

Much of receptor sensitivity to odorants is accounted for by a single geometrical 36

property of molecular structure. Similarity in the shape of temporal response filters 37

across odorants and ORNs extend these relationships to fluctuating environments. 38

These results uncover shared individual and population level patterns that together lend 39

structure to support odor perceptions. 40

41

4

Introduction 42

The ability to identify odorants across a wide range of concentrations and detect changes in 43

odorant concentration are essential for olfactory perception and behavior. How olfactory 44

representations are organized to support these distinct functions is not yet fully understood. In 45

mammalian and insect olfactory systems, combinatorial receptor codes allow a limited number 46

of olfactory receptor neurons (ORNs) to encode a large number of odorants (Malnic et al., 47

1999). Each ORN typically expresses one olfactory receptor (Or) type (Buck and Axel, 1991), 48

but a single Or can be activated by many odorants and a single odorant can activate many Ors 49

(Friedrich and Korsching, 1997). Different odorants can be discriminated by distinct activity 50

patterns across a population of olfactory neurons (Hallem and Carlson, 2006; Kreher et al., 51

2008; Nara et al., 2011). The olfactory receptor code also conveys information about odorant 52

intensity, as odorants at higher concentrations typically activate more ORNs (Kajiya et al., 2001; 53

Wang et al., 2003). Odorants may also evoke different temporal response patterns in ORNs, 54

augmenting information coding using time (De Bruyne et al., 2001; Friedrich and Laurent, 2001; 55

Grillet et al., 2016; Junek et al., 2010; Raman et al., 2010). 56

57

Behavioral experiments in mammals and insects indicate that animals can distinguish and learn 58

differences between odor identities and concentrations (Apostolopoulou et al., 2013; Chen et 59

al., 2011; Mishra et al., 2013; Pelz et al., 1997; Uchida, 2008; Wang, 2004). Thus, olfactory 60

perception likely requires the ability to independently represent both odor identity and intensity. 61

Distinct and invariant representations of odor identity and intensity appear in neurons in central 62

olfactory processing regions (Bolding and Franks, 2017; Roland et al., 2017; Sachse and 63

Galizia, 2003; Stopfer et al., 2003; Wang, 2004). However, ORN activity patterns always reflect 64

both odor identity and intensity in a seemingly intertwined manner. Increasing the concentration 65

of any odor typically recruits more ORNs, altering the combinatorial activity pattern in a complex 66

manner. Two previously described properties of ORN responses may help disentangle odor 67

identity and intensity. First, the relative activity of different ORNs largely persists across 68

concentrations of an odor (Cleland et al., 2007; Wachowiak et al., 2002). Second, normalized 69

ORN responses exhibit a stereotyped distribution across odors. The mean population activity, 70

used to normalize ORN responses, scales on average with odor concentration (Stevens, 2016). 71

However, these studies lack the full dynamic range of individual ORN responses, leaving it 72

unclear what and how properties of individual ORNs give rise to the emergent properties 73

characteristic of the population. To assess the underlying mechanisms and further develop a 74

statistical description of population responses requires a comprehensive analysis of ORN 75

5

population activity, with single cell resolution, over a broad stimulus space that spans many 76

odorants across varying intensities. 77

78

Here, we asked whether a complete olfactory receptor population has cellular and systems level 79

properties that structure the responses to help disentangle odorant type and concentration 80

dependent variations. We hypothesized that such properties might be reflected in functional 81

relationships between individual and complete ORN responses to a broad set of odorant 82

concentrations and types. We leveraged the experimental accessibility of the olfactory system of 83

the Drosophila larva to search for quantitative structure in the responses of a complete primary 84

olfactory population as well as in the sensitivity and dynamics of individual ORNs. The 85

numerical simplicity of Drosophila olfactory neurons facilitates such a system-level dissection of 86

a full set of ORNs. Furthermore, the larval ORNs form the first layer of an olfactory circuit that 87

also shares glomerular organization with adult insects and vertebrates (Ramaekers et al., 2005; 88

Su et al., 2009; Vosshall and Stocker, 2007). 89

90

To perform our study, we developed an in vivo imaging setup with microfluidics to 91

simultaneously deliver highly controlled stimuli from a panel of 34 odorants while monitoring the 92

responses of all ORNs with cellular resolution. Our odorant panel elicits activity in all 21 ORNs, 93

allowing us to characterize the functional structure of the entire population. We find that all 94

ORN-odorant pairs share the same activation or dose-response function: ORN activity 95

increases with odorant concentration along the same Hill curve for any odorant, but with 96

different sensitivities. Thus, the principal free variable that characterizes the interaction between 97

any odorant molecule and any receptor is the sensitivity. We find that the statistical distribution 98

of these sensitivities follows a power-law. The consequence of this power law is that the relative 99

change of overall ORN activity becomes proportional to the relative change of odorant 100

concentration, potentially simplifying the problem of how downstream neurons process 101

information about odor intensity. In addition, the primary principal component of ORN 102

sensitivities is correlated with a molecular property that describes the geometric and 103

electrotopological states of odorants (Haddad et al., 2008). Finally, ORNs share a stereotyped 104

temporal filter shape such that the observed similarities in odorant response patterns may also 105

extend to fluctuating environments, underscoring the computational significance of these shared 106

features. These structured ORN response patterns may constitute a simple strategy to 107

represent odors. 108

6

Results 109

A microfluidic setup for in vivo calcium imaging of larval ORNs 110

To record from a population of olfactory neurons with single cell resolution, we developed a 111

microfluidics setup for delivering a large range of odorant inputs while simultaneously imaging 112

neural activity. Small size and optical transparency make the larva’s olfactory system – like that 113

of C. elegans – suitable for in vivo multineuronal calcium imaging with microfluidic control of 114

olfactory inputs (Chronis et al., 2007). Furthermore, fluid delivery of odorants allows for precise 115

control of odorant concentration, stimulus waveform, and timing between stimulus delivery 116

compared to gaseous odorant delivery (Andersson et al., 2012). Our microfluidic device allows 117

for recording from an intact, immobilized, and un-anesthetized larva with as many as 24 fluid 118

delivery channels (Figures 1A-1D and S1A-S1D). Calcium imaging and genetic labeling allow 119

us to record the activity of any individual ORN alone or the activity of all ORNs simultaneously, 120

by expressing the calcium indicator GCaMP6m (Chen et al., 2013) under the control of either a 121

specific ORN Gal4 driver or the Orco-Gal4 driver, respectively (Vosshall et al., 1999). We use 122

this microfluidic setup to perform single cell and population level recordings of olfactory 123

processing in single animals exposed to inputs spanning a broad range of odorant types and 124

concentrations. 125

126

Single ORN identification from population recordings 127

We sought to efficiently record ORN population responses to many odorant stimuli in the same 128

animal with single cell identification. To do this, we developed a method to unambiguously 129

identify all 21 ORNs during population recordings. First, we assessed the stereotypy of larval 130

ORN anatomical organization. We found that the layout of ORN dendrites aids in segmenting 131

and identifying cells during calcium imaging. The larva has 21 ORNs located in each bilaterally 132

symmetric dorsal organ ganglion (DOG). The 21 ORN dendrites are organized into seven 133

parallel bundles, each containing three dendrites, that project from an ORN cell body to a 134

perforated dome structure on the animal’s head called the dorsal organ (Singh and Singh, 135

1984). When a larva is immobilized in the microfluidic device, four ventral and three dorsal 136

dendritic bundles are easily distinguished (Figure 1E). We mapped individual ORNs to each 137

bundle by expressing RFP in all ORNs and GFP in a selected ORN using a cell-specific Gal4 138

driver (Fishilevich et al., 2005; Kreher et al., 2005) ( Figure S2). We found that the three ORN 139

dendrites located in each bundle were stereotyped (confirmed in n ≥ 5 animals for each ORN). 140

Thus, by following the activation of any cell body in the DOG to its corresponding dendritic 141

bundle, we narrowed its possible identity to one of three ORNs. 142

7

143

To complete the identification of individual ORNs, we used a set of odorants that activate single 144

ORNs of known identity at low concentrations (Mathew et al., 2013) (see Methods). We 145

delivered these odorants to larvae expressing GCaMP6m in all ORNs and found that 15 of 146

these odorants are sufficient to identify each ORN when examined in conjunction with dendritic 147

bundle location (Figure 1F). Together, the anatomical map and functional responses to this 148

subset of odorants provides a comprehensive means of identifying and segmenting the ORNs 149

responsive to any olfactory input during multi-neuronal calcium imaging. 150

151

Orthogonality of odorant identity and intensity in ORN population activity 152

We next searched for general patterns in population level olfactory responses that might 153

emerge from a sufficiently broad analysis of odorant types and intensities. We assembled a 154

panel of 34 odorants that broadly samples the full olfactory sensitivity of larval ORNs (see 155

Methods and Figure S3A). This panel primarily contains odorants that are components of fruits 156

or plant leaves, which make up the larva’s natural environment (Dweck et al., 2018), and have 157

been shown to elicit innate attractive or aversive behavior in the larva (Table S1). Furthermore, 158

these odorants have molecular structures that span a variety of functional groups, such as 159

esters, alcohols, aromatics, aldehydes, ketones, pyrazines, thiazoles, and phenyl groups. To 160

characterize the ORN representation of these stimuli, we simultaneously recorded from all 161

ORNs while exposing larvae to all 34 odorants across the concentration range of olfactory 162

sensitivity. Using our anatomical and functional ORN identity mapping method in addition to 163

intensity based registration and segmentation methods (Pnevmatikakis et al., 2016; Thévenaz 164

et al., 1998), we quantified each ORN’s peak response (Figures S3B). The peak response was 165

defined as the highest ORN activity during odorant delivery. We found that all 21 ORNs in each 166

DOG were responsive to at least one odorant in the panel. 167

168

We measured the response amplitude of every ORN to step stimuli across five orders of 169

magnitude in concentration, from 10-8 dilution (where all odorants were at or below the threshold 170

of ORN detection) to 10-4 dilution (where many ORNs reached saturation). We used five second 171

step pulses interleaved with 20-60 seconds of water, a protocol that allowed for us to measure 172

peak responses and for full recovery of neural activity (Figure S1E-S1F). We measured ORN 173

activity using the calcium indicator GCaMP6m, a sensitive reporter of neural excitation. Previous 174

studies have reported that some odor molecules can inhibit some ORNs in Drosophila and other 175

insects (Cao et al., 2017; Hallem and Carlson, 2006; Tichy et al., 2005). We did not observe 176

8

odorant induced inhibition across ORNs, although the low background intensity of GCaMP6m 177

makes it much more sensitive to excitatory than inhibitory responses. We note that calcium 178

imaging also limits our temporal resolution to that of the GCaMP6m indicator (Chen et al., 179

2013). 180

181

We found that ORNs that are most sensitive to particular odorant are also more sensitive to 182

molecules with similar chemical structure. For example, the long chain alcohol 1-pentanol 183

slightly evoked activity specifically in the Or35a-ORN at a 10-7 dilution. Higher concentrations of 184

1-pentanol gradually saturated the Or35a-ORN, while also activating four other ORNs 185

expressing either Or67b, Or85c, O13a, or Or1a (Figure 1G and Video S1). The additional 186

ORNs recruited by 1-pentanol were also activated at low concentrations by other long chain 187

alcohol odorants (Mathew et al., 2013). We next examined the population-wide dose-response 188

curves for these additional alcohol odorants. Low concentrations of each alcohol specifically 189

activated a distinct ORN. Higher concentrations reliably activated the Or35a, Or13a, Or67b, and 190

Or85c ORNs to varying degrees (Figure S3C). We then collected dose-dependent responses 191

across the entire ORN population for all 34 odorants, with at least five animals per odorant 192

(Figure 2A). Specific activation of single ORNs at low concentrations (10-7 and 10-6 dilutions) is 193

in agreement with previous reports (Mathew et al., 2013). We found a similar pattern of 194

overlapping activation for ORNs that were selectively responsive to odorants with similar 195

molecular structures at low concentrations. 196

197

As in other animals, the combinatorial olfactory activity pattern changes with increasing odorant 198

intensity (Malnic et al., 1999), and with a pattern of ORN recruitment that is correlated with 199

molecular selectivity. To discern this pattern, we performed principal component analysis (PCA) 200

on the ORN population activity responses to all 34 odorants across all five concentrations (i.e., 201

PCA across 170 odorant-concentration pairs). We projected ORN activity responses in the 202

space of the first three principal components, which accounts for 60% of the variance in the data 203

(Figure 2B). At the lowest concentrations, olfactory representations at or below detection 204

thresholds were tightly clustered at a central point in the PCA space. At higher concentrations, 205

olfactory representations diverged, increasing distance monotonically from the central point. 206

Interestingly, the trajectory of each odorant tended to follow its own direction in PCA space. This 207

pattern is particularly clear for aliphatic and aromatic odorants. Aliphatic odorants with long 208

carbon chains form trajectories projecting in a similar direction of PCA space, since higher 209

concentrations of these odorants tend to selectively recruit ORNs that are also sensitive to 210

9

aliphatic odorants. The same was true for aromatic odorants and their corresponding group of 211

sensitive ORNs. Clustering of ORNs as primarily responsive to aliphatic or aromatic odorant 212

types agrees with previous Drosophila ORN electrophysiology recordings using large panels of 213

odorant and natural odor stimuli (Dweck et al., 2018; Kreher et al., 2008). The trajectories 214

corresponding to structurally similar molecules are separated by small angles (Figure 2B). 215

Visualization of ORN responses in PCA space reveals structure in the population representation 216

of odorant identity over a large range of intensities. The population wide response maintains a 217

fixed direction in the representation of each odorant with rising concentration. This property also 218

holds true for temporal responses of the ORN population over the course of stimulus delivery 219

(see Methods and Figure S3D). 220

221

A Hill function with variable sensitivity describes dose-response relationships 222

We uncovered shared structure in each ORN’s activation function when we compared all 223

odorant-ORN pairs that reached saturation (n=36 pairs). We found that the dose-response 224

curves were well described by a Hill function: 225

𝑦 = 𝐴𝑚𝑎𝑥𝑐𝑛

𝑐𝑛+𝐸𝐶50𝑛 , 226

where 𝐴𝑚𝑎𝑥 is the maximum response amplitude measured by the calcium indicator, 𝑐 is the 227

odorant concentration, 𝑛 is the Hill coefficient or steepness of the linear portion of the curve, and 228

𝐸𝐶50 is the half-maximal effective concentration. The Hill function canonically describes binding 229

affinities in ligand-receptor interactions such as that between odorants and olfactory receptors. 230

We performed a population fit on the 36 saturated dose-response curves to the Hill function and 231

find that all curves were well fit by a shared Hill coefficient with 𝑅2 = 0.99 (see Methods, 232

Figure S4A). Each recorded neuron reaches a similar saturated response amplitude (𝐴𝑚𝑎𝑥) 233

across different molecules within each experiment (Figure S4B-S4C). The amplitude 234

normalized dose-response curves, aligned by the 𝐸𝐶50, collapse onto a single Hill function with 235

𝑛 = 1.42 (Figure 3A). 236

237

We further sought to determine the 𝐸𝐶50 values for unsaturated odorant-ORN pairs. To this end, 238

we used a maximum-likelihood based method (see Methods) to estimate the mean 𝐸𝐶50 value 239

for each odorant-ORN pair. Our only constraint on the estimation was that the amplitude and Hill 240

coefficients had shared mean values across all measurements, an assumption supported by the 241

36 activity curves that reached saturation. Using this method, we were able to extract 𝐸𝐶50 242

values for the majority of odorant-ORN pairs. 𝐸𝐶50 values spanned several orders of magnitude. 243

10

We assembled a matrix of odorant-ORN sensitivity (the inverse of 𝐸𝐶50), which is most easily 244

visualized on a logarithmic scale (Figure 3B). This sensitivity matrix, combined with the 245

activation functions with shared Hill coefficients, was able to account for 99% of the variance in 246

the entire dataset of odorant-ORN interactions across all concentrations (Figure S4D). 247

248

We conclude that a common Hill function, with the sensitivity as the principal free parameter, 249

describes the dose-response relationship for any odorant-ORN interaction. For each odorant, 250

the vector of sensitivity (a row in the matrix in Figure 3B) specifies the identity and threshold of 251

each activated ORN with increasing odorant concentration. A corollary of having a unique 252

sensitivity vector for each odorant is having a unique direction for the trajectory of population 253

responses in principal component space across concentrations (Figure 2B). 254

255

ORN population sensitivities follow a power law distribution 256

Next, we examined the probability distribution of ORN sensitivities. For each odorant, 257

responsive ORNs were distributed along a sensitivity axis, with most ORNs in the low sensitivity 258

region (Figure 3C). The density of ORNs diminishes with increasing sensitivities, generating a 259

heavy-tailed probability distribution. To quantify this distribution, we constructed a cumulative 260

density function of ORN sensitivities that is well fit by a line in a log-log plot, indicative of a 261

power law (Figure 3D). The probability density function is described by: 𝑃(𝑥) ∝ 𝑥−𝜆−1 , 𝜆 = 0.42 262

for 𝑥 > 4.2 × 104 (methods from (Clauset et al., 2009)). We compared the fitting of the power 263

law with other heavy-tailed distributions and concluded that the power law is the simplest form 264

with a good fit to the probability density of odorant-ORN sensitivities (see Methods and Table 265

S2). 266

267

A power law distribution of ORN sensitivities means that a relative change of concentration of 268

any odorant will trigger, on average, the same relative change in the number of activated ORNs. 269

The power law distribution of ORN sensitivities, together with a common Hill function, should 270

give rise to population-wide activity that follows a power law relationship with respect to 271

concentration, with exponent 𝜆 (See Methods). We confirmed this prediction in our 272

experimental data (Figure 3E). The mean activity of the ORN population grows with odorant 273

concentration following a power law with an exponent of 0.32±0.06, which is close to the 274

exponent found from fitting the sensitivity distribution (Figure 3D). The power function in a log 275

scale is a linear relationship, such that log(𝐴) ∝ log(𝑐), where 𝐴 is ORN population activity and 𝑐 276

is odorant concentration. Thus, on average, the relative change of the ORN population activity is 277

11

proportional to the relative change of odorant concentration, 𝑑(log(𝐴)) ∝𝑑(log(𝑐)) = 𝑑𝐴

𝐴∝𝑑𝑐

𝑐 (as 278

shown in Figure 3E). 279

280

Correlations in ORN sensitivities correspond to molecular structure 281

We used principal component analysis to study the structure of the sensitivity matrix (see 282

Methods). We found that the first principal component in odorant space of the logarithmically 283

scaled matrix explains a significant portion of the variance compared to shuffled data (Figure 284

4A). The eigenvector of the 1st principal component indicates the relative weights of different 285

ORNs along its axis. As shown in Figure 4B, neurons that prefer long chain alcohols (e.g., 286

Or35a, Or13a, and Or85c) and neurons that prefer aromatic odorants (e.g., Or45b, Or59a, and 287

Or24a) are at opposite extremes. Arranging all 34 odorants by their projection on the 1st 288

principal component, a clear trend based on molecular structure emerges, progressing from 289

long carbon chains on one end to aromatic molecules on the other (Figure 4C). 290

291

To further examine the correlation with molecular structure, we considered an extensive list of 292

molecular descriptors that were found to be relevant to odor discrimination across animal 293

species (Haddad et al., 2008). We found that one of these molecular descriptors, “P1s”, has the 294

highest correlation to the first principal component (Figure 4D, correlation coefficient(r) = -0.8). 295

“P1s” is a geometric descriptor of molecular structure weighted by atomic electrotopological 296

state. 1D long-chain molecules have large P1s values and 2D ring-like molecules have small 297

values (see examples in Figure 4D) (Todeschini and Lasagni, 1994). The dominance of the 298

“P1s” molecular descriptor is in agreement with the trajectories of aromatic and aliphatic odorant 299

representations pointing in opposite directions in Figure 2B. 300

301

ORN-odorant responses share similar temporal characteristics 302

An additional challenge to olfactory coding of a wide variety of odorant types across 303

concentrations arises from complex temporal dynamics due to physical fluctuations, such as 304

turbulence or convection, in the stimulus itself. To examine how fluctuations affect ORN 305

responses, we compared the conversion of temporal patterns of olfactory input for different 306

odorant-ORN pairs across odorant intensities (Figure 5). To do this, we used reverse-307

correlation analysis, subjecting larvae to “white noise” olfactory input by stochastically switching 308

between odorant and water delivery and seeking the temporal filter that best maps olfactory 309

inputs into calcium dynamics (Figure 5A) (Geffen et al., 2009; Kato et al., 2014). We found that 310

random olfactory input could evoke fluctuating calcium activity in an ORN, and repeated 311

12

presentation of the same input pattern would evoke consistent normalized responses from 312

animal to animal (Figure S5A). The systematic conversion of the stimulus to response 313

waveform is well characterized by a linear-nonlinear (LN) model. A linear transfer function 314

estimates the relative weight of each time point in stimulus history to determine the time-varying 315

response amplitude (Figure 5B). The convolution of the linear transfer function with stimulus 316

history is then passed through a static nonlinearity to correct for saturation (Figure S5B). We 317

verified the LN model by predicting the response to a novel random input using a filter 318

calculated from different random inputs (Figure S5C). To make comparisons across odorants, 319

concentrations, and ORNs we used normalized response amplitudes that preserve temporal 320

characteristics encoded in the filter shape. 321

322

We measured the linear transfer function for 3-octanol as it has a relatively broad recruitment of 323

ORNs across the concentration range studied. At the lowest concentrations of 3-octanol, a filter 324

describing ORN activity only emerges for the Or85c-ORN (Figure 5C). At higher concentrations, 325

filters begin to emerge for additional ORNs. These filters for each ORN, when normalized for 326

response amplitude, were virtually identical in their temporal response profiles as single lobed 327

functions with similar peak and decay times (Figures 5D-5E and S5D). The shapes of the filters 328

for different odorants activating the same ORN are also virtually indistinguishable, on the order 329

of ~100 ms (Figure 5F-5G). Thus, the rank order of the ORN responses could be preserved in 330

an environment with fluctuating odor intensities. 331

13

Discussion 332

Olfactory receptor neurons have diverse tuning properties to odorant molecules. This diversity 333

forms a combinatorial receptor code that represents a broad range of odorant molecules across 334

concentrations. Changes in concentration lead to changes in the combinatorial pattern of 335

activated ORNs, potentially complicating the ability to separate differences in odor identity and 336

odor intensity. Here, we asked whether olfactory receptor neurons, at the individual and 337

population levels, display structured response patterns that might facilitate in separating such 338

differences. Previous efforts at characterizing Drosophila ORNs necessarily focused analysis on 339

particular ORN types, odorants or odorant concentrations (Asahina et al., 2009; Hallem and 340

Carlson, 2006; Martelli et al., 2013; Mathew et al., 2013; Nagel and Wilson, 2011). 341

Electrophysiological studies of larval olfactory receptors, expressed in “empty olfactory neurons” 342

of the adult fly, revealed the tuning of individual receptors to large numbers of odorant 343

molecules (Kreher et al., 2008; Mathew et al., 2013). Calcium imaging of larval ORNs directly 344

revealed how subsets of larval ORNs are activated by selected odorants and concentrations 345

(Asahina et al., 2009). This study bridges these pioneering efforts with an analysis of 346

population-wide olfactory responses with single cell resolution across a broad olfactory space. 347

The small size of the Drosophila larva, combined with multi-neuronal imaging and new 348

microfluidic tools, has allowed us to characterize the responses of a complete ORN population 349

to a panel of odorant types and concentrations that fully spans the sensitivity of larval ORNs. 350

351

Our broad characterization has uncovered features in the response patterns of individual ORNs 352

and of the ORN population that are shared across different ORNs and across odorants. At the 353

level of individual neurons, each ORN response to odorants exhibits the same activation 354

function shape with varying sensitivity levels. At the level of the population, the relative change 355

of ORN activity across all tested odorants is proportional to the relative change of odorant 356

concentration, a consequence of a power law that describes the sensitivity distribution of ORNs. 357

Furthermore, a common temporal filter shape converts different stimulus waveforms into ORN 358

calcium activity patterns. Here, we discuss the functional implications of such shared structure 359

in ORN responses, and we connect our findings to computations thought to occur in the early 360

olfactory system. 361

362

Common activation function across ORN-odorant pairs 363

We confirmed that a shared activation function across ORNs and odorants is consistent with 364

available electrophysiological recordings of Drosophila ORNs, which also reveal activation 365

14

functions across concentrations that can be fit to Hill curves with shared coefficients (see 366

Methods and Figure S4H-S4I)(Kreher et al., 2008). The exact value of the Hill coefficient for 367

activation functions corresponding to spike rates (~0.7) and calcium dynamics (~1.4) are 368

different, likely reflecting the nonlinear transformation between electrical activity and calcium 369

activity across ORNs. Common activation functions have also been observed for subsets of 370

ORNs responding to molecularly similar odorants in the cockroach antenna and the rat olfactory 371

bulb (Meister and Bonhoeffer, 2001; Sass, 1976). 372

373

The underlying molecular basis for shared activation functions across receptor types may be 374

similar stoichiometries in odorant/receptor binding across Ors and/or similar internal signaling 375

pathways for spike generation across ORNs. Nagel and Wilson, 2011 found that dynamics of 376

spike generation were highly stereotyped from ORN to ORN. Common signaling pathways 377

would be consistent with the observation that expressing an Or in different ORNs or the 378

endogenous ORN lead to similar tuning properties and background firing rates (Hallem et al., 379

2004). Such a property could be analogous to that of photoreceptors in the visual system. Red 380

and green cones share a similar shape in their response profiles across wavelengths. The only 381

difference in their stimulus evoked activity patterns is their spectral sensitivity (Naka and 382

Rushton, 1966). 383

384

A common activation function across odorant and receptor types is one component of 385

correlations in ORN response patterns across olfactory space. Shared activation functions and 386

the sensitivity distribution across ORNs likely allow the vector representation of any odorant to 387

maintain a similar direction across concentrations. An uniform intraglomerular transfer function 388

from ORNs to second order projection neurons (PNs) has been described in the adult 389

Drosophila antennal lobe (Olsen et al., 2010). Together, these common functions allow the 390

olfactory system to maintain distinct representations of different odorants that are stable across 391

concentrations as has been demonstrated by (Sachse and Galizia, 2003; Stopfer et al., 2003). 392

Our work suggests that an intensity invariant aspect of odor representation – the direction of the 393

activity vector in principal component space – already emerges at the ORN layer, and this 394

property may be carried forward to the PN layer. 395

396

Power law distribution in olfactory sensitivities across ORNs 397

Different neurons are required to sense odorants in different regimes of odorant concentration 398

needed for long-range chemotaxis, in the Drosophila larva (Asahina et al., 2009). Encoding a 399

15

broad concentration range requires a distribution of ORNs with varying sensitivities. We find that 400

these olfactory sensitivities are drawn from a power law distribution. 401

402

To our knowledge, a power law distribution of olfactory sensitivities has not yet been described 403

in any animal. One possibility for the power law in olfactory sensitivity is to match the 404

distributions of odorant concentrations found in natural olfactory environments. Natural odors 405

are mixed by convection and turbulence, physical processes that are rich in power law 406

dynamics (Celani et al., 2014; Murlis, 1992; Riffell et al., 2008). Power laws appear in the 407

statistics of other natural stimuli as well. Natural visual scenes exhibit a power law relationship 408

between spectral power and spatial frequency (Field, 1987; Simoncelli and Olshausen, 2001). 409

The loudness of natural sounds across frequencies are distributed by power laws (Theunissen 410

and Elie, 2014). Sensory systems, in general, may adapt the statistical distribution of their 411

sensitivities to their natural environments. 412

413

The natural environment may drive the selection of a molecular recognition mechanism for 414

olfactory receptors that gives rise to a power law distribution in ORN sensitivities. Lancet et al., 415

1993 proposed a molecular recognition system in which a receptor has multiple selective 416

binding subsites. Each binding subsite contributes in a combinatorial manner to the binding 417

strength between a receptor and molecule. This simple probabilistic model generates a power 418

law sensitivity distribution for receptors with random sets of binding subsites. The statistics of an 419

olfactory code using such a molecular recognition system would be preserved with expansion of 420

the ORN periphery, as occurs with Drosophila in which the adult has nearly triple the number of 421

receptor types as that found in the larva. 422

423

A power law distribution implies a fixed ratio between the relative change in ORN population 424

activity for a relative change in odorant concentration. Detection of relative change in stimulus 425

intensities has been observed in psychophysical studies of diverse sensory modalities. A 426

notable example is Stevens’s Law in human psychophysics, where perceived response 427

magnitudes have been shown as a power function of actual stimulus intensities (𝐹(𝐼)~𝐼𝑛), 428

across many sensory modalities including olfaction (Stevens, 1957). Our results reveal that a 429

phenomenon analogous to Stevens’s Law can be attributed to the olfactory sensory periphery 430

itself, a direct outcome of the statistical distribution of response sensitivities across the ORN 431

population. Cao et al. (Cao et al., 2016; Gorur-Shandilya et al., 2017) find that at the individual 432

ORN level, adaptation scales ORN gain with respect to odorant concentration according to the 433

16

Weber-Fechner Law. These findings are complementary to our study, which expands the 434

scaling observation to the entire ORN population and across several orders of magnitude in 435

concentration. Together these results suggest that multiple complementary mechanisms 436

underlie the Weber-Fechner and Steven’s Law. 437

438

Common temporal signal processing across ORNs 439

We observed a common temporal filter shape across ORNs in an environment with fluctuating 440

odorant concentrations. Although calcium imaging limited our ability to resolve possible 441

temporal differences in these filters faster than ~100 ms, we note that Martelli et al. (2013) also 442

reported remarkable similarities in their measurement of temporal filters for LN models that 443

connect odorant dynamics to electrical activity in adult Drosophila ORNs (Martelli et al., 2013). 444

Common temporal filters are also consistent with the observation of a fixed degree and kinetics 445

of adaptation in ORNs (Martelli et al., 2013). 446

447

A constant temporal filter in conjunction with uniform ORN activation function over 448

concentrations could allow a population of responsive neurons to maintain the same relative 449

amplitudes of activation in a static or fluctuating odorant environment. Instantaneous olfactory 450

representations would not change simply because each ORN has a different rate of activation or 451

inactivation in response to the same changes in odor concentration. Constant temporal filters 452

across ORNs may arise from stereotyped transduction dynamics among ORNs. Thus, a 453

common temporal filter shape across ORNs could preserve the olfactory code in an 454

environment with fluctuating odor concentrations. 455

456

Implications for olfactory processing 457

Shared patterns in single and population level ORN responses across olfactory space are in 458

agreement with previously described circuit mechanisms in the antennal lobe, the first olfactory 459

processing center in Drosophila. Normalization is thought to occur through a class of local 460

interneurons that receive and pool inputs from all ORNs, thereby normalizing the olfactory 461

representation across concentrations through inhibitory feedback (Olsen and Wilson, 2008; 462

Olsen et al., 2010). Anatomical studies in the larva have revealed the class of local interneurons 463

that could carry out this normalization function (Berck et al., 2016). The shared activation 464

function across ORNs and across odorants could benefit this circuit mechanism by preserving 465

the rank order of ORN activity before and after normalization, helping maintain odorant 466

identification and discrimination across varying intensities. The power law distribution of 467

17

sensitivities means that most ORNs will be weakly active for most odors at most concentrations. 468

A nonlinear transformation from ORNs to PNs during divisive normalization helps amplify the 469

signal of weakly active ORNs (Olsen et al., 2010), benefitting the population level 470

representation. 471

472

For animals that sniff, the change in concentration through inhalation generates a temporal 473

sequence of ORN activity, reflecting the order of olfactory sensitivities for the inhaled odorant 474

molecules. In a primacy code for odorant identification, animals use the first set of activated 475

ORNs to recognize an odor (Wilson et al., 2017). Such a code would benefit from shared 476

activation functions across ORNs by preserving the rank order of ORN activation as odorants 477

are sniffed at different concentrations. Furthermore, high sensitivity ORNs are generally less 478

densely distributed along the sensitivity spectrum, in relation to other ORNs. Thus, the sensitive 479

ORNs would be the most informative for odor identification by the primacy code since they are 480

more separated. 481

482

In conclusion, this study highlights properties of individual ORNs (common activation functions 483

with different sensitivities) and of the ORN population (a power law distribution of sensitivities) 484

that may reflect simple strategies for representing odor identity and intensity. In the future, it will 485

be interesting to examine how downstream olfactory neurons ultimately extract and use the 486

described structure in ORN responses. 487

488

Acknowledgments 489

The authors would like to thank valuable discussions with Kenny Blum, Benjamin de Bivort, 490

John Carlson, Dennis Mathew, Venki Murthy, Dmitri Chklovskii, Cengiz Pehlevan, Yuhai Tu, 491

Marta Zlatic, Betty Hong, Mei Zhen, Armin Bahl, and members of the Samuel lab. This work was 492

performed in part at the Harvard Center for Nanoscale Systems, a member of the National 493

Nanotechnology Infrastructure Network (NNIN), which is supported by the National Science 494

Foundation under NSF award no. ECS-0335765. This work was also supported by a National 495

Science Foundation Graduate Research Fellowship (DGE1144152), NSF Brain Initiative grant 496

(NSF-IOS-1556388), and grants from the NIH (8DP1GM105383, P01GM103770, 497

F31DC015704). YH acknowledges support from the Swartz Program in Theoretical 498

Neuroscience at Harvard. 499

500

Author Contributions 501

18

G.S., J.K., and A.S. designed the experiments. G.S., M.B., and J.K. designed and fabricated the 502

microfluidic chips. G.S. and J.B. designed and customized the odor delivery and imaging setup. 503

C.T. made biological reagents. G.S. and J.K. collected the data. Y.H. derived power law scaling 504

from the sensitivity distribution. J.B. developed the maximum likelihood fitting method. G.S., 505

J.K., Y.H, J.B., and G.V. analyzed the data. G.S., J.K., and A.S. wrote the paper. 506

507

Declaration of Interests 508

The authors declare no competing interests. 509

19

References 510

Andersson, M.N., Schlyter, F., Hill, S.R., and Dekker, T. (2012). What Reaches the Antenna ? 511 How to Calibrate Odor Flux and Ligand – Receptor Affinities. Chem. Senses 37, 403–420. 512 Apostolopoulou, A.A., Widmann, A., Rohwedder, A., Pfitzenmaier, J.E., and Thum, A.S. (2013). 513 Appetitive Associative Olfactory Learning in Drosophila Larvae. J. Vis. Exp. 1–11. 514 Asahina, K., Louis, M., Piccinotti, S., and Vosshall, L.B. (2009). A circuit supporting 515 concentration-invariant odor perception in Drosophila. J. Biol. 8, 9. 516 Berck, M.E., Khandelwal, A., Claus, L., Hernandez-Nunez, L., Si, G., Tabone, C.J., Li, F., 517 Truman, J.W., Fetter, R.D., Louis, M., et al. (2016). The wiring diagram of a glomerular olfactory 518 system. Elife 5, 1–21. 519 Bolding, K.A., and Franks, K.M. (2017). Complementary codes for odor identity and intensity in 520 olfactory cortex. Elife 6, 1–26. 521 De Bruyne, M., Foster, K., and Carlson, J.R. (2001). Odor coding in the Drosophila antenna. 522 Neuron 30, 537–552. 523 Buck, L., and Axel, R. (1991). A novel multigene family may encode odorant receptors: A 524 molecular basis for odor recognition. Cell 65, 175–187. 525 Cao, L.-H., Jing, B.-Y., Yang, D., Zeng, X., Shen, Y., Tu, Y., and Luo, D.-G. (2016). Distinct 526 signaling of Drosophila chemoreceptors in olfactory sensory neurons. Proc. Natl. Acad. Sci. 527 113, E902–E911. 528 Cao, L.H., Yang, D., Wu, W., Zeng, X., Jing, B.Y., Li, M.T., Qin, S., Tang, C., Tu, Y., and Luo, 529 D.G. (2017). Odor-evoked inhibition of olfactory sensory neurons drives olfactory perception in 530 Drosophila. Nat. Commun. 8, 1–13. 531 Celani, A., Villermaux, E., and Vergassola, M. (2014). Odor landscapes in turbulent 532 environments. Phys. Rev. X 4, 1–17. 533 Chen, T.W., Wardill, T.J., Sun, Y., Pulver, S.R., Renninger, S.L., Baohan, A., Schreiter, E.R., 534 Kerr, R.A., Orger, M.B., Jayaraman, V., et al. (2013). Ultrasensitive fluorescent proteins for 535 imaging neuronal activity. Nature 499, 295–300. 536 Chen, Y.C., Mishra, D., Schmitt, L., Schmuker, M., and Gerber, B. (2011). A behavioral odor 537 similarity “space” in larval Drosophila. Chem. Senses 36, 237–249. 538 Chronis, N., Zimmer, M., and Bargmann, C.I. (2007). Microfluidics for in vivo imaging of 539 neuronal and behavioral activity in Caenorhabditis elegans. Nat. Methods 4, 727–731. 540 Clauset, A., Shalizi, C.R., and Newman, M.E.J. (2009). Power-Law Distributions in Empirical 541 Data. SIAM Rev. 51, 661–703. 542 Cleland, T. a, Johnson, B. a, Leon, M., and Linster, C. (2007). Relational representation in the 543 olfactory system. Proc. Natl. Acad. Sci. U. S. A. 104, 1953–1958. 544 Dweck, H.K.M., Ebrahim, S.A.M., Retzke, T., Grabe, V., Weißflog, J., Svatoš, A., Hansson, 545 B.S., and Knaden, M. (2018). The Olfactory Logic behind Fruit Odor Preferences in Larval and 546 Adult Drosophila. Cell Rep. 23, 2524–2531. 547 Field, D.J. (1987). Relations between the statistics of natural images and the response 548 properties of cortical cells. J. Opt. Soc. Am. A. 4, 2379–2394. 549 Fishilevich, E., Domingos, A.I., Asahina, K., Naef, F., Vosshall, L.B., and Louis, M. (2005). 550 Chemotaxis behavior mediated by single larval olfactory neurons in Drosophila. Curr. Biol. 15, 551 2086–2096. 552 Friedrich, R.W., and Korsching, S.I. (1997). Combinatorial and chemotopic odorant coding in the 553 zebrafish olfactory bulb visualized by optical imaging. Neuron 18, 737–752. 554 Friedrich, R.W., and Laurent, G. (2001). Dynamic optimization of odor representations by slow 555 temporal patterning of mitral cell activity. Science 291, 889–894. 556 Geffen, M.N., Broome, B.M., Laurent, G., and Meister, M. (2009). Neural Encoding of Rapidly 557 Fluctuating Odors. Neuron 61, 570–586. 558 Gorur-Shandilya, S., Demir, M., Long, J., Clark, D.A., and Emonet, T. (2017). Olfactory receptor 559

20

neurons use gain control and complementary kinetics to encode intermittent odorant stimuli. 560 Elife 6, 1–30. 561 Grillet, M., Campagner, D., Petersen, R., McCrohan, C., and Cobb, M. (2016). The peripheral 562 olfactory code in Drosophila larvae contains temporal information and is robust over multiple 563 timescales. Proc. R. Soc. B Biol. Sci. 283. 564 Haddad, R., Khan, R., Takahashi, Y.K., Mori, K., Harel, D., and Sobel, N. (2008). A metric for 565 odorant comparison. Nat. Methods 5, 425–429. 566 Hallem, E.A., and Carlson, J.R. (2006). Coding of Odors by a Receptor Repertoire. Cell 125, 567 143–160. 568 Hallem, E. a, Ho, M.G., and Carlson, J.R. (2004). The molecular basis of odor coding in the 569 Drosophila antenna. Cell 117, 965–979. 570 Junek, S., Kludt, E., Wolf, F., and Schild, D. (2010). Olfactory Coding with Patterns of Response 571 Latencies. Neuron 67, 872–884. 572 Kajiya, K., Inaki, K., Tanaka, M., Haga, T., Kataoka, H., and Touhara, K. (2001). Molecular 573 bases of odor discrimination: Reconstitution of olfactory receptors that recognize overlapping 574 sets of odorants. J. Neurosci. 21, 6018–6025. 575 Kato, S., Xu, Y., Cho, C.E., Abbott, L.F., and Bargmann, C.I. (2014). Temporal Responses of C. 576 elegans Chemosensory Neurons Are Preserved in Behavioral Dynamics. Neuron 81, 616–628. 577 Kreher, S.A., Kwon, J.Y., and Carlson, J.R. (2005). The Molecular Basis of Odor Coding in the 578 Drosophila Larva. Neuron 46, 445–456. 579 Kreher, S.A., Mathew, D., Kim, J., and Carlson, J.R. (2008). Translation of Sensory Input into 580 Behavioral Output via an Olfactory System. Neuron 59, 110–124. 581 Lancet, D., Sadovsky, E., and Seidemann, E. (1993). Probability model for molecular 582 recognition in biological receptor repertoires: significance to the olfactory system. Proc. Natl. 583 Acad. Sci. 90, 3715–3719. 584 Malnic, B., Hirono, J., Sato, T., and Buck, L.B. (1999). Combinatorial Receptor Codes for Odors. 585 Cell 96, 713–723. 586 Martelli, C., Carlson, J.R., and Emonet, T. (2013). Intensity invariant dynamics and odor-specific 587 latencies in olfactory receptor neuron response. J. Neurosci. 33, 6285–6297. 588 Mathew, D., Martelli, C., Kelley-Swift, E., Brusalis, C., Gershow, M., Samuel, A.D.T., Emonet, 589 T., and Carlson, J.R. (2013). Functional diversity among sensory receptors in a Drosophila 590 olfactory circuit. Proc. Natl. Acad. Sci. 110, E2134–E2143. 591 Mazor, O., and Laurent, G. (2005). Transient Dynamics versus Fixed Points in Odor 592 Representations by Locust Antennal Lobe Projection Neurons. Neuron 48, 661–673. 593 Meister, M., and Bonhoeffer, T. (2001). Tuning and topography in an odor map on the rat 594 olfactory bulb. J. Neurosci. 21, 1351–1360. 595 Mishra, D., Chen, Y.-C., Yarali, A., Oguz, T., and Gerber, B. (2013). Olfactory memories are 596 intensity specific in larval Drosophila. J. Exp. Biol. 216, 1552–1560. 597 Münch, D., and Galizia, C.G. (2016). DoOR 2.0--Comprehensive Mapping of Drosophila 598 melanogaster Odorant Responses. Sci. Rep. 6, 21841. 599 Murlis, J. (1992). Odor Plumes And How Insects Use Them. Annu. Rev. Entomol. 37, 505–532. 600 Nagel, K.I., and Wilson, R.I. (2011). Biophysical mechanisms underlying olfactory receptor 601 neuron dynamics. Nat. Publ. Gr. 14. 602 Naka, K.I., and Rushton, W.A. (1966). S-potentials from colour units in the retina of fish 603 (Cyprinidae). J. Physiol. 185, 536–555. 604 Nara, K., Saraiva, L.R., Ye, X., and Buck, L.B. (2011). A large-scale analysis of odor coding in 605 the olfactory epithelium. J Neurosci 31, 9179–9191. 606 Olsen, S.R., and Wilson, R.I. (2008). Lateral presynaptic inhibition mediates gain control in an 607 olfactory circuit. Nature 452, 956–960. 608 Olsen, S.R., Bhandawat, V., and Wilson, R.I. (2010). Divisive normalization in olfactory 609 population codes. Neuron 66, 287–299. 610

21

Pelz, C., Gerber, B., Menzel, R., Schleyer, M., and Gerber, B. (1997). Odorant intensity as a 611 determinant for olfactory conditioning in honeybees: roles in discrimination, overshadowing and 612 memory consolidation. J. Exp. Biol. 200, 837–847. 613 Pnevmatikakis, E.A., Soudry, D., Gao, Y., Machado, T.A., Merel, J., Pfau, D., Reardon, T., Mu, 614 Y., Lacefield, C., Yang, W., et al. (2016). Simultaneous Denoising, Deconvolution, and Demixing 615 of Calcium Imaging Data. Neuron 89, 299. 616 Ramaekers, A., Magnenat, E., Marin, E.C., Gendre, N., Jefferis, G.S.X.E., Luo, L., and Stocker, 617 R.F. (2005). Glomerular maps without cellular redundancy at successive levels of the 618 Drosophila larval olfactory circuit. Curr. Biol. 15, 982–992. 619 Raman, B., Joseph, J., Tang, J., and Stopfer, M. (2010). Temporally Diverse Firing Patterns in 620 Olfactory Receptor Neurons Underlie Spatiotemporal Neural Codes for Odors. J. Neurosci. 30, 621 1994–2006. 622 Riffell, J.A., Abrell, L., and Hildebrand, J.G. (2008). Physical processes and real-time chemical 623 measurement of the insect olfactory environment. J. Chem. Ecol. 34, 837–853. 624 Roland, B., Deneux, T., Franks, K.M., Bathellier, B., and Fleischmann, A. (2017). Odor identity 625 coding by distributed ensembles of neurons in the mouse olfactory cortex. Elife 6, 1–26. 626 Sachse, S., and Galizia, C.G. (2003). The coding of odour-intensity in the honeybee antennal 627 lobe: local computation optimizes odour representation. Eur. J. Neurosci. 18, 2119–2132. 628 Sass, H. (1976). Sensory Encoding of Odor Stimuli in Periplaneta americana. J. Comp. Physiol. 629 A 107, 49–65. 630 Simoncelli, E.P., and Olshausen, B.A. (2001). Natural image statistics and neural 631 representation. Annu. Rev. Neurosci. 24, 1193–1216. 632 Singh, R.N., and Singh, K. (1984). Fine structure of the sensory organs of Drosophila 633 melanogaster Meigen larva (Diptera : Drosophilidae). Int. J. Insect Morphol. Embryol. 13, 255–634 273. 635 Stevens, C.F. (2016). A statistical property of fly odor responses is conserved across odors. 636 Proc. Natl. Acad. Sci. U. S. A. 113, 6737–6742. 637 Stevens, S.S. (1957). On the psychophysical law. Psychol. Rev. 64, 153–181. 638 Stopfer, M., Jayaraman, V., and Laurent, G. (2003). Intensity versus identity coding in an 639 olfactory system. Neuron 39, 991–1004. 640 Su, C.-Y., Menuz, K., and Carlson, J.R. (2009). Olfactory Perception: Receptors, Cells, and 641 Circuits. Cell 139, 45–59. 642 Theunissen, F.E., and Elie, J.E. (2014). Neural processing of natural sounds. Nat. Rev. 643 Neurosci. 15, 355–366. 644 Thévenaz, P., Ruttimann, U.E., and Unser, M. (1998). A pyramid approach to subpixel 645 registration based on intensity. IEEE Trans. Image Process. 7, 27–41. 646 Tichy, H., Hinterwirth, A., and Gingl, E. (2005). Olfactory receptors on the cockroach antenna 647 signal odour ON and odour Off by excitation. Eur. J. Neurosci. 22, 3147–3160. 648 Todeschini, R., and Lasagni, M. (1994). New Molecular Descriptors for 2D and 3D Structures. 649 Theory. J. Chemom. 8, 263–272. 650 Uchida, N. (2008). Odor concentration invariance by chemical ratio coding. Front. Syst. 651 Neurosci. 1, 3. 652 Vosshall, L.B., and Stocker, R.F. (2007). Molecular architecture of smell and taste in Drosophila. 653 Annu. Rev. Neurosci. 30, 505–533. 654 Vosshall, L.B., Amrein, H., Morozov, P.S., Rzhetsky, A., and Axel, R. (1999). A spatial map of 655 olfactory receptor expression in the Drosophila antenna. Cell 96, 725–736. 656 Wachowiak, M., Cohen, L.B., and Zochowski, M.R. (2002). Distributed and concentration-657 invariant spatial representations of odorants by receptor neuron input to the turtle olfactory bulb. 658 J. Neurophysiol. 87, 1035–1045. 659 Wang, Y. (2004). Stereotyped Odor-Evoked Activity in the Mushroom Body of Drosophila 660 Revealed by Green Fluorescent Protein-Based Ca2+ Imaging. J. Neurosci. 24, 6507–6514. 661

22

Wang, J.W., Wong, A.M., Flores, J., Vosshall, L.B., and Axel, R. (2003). Two-photon calcium 662 imaging reveals an odor-evoked map of activity in the fly brain. Cell 112, 271–282. 663 Wilson, C.D., Serrano, G.O., Koulakov, A.A., and Rinberg, D. (2017). A primacy code for odor 664 identity. Nat. Commun. 8, 1477. 665

666

23

Main Figures 667

668

Figure 1. Anatomical and functional identification of individual ORNs within the 669

complete population. 670

(A) Schematic of the microfluidic setup for odorant delivery and larval ORN calcium imaging. 671

24

(B) 16-channel microfluidic chip. Arrowhead marks inlet channel for loading a larva, arrow marks 672

outlet channel for fluid waste, and * marks odorant stimuli delivery channels. 673

(C-D) Magnified views (10X in C and 40X in D) of an immobilized larva in the inlet channel. Red 674

indicates RFP labeling of all ORN dendrites and cell bodies. 675

(E) Organization of the seven ORN dendritic bundles (numbered) in the larva. Or35a>GFP; 676

Orco>RFP used to label all ORNs in red and the Or35a-ORN in green. Dashed line in lateral 677

view marks separation between ventral and dorsal bundles. 678

(F) Functional mapping between each of 15 odorants that primarily activate a single ORN within 679

each dendritic bundle, at low concentrations. Size of shaded circles indicates normalized neural 680

activity level (F/F) of the specified ORN to an odorant. * indicates inferred location of the 681

Or33a-ORN based on dendritic bundle 2 vacancy (Figure S2). 682

(G) DOG cell body locations and GCaMP6m responses of four ORNs responsive to 1-pentanol 683

(left). Fluorescence intensity changes for each of the four ORN cell bodies during pulsed 684

presentations of increasing 1-pentanol concentrations (right). Arrow indicates time point at 685

which left panel in G was captured. See also Figure S1, Figure S2 and Video S1. 686

25

687

26

Figure 2. Orthogonality of odorant identity and intensity in ORN population activity. 688

(A) Averaged peak responses of all 21 ORNs to a panel of 34 odorants, each delivered at five 689

concentrations (n ≥ 5 for each odorant type and concentration; odorant pulse duration = 5 s). 690

(B) PCA of ORN population responses. Colored dots represent the projection of ORN 691

population activity onto the first three principal components. Size and color of dots correspond 692

to odorant concentration and type, respectively. Dots from the same odorant are linked and the 693

molecular structure of the odorant is shown adjacent to each trajectory. Aromatic versus 694

aliphatic odorants cluster in separate regions of PCA space. See also Figure S3, Table S1 and 695

Data S1. 696

27

697

Figure 3. Individual ORNs share a common activation function and population 698

sensitivities follow a power law distribution. 699

(A) Normalized ORN responses across relative odorant concentration (actual concentration 700

divided by 𝐸𝐶50), for odorant-ORN pairs reaching saturation. Individual curves for plotted 701

28

odorant-ORN pairs collapse onto a single curve described by a Hill equation with a shared Hill 702

coefficient of 1.42. Black line indicates the Hill equation fit. Each distinct colored and shaped 703

point represents data from an unique odorant-ORN pair. 704

(B) Heatmap of the logarithm of sensitivity values, log10(1/𝐸𝐶50), from each odorant-ORN pair. * 705

for black elements indicates odorant-ORN pairs that had no response within the tested 706

concentration range. 707

(C) Raster plot of ORN sensitivities (defined as 1/𝐸𝐶50) for each odorant. Each tick mark 708

represents an ORN. 709

(D) Log-log plot of the cumulative distribution function of ORN sensitivities across all odorants. 710

Dashed line is a linear fit to the data with slope = −0.42. 711

(E) Log-log plot of average neuron activity (∆𝐹/𝐹) across all odorant-ORN pairs for each 712

concentration. Error bars = SEM. Slope of least-squares fit line = 0.32 ± 0.06 (𝑅2 = 0.99). See 713

also Figure S4 and Table S2. 714

29

715

Figure 4. Correlation between ORN-odorant sensitivities and odorant molecular 716

structure. 717

(A) Percentage of variance explained by each principal component of the ORN sensitivity matrix 718

in Figure 3B. Data compared with the results from 1000 randomly shuffled matrices. 719

(B) Weights indicating each ORN’s contribution to the 1st principal component. 720

(C) Each odorant’s (indicated by molecular structure) projection onto the 1st principal 721

component. 722

(D) Correlation between each odorant’s projection on the 1st principal component and the most 723

correlated molecular descriptor, P1s. 724

30

725

31

Figure 5. Common temporal signal processing across ORNs. 726

(A) Or42a-ORN response to an m-sequence delivery of 3-pentanol at 10-7 dilution. Red 727

indicates on-off stimulus sequence over time and black curve indicates ORN response. 728

(B) Linear filter calculated via reverse-correlation analysis from the data shown in A. 729

(C) Linear filters of seven ORNs responding to 3-octanol across five concentrations. Black curve 730

indicates the averaged filter from data across multiple animals (individual filters shown in gray). 731

(D-E) Comparison of filter waveforms for the same odorant (10-4 dilution of 3-octanol) activating 732

different ORNs (D), and the same ORN (Or85c) responding to different odorants and 733

concentrations (E). All filters were normalized by their peak amplitude. 734

(F-G) Distribution of peak time (F) and decay time (G) of 31 averaged filters measured from 735

various ORN and odorant stimuli. Distributions of peak and decay times were fit to Gaussian 736

distributions with mean and variance indicated below each histogram. See also Figure S5 and 737

Video S2. 738

32

STAR METHODS 739

KEY RESOURCES TABLE 740

REAGENT or RESOURCE SOURCE IDENTIFIER

Chemicals

4,5-Dimethylthiazole Sigma-Aldrich CAS 3581-91-7

Geranyl acetate Sigma-Aldrich CAS 105-87-3

2,5-dimethylpyrazine Sigma-Aldrich CAS 123-32-0

2-acetylpyridine Sigma-Aldrich CAS 1122-62-9

3-octanol Sigma-Aldrich CAS 589-98-0

6-methyl-5-hepten-2-ol Sigma-Aldrich CAS 1569-60-4

1-pentanol Sigma-Aldrich CAS 71-41-0

Isoamyl acetate Sigma-Aldrich CAS 123-92-2

Butyl acetate Sigma-Aldrich CAS 123-86-4

Ethyl butyrate Sigma-Aldrich CAS 105-54-4

Benzaldehyde Sigma-Aldrich CAS 100-52-7

Methyl salicylate Sigma-Aldrich CAS 119-36-8

2-heptanone Sigma-Aldrich CAS 110-43-0

4-methylcyclohexanol Sigma-Aldrich CAS 5899-91-3

Ethyl acetate Sigma-Aldrich CAS 141-78-6

2-phenylethanol Sigma-Aldrich CAS 60-12-8

Pentyl acetate Sigma-Aldrich CAS 628-63-7

3-Pentanol Sigma-Aldrich CAS 584-02-1

Anisole Sigma-Aldrich CAS 100-66-3

Methyl phenyl sulfide (Thioanisole) Sigma-Aldrich CAS 100-68-5

Trans-3-Hexen-1-ol Sigma-Aldrich CAS 928-97-2

Acetal Sigma-Aldrich CAS 105-57-7

2-Nonanone Sigma-Aldrich CAS 821-55-6

4-Hexen-3-one Sigma-Aldrich CAS 2497-21-4

4-Methyl-5-vinylthiazole Sigma-Aldrich CAS1759-28-0

Trans, trans-2,4-Nonadienal Sigma-Aldrich CAS 5910-87-2

2-Methoxyphenyl acetate Sigma-Aldrich CAS 613-70-7

Hexyl acetate Sigma-Aldrich CAS 142-92-7

Benzyl acetate Sigma-Aldrich CAS 140-11-4

Linalool Sigma-Aldrich CAS 78-70-6

(1R)-(-)-myrtenal Sigma-Aldrich CAS 18484-69-6

4-phenyl-2-butanol Sigma-Aldrich CAS 2344-70-9

Pentanoic acid (valeric acid) Sigma-Aldrich CAS 109-52-4

Nonane Sigma-Aldrich CAS 111-84-2

Menthol Sigma-Aldrich CAS 89-78-1

Experimental Models: Organisms/Strains

UAS-mCherry.NLS; UAS-GCaMP6m This study

UAS-mCD8::GFP; Orco::RFP Bloomington Drosophila Stock Center (BDSC)

RRID:BDSC_63045

Orco-Gal4 BDSC RRID:BDSC_23292

Or1a-Gal4 BDSC RRID:BDSC_9949

Or7a-Gal4 BDSC RRID:BDSC_23907 RRID:BDSC_23908

33

Or13a-Gal4 BDSC RRID:BDSC_9945

Or22c-Gal4 BDSC RRID:BDSC_9953

Or24a-Gal4 BDSC RRID:BDSC_9958

Or30a-Gal4 BDSC RRID:BDSC_9960

Or33b-Gal4 BDSC RRID:BDSC_9963

Or35a-Gal4 BDSC RRID:BDSC_9968

Or42a-Gal4 BDSC RRID:BDSC_9970

Or42b-Gal4 BDSC RRID:BDSC_9971

Or45a-Gal4 BDSC RRID:BDSC_9976

Or45b-Gal4 BDSC RRID:BDSC_9977

Or47a-Gal4 BDSC RRID:BDSC_9982

Or49a-Gal4/Cyo; Dr/TM3 Gift from John Carlson

Or59a-Gal4 BDSC RRID:BDSC_9990

Or63a-Gal4 BDSC RRID:BDSC_9992

Or67b-Gal4 BDSC RRID:BDSC_9995

Or74a-Gal4 BDSC RRID:BDSC_23123

Or82a-Gal4 BDSC RRID:BDSC_23125

Or83a-Gal4 BDSC RRID:BDSC_23128

Or85c-Gal4 BDSC RRID:BDSC_23913

Or94b-Gal4 BDSC RRID:BDSC_23916

Deposited Data

Raw and analyzed ORN dose-response data

This paper https://github.com/samuellab/Larval-ORN

Raw activity data of 21 ORNs responding to 34 odorants.

This paper https://data.mendeley.com/datasets/7kbsmx94zm/draft?a=8ad617e6-e9db-4e4f-b1a2-776b00fb4c58

Software and Algorithms

TurboReg Thévenaz et al., 1998 http://bigwww.epfl.ch/thevenaz/turboreg/

CaImAn-MATLAB Pnevmatikakis et al., 2016 https://github.com/flatironinstitute/CaImAn-MATLAB

Power-law distribution in empirical data Clauset et al., 2009 http://tuvalu.santafe.edu/~aaronc/powerlaws/

Algorithms for analyzing ORN ensemble dose-response data

This paper https://github.com/samuellab/Larval-ORN

Linear-Nonlinear model analysis of ORNs’ response to m-sequence stimuli

This paper https://github.com/samuellab/Larval-ORN

741

Contact for Reagent and Resource Sharing 742

Further information and requests for resources and reagents should be directed to and will be 743

fulfilled by the lead contact, Dr. Aravinthan Samuel (samuel@physics.harvard.edu). 744

745

Experimental Model and Subject Details 746

Drosophila melanogaster flies were reared at 22°C under a 12:12 hour light/dark cycle in vials 747

containing conventional cornmeal-agar based medium. Adult flies were transferred to a larvae 748

34

collection cage (Genesee Scientific) containing a grape juice agar plate and a dime-sized 749

amount of fresh yeast paste. Flies could lay eggs on the grape juice agar plate for two days and 750

then the plate was removed for collection of first instar larvae. Both female and male larvae 751

were used in all experiments. Transgenic stocks were obtained from the Bloomington 752

Drosophila Stock Center (BDSC). The following fly strains were used in this study: UAS-753

mCherry.NLS; UAS-GCaMP6m (a combination of UAS-mCherry.NLS:BL38425, and UAS-754

GCaMP6m:BL42750), UAS-mCD8::GFP; Orco::RFP (BL63045), Orco-Gal4 (BL23292), Or1a-755

Gal4 (BL9949), Or7a-Gal4 (BL23908 and BL23907), Or13a-Gal4 (BL9945), Or22c-Gal4 756

(BL9953), Or24a-Gal4 (BL9958), Or30a-Gal4 (BL9960), Or33b-Gal4 (BL9963), Or35a-Gal4 757

(BL9968), Or42a-Gal4 (BL9970), Or42b-Gal4 (BL9971), Or45a-Gal4 (BL9976), Or45b-Gal4 758

(BL9977), Or47a-Gal4 (BL9982), Or49a-Gal4/Cyo; Dr/TM3 (gift from John Carlson lab), Or59a-759

Gal4 (BL9990), Or63a-Gal4 (BL9992), Or67b-Gal4 (BL9995), Or74a-Gal4 (BL23123), Or82a-760

Gal4 (BL23125), Or83a-Gal4 (BL23128), Or85c-Gal4 (BL23913), and Or94b-Gal4 (BL23916). 761

762

Method Details 763

Microfluidic device design, fabrication, and specifications 764

The microfluidic device pattern was designed using AutoCAD. Devices were designed to have 765

either 8, 16, or 24 channels, but all devices operate using a similar strategy (Figure S1A). For 766

example, the 16-channel device includes two control channels to direct odorant flow, 13 odorant 767

channels, one water channel to remove odorant residue, a larva loading channel, and a waste 768

channel. The odorant, water, and control channels are of equal length to ensure equal 769

resistance. The device was designed to function using a directed flow strategy similar to that 770

described in (Chronis et al., 2007). Briefly, the device always has three channels open: the 771

water, an odorant and a control channel (Figure S1C). Switching between the two control 772

channels directs either water or an odorant to flow past the larva, as demonstrated in Figure 773

S1D. The loading channel is 70 µm high with a width starting at 300 µm and gradually tapering 774

to 60 µm in order to immobilize a first instar larva. The tapered end of the loading channel is 775

positioned perpendicular to a stimulus delivery channel to allow for odorant flow past larval 776

ORNs (Figure 1D). 777

778

The design pattern was sent to a mask-making service (outputcity.com), which provided the 779

photomask. The mask was then transferred onto a silicon wafer using photolithography. The 780

wafer was used to fabricate microfluidic devices using polydimethylsiloxane (PDMS) and the 781

standard soft lithography approach. The resulting PDMS molds were cut and bonded to glass 782

35

cover slips. Each microfluidic device was used with only a single panel of odorants to prevent 783

odorant contamination. 784

785

We used fluorescein dye to measure the switching time between water and odorants as well as 786

to verify the spatial odorant profile in the device during stimulus delivery. Our standard air 787

pressure for stimulus delivery was 6 psi, which led to a flow rate of 0.5 mL/min or 0.2m/s in the 788

microfluidic device. With these conditions, the switching time between water and odorant was 789

~20 ms (Figure S1B). 790

791

Odorant delivery 792

Odorants were obtained from Sigma-Aldrich, diluted in deionized (DI) water (Millipore) and 793

stored for no more than one week. To prevent contamination, each odorant concentration was 794

stored in a separate glass bottle and delivered through its own syringe and tubing set. Panels of 795

odorants were delivered using a 16-channel pinch valve perfusion system (AutoMate Scientific, 796

Inc.). Each syringe and tubing set contained a 30 mL luer lock glass syringe (VWR) connected 797

to Tygon FEP-lined tubing (Cole-Parmer), which in turn was connected to silicone tubing 798

(AutoMate Scientific. Inc.). The silicone tubing was placed through the pinch valve region of the 799

perfusion system since its flexibility could allow for the passage or blockage of fluid flow to the 800

microfluidics device. The silicone tubing was then connected to PTFE tubing (Cole-Parmer), 801

which was then inserted into the microfluidic device. A microcontroller and custom written 802

Matlab code were used to control the on/off sequence of the valves and to synchronize valve 803

control with the onset of recording in the imaging software (NIS Elements). 804

805

The larva experienced continuous fluid flow at a rate of 0.5 mL/min for the entire duration of a 806

recording. In dose-response experiments, the stimulus sequence consisted of five second 807

odorant pulses interleaved by a water washout period. The odorant pulse duration time was 808

chosen such that ORN soma responses reached maximum amplitude (Figure S1E). The water 809

washout duration was adjusted based on stimulus concentration to allow for ORN recovery back 810

to baseline activity levels, and thus ensured that measurements of ORN responses were 811

independent of stimulus sequence (Figure S1F and Video S1). For white noise experiments, a 812

1024-step m-sequence of odorant and water was delivered with a time step of 0.2 s (Video S2). 813

814

Odor selection 815

36

We pooled all 690 odorants described on the database for odor responses (DoOR), a repository 816

of odor response measurements from studies on the fruit fly Drosophila melanogaster and the 817

honeybee Apis mellifera (Münch and Galizia, 2016). For each of the 690 odorants, we used the 818

E-Dragon software (http://www.vcclab.org/lab/edragon/) to determine the values corresponding 819

to 32 molecular descriptors of the multidimensional odor metric described in (Haddad et al., 820

2008). We next performed PCA on all odorants with their descriptors and visualized odorant 821

distribution using the first three PCs. We first picked the 19 odorants described in (Mathew et 822

al., 2013) for our panel. To select the remaining odorants, we picked ones such that our odorant 823

panel closely matched the density profile and distribution of the 690 odorants represented along 824

the first 3 PCs of the PCA space. We found that 35 odorants were sufficient to meet our 825

distribution matching requirements and for the panel to include stimuli that cover a broad range 826

of molecular functional groups, are relevant environmental odorants for the larva, have been 827

previously shown to elicit behavioral responses in the larva (see Figure S3A and Table S1). 828

During our imaging experiments, we found that one odorant, pentanoic acid, did not elicit any 829

ORN responses and thus removed it from the panel. 830

831

Calcium imaging 832

First instar larvae were loaded into a microfluidic device using a 1 mL syringe filled with a 0.1% 833

triton-water solution. Each larva was gently pushed to the end of the loading channel where it 834

was mechanically trapped. The larva was positioned such that its left and right dorsal organs 835

(nose) were exposed to the stimulus delivery channel and its dorsal side (location of ORN cell 836

bodies) was closest to the objective. 837

838

Larvae were imaged using an inverted Nikon Ti-e spinning disk confocal microscope with a 60X 839

water immersion objective (NA 1.2). A charged-coupled device (CCD) microscope camera 840

(Andor iXon EMCCD) captured images at 30 frames/sec. ORN cell bodies were recorded by 841

scanning the entire volume (~20 slices with a step size of 1.5 µm) of the dorsal organ ganglion 842

(Video S1), while ORN axon terminals were recorded from a single slice of the antennal lobe 843

(Video S2). 844

845

Initial experiments were performed to verify the identity of ORNs activated by each of the 19 846

odorants that were described in (Mathew et al., 2013) and used in our panel for ORN 847

identification (data not shown). We used single ORN Gal4 drivers expressing GCaMP6m to 848

confirm the ORN(s) responsive to each of the 19 odorants. 849

37

850

Dose-response experiments (data shown in Figures 1-2 and S3B-S3C, Video S1) were 851

performed using larvae of the Orco>GCaMP6m, Orco>mCherry.NLS genotype and recording 852

from ORN cell bodies. White noise experiments (data shown in Figures 5 and S5, Video S2) 853

were performed using larvae expressing GCaMP6m in a single ORN (e.g. Or42a>GCaMP6m 854

used in Figure S5) and recording from ORN axon terminals. For all experiments, both the left 855

and right sides of ORN soma or axon responses were recorded simultaneously and used for 856

analysis. Recordings from at least five larvae were collected for each odorant. All samples were 857

used for analysis unless dendritic varicosities developed over the course of the recording, a sign 858

of unhealthy neurons likely due to mechanical stress. 859

860

Quantification and Statistical Analysis 861

Data were analyzed with custom scripts written in MATLAB, available at 862

https://github.com/samuellab/Larval-ORN. The statistical tests, representations of sample 863

sizes ("n"), what each n value represents, and other related measures are shown in the legend 864

of each relevant figure. No statistical methods were used to determine sample sizes in advance, 865

but sample sizes are similar to those reported in other studies in the field. 866

867

Quantification of dose-response recordings 868

Custom code written in ImageJ and published Matlab code were used to track and identify each 869

ORN as well as its responses to odorant stimuli. Slight movement artifacts were corrected by 870

aligning frames using mCherry NLS labeling of ORN cell bodies and the ImageJ TurboReg 871

plugin (Thévenaz et al., 1998). ROI segmentation was performed using the method of 872

constrained nonnegative matrix factorization (CNMF) (Pnevmatikakis et al., 2016). Each ORN 873

activated in response to an odorant stimulus was visually identified using both the anatomical 874

location of its dendritic bundle and the functional map of cognate odorant to ORN activation 875

(Figures 1 E-G, and S2). ORN identification was performed independently by two 876

experimenters to ensure accuracy. Changes in fluorescence were then quantified as (𝐹𝑝𝑒𝑎𝑘 −877

𝐹0)/𝐹0, where 𝐹0 was the average ORN intensity sampled from the frames immediately 878

preceding odorant delivery and 𝐹𝑝𝑒𝑎𝑘 was the highest intensity in ORN fluorescence during 879

odorant delivery. Each odorant stimulus was repeated across at least 5 animals. The raw 880

response data is summarized in Data S1. 881

882

38

The heatmap in Figure 2A was generated by directly averaging the peak responses across 883

animals (no normalization was performed). Simulated annealing was used to optimize the order 884

of ORNs and odorants presented in this heatmap, such that it minimized a loss function in which 885

cost increased linearly with the distance that activated odorant-ORN pairs were from the matrix 886

diagonal. Matrices of neural responses at each concentration were concatenated, such that 887

columns corresponded to ORNs and rows corresponded to odorants, for all five concentrations. 888

We then centered the data and performed PCA using the SVD method, along the ORN axis 889

(Figure 2B). 890

891

To visualize temporal dynamics of ORN responses over the duration of stimulus delivery, we 892

performed PCA on ORN responses over time to two odorants, benzaldehyde and ethyl butyrate, 893

across five concentrations (Figure S3D). The time period starts from odorant pulse delivery and 894

continues up to 15 seconds after delivery offset. Similar methods and findings have been 895

described in insect olfactory projection neurons (Mazor and Laurent, 2005; Stopfer et al., 2003). 896

897

Estimation of EC50 values 898

Dose-response data were modeled as following a Hill equation, the general form of which is as 899

follows: 900

y = A𝑚𝑎𝑥cn

cn+EC50n (1)901

where A𝑚𝑎𝑥 is the maximum ORN response level across concentrations, 𝑐 is the odorant 902

concentration, EC50 is the half-maximal effective concentration, and 𝑛 is the Hill coefficient. 903

There were a total of 259 odorant-ORN pairs in which some odorant response was observed. 904

Saturation of the dose-response curve was observed in 36 of these odorant-ORN pairs. We first 905

performed a least-squares fit on each of the individual 36 curves and found that the curves had 906

similar Hill coefficients. However, with only five data points to describe each curve and three 907

free parameters to fit the Hill function, there was large uncertainty in the fit estimates. We 908

instead perform a population fit on all 36 saturated curves to a Hill function with a common Hill 909

coefficient, by optimizing the following equation: 910

min𝐴𝑚𝑎𝑥𝑖, 𝐸𝐶50𝑖,𝑛

∑|𝒀𝑖 − 𝑓(𝐴𝑚𝑎𝑥𝑖, 𝐸𝐶50𝑖, 𝑛, 𝒄)|

36

𝑖=1

911

where 𝑖 indicates each of the 36 saturated odorant-ORN pairs, 𝒀𝑖 is the average ORN 912

response for an odorant-ORN pair, 𝑓 is the Hill equation with parameters 𝐴𝑚𝑎𝑥𝑖 and 𝐸𝐶50𝑖 for 913

each pair, 𝑛 is the shared Hill coefficient, and 𝒄 is the five concentration levels used in this 914

39

study. The optimization method uses the standard Matlab fminsearch function. We find that this 915

approach yields a R2=0.99, which is comparable to the goodness of fit from 36 independent Hill 916

coefficients (Figure S4A). 917

918

For the remaining 223 non-saturated pairs, there were not enough points on the dose-response 919

curve to estimate parameters of the Hill equation using the previous method. To make estimates 920

of the Hill equation parameters for pairs in which saturation was not observed, we modeled a 921

single animal response using the Hill function. For a given odorant-ORN pair, for a single 922

animal, the ORN’s 𝐴, 𝑛, and 𝑘 = log10(1/𝐸𝐶50) parameters were each drawn from normal 923

distributions. Additionally, we assumed that the distributions for 𝐴 and 𝑛 as well as the variance 924

of the 𝐸𝐶50 distributions were independent of odorant and ORN. Thus, the distributions for A and 925

𝑛 across all odorant-ORN pairs were the same, and the distributions of 𝑘 across all odorant-926

ORN pairs differed only in their mean value �̅�. 927

928

The data were modeled in this way because the primary source of variance in the data was from 929

animal-to-animal response variation. This variation was much larger than the trial-to-trial 930

variation in a single animal. For a single odorant-ORN pair across different animals, different 931

values for 𝐴, 𝑛, and 𝑘 may be resolved. Further, the mean and variance in A and 𝑛 across 932

animals for the same odorant-ORN pair appeared comparable to the mean and variance in 𝐴 933

and 𝑛 for different odorant-ORN pairs. Similarly, the variance in 𝑘 across animals for the same 934

odorant-ORN pair appeared comparable to the variance in 𝑘 across animals for different 935

odorant-ORN pairs. 936

937

We perform maximum likelihood estimation on this model by writing down the probability that 938

the data might be generated by the model with particular values for the moments of the 939

distributions of 𝐴, 𝑛, and 𝑘. For this model, we use the log scale of concentration and 𝐸𝐶50: 𝑥 =940

log10(𝑐) , 𝑘 = log10(1/𝐸𝐶50), and the logistic function in place of the Hill equation: 𝑦(𝑥) =941

𝐴

1+exp (−𝜈(𝑥+𝑘)), with 𝜈 = 𝑛 ln 10. Deviations from the logistic function are modeled as Gaussian 942

noise 𝒩(𝑦𝑞𝑖(𝑥), 𝜎𝑁), where 𝜎𝑁 represents experiment noise. Specifically, for a single-animal 𝑖 of 943

one odorant-ORN pair (indexed by 𝑞), the probability of generating responses 𝑌𝑖 given normal 944

distributions for A, 𝜈, and 𝑘 is 945

40

𝑃(𝑌𝑞𝑖|�̅�; 𝜎𝐴; 𝑘𝑞̅̅ ̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁)946

=∭𝑃(𝑌𝑞𝑖|𝐴𝑗; 𝑘𝑗; 𝜈𝑗; 𝜎𝑁) 𝑃(𝐴𝑗|�̅�; 𝜎𝐴)𝑃(𝑘𝑗|𝑘𝑞̅̅ ̅; 𝜎𝑘)𝑃(𝜈𝑗|�̅�; 𝜎𝜈)𝑑𝐴𝑗𝑑𝑘𝑗𝑑𝜈𝑗 . (2) 947

Here, the first term 𝑃(𝑌𝑞𝑖|𝐴𝑗; 𝑘𝑗;𝜈𝑗; 𝜎𝑁) represents the probability of generating the data from a 948

logistic function with parameters 𝐴𝑗, 𝑘𝑗,𝜈𝑗, given Gaussian noise with variance 𝜎𝑁2. The other 949

three terms on the right side are also Gaussians. Note that the only dependence on the odorant-950

ORN pair 𝑞 is in the mean value 𝑘𝑞̅̅ ̅. For one odorant-ORN pair, across all animals, the 951

probability is then the product of the individual animals 952

𝑃(𝑌𝑞|�̅�; 𝜎A; 𝑘𝑞̅̅ ̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁) =∏𝑃(𝑌𝑞𝑖|�̅�; 𝜎A; 𝑘𝑞̅̅ ̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁)

𝑀

𝑖

953

where M is the number of animals for that odorant-ORN pair. The probability of generating all 954

the data across odorant-ORN pairs from this model is then 955

𝑃(𝑌|�̅�; 𝜎A; k̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁) = ∏𝑃(𝑌𝑞|�̅�; 𝜎A; 𝑘𝑞̅̅ ̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁),

𝑄

𝑞

956

where 𝑄 is the total number of odorant-ORN pairs and k̅ = (𝑘1, 𝑘2, … 𝑘𝑄) is the vector of 957

log10(1/𝐸𝐶50) values for all odorant-ORN pairs. Maximum likelihood estimation amounts to 958

maximization of this expression, or, typically, the logarithm of this expression 959

max�̅�;𝜎A;k̅;𝜎𝑘;�̅�;𝜎𝜈;𝜎𝑁

ln[𝑃(𝑌|�̅�; 𝜎A; k̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁)] (3) 960

961

We characterize (�̅�; 𝜎A; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁) as being global (applying to all odorant-ORN pairs) and k̅ 962

as being local (different for every odorant-ORN pair). The moments of the distribution are 963

optimized by alternatively optimizing for the global moments while fixing the local moments and 964

optimizing the local moments while fixing the global moments. This is done to limit the number 965

of optimization parameters for any given optimization step. Function maximization was 966

performed using MATLAB’s fminsearch function, and initial values for the distribution moments 967

were chosen based on a least-squares regression assuming fixed values for 𝜈 and 𝐴 across all 968

odorant-ORN pairs. The result of the optimization in equation (3) was found to be robust to 969

choices of initial values. 970

971

The estimated k̅ for all odorant-ORN pairs is summarized in Figure 3B. The black elements in 972

the matrix indicate that the corresponding ORN showed no activity within the tested 973

concentration range, i.e., we were unable to fit a �̅� value for these odorant-ORN pairs. The 974

41

standard deviation for the set of k’s was found to be 𝜎𝑘 = 0.45. Because the primary source of 975

variance in the data is variation in animal responses, we can treat this as an error bar on a 976

single animal-trial measurement of �̅�. The moments of the distribution for 𝐴 were found to be 977

�̅� = 4.1, 𝜎𝐴 = 2.3, and the moments for the distribution of 𝑛 = 𝜈/ ln 10 were found to be �̅� =978

1.6, 𝜎𝑛 = 0.6. 979

980

Using these estimated distributions, single animal-trial values for the logistic function 981

parameters 𝐴𝑖 , 𝑘𝑖, 𝜈𝑖 can be estimated using maximum likelihood estimation. Maximization is 982

performed on the probability of single animal data for a given odorant-ORN pair 𝑞 and animal 983

𝑖, which is similar to equation 2, but without the integrals: 984

𝑃(𝑌𝑞𝑖|�̅�; 𝜎𝐴; 𝑘𝑞̅̅ ̅; 𝜎𝑘; �̅�; 𝜎𝜈; 𝜎𝑁) = 𝑃(𝑌𝑞𝑖|𝐴𝑖; 𝑘𝑖; 𝜈𝑖; 𝜎𝑁)𝑃(𝐴𝑖|�̅�; 𝜎𝐴)𝑃(𝑘𝑖|𝑘𝑞̅̅ ̅; 𝜎𝑘)𝑃(𝜈𝑖|�̅�; 𝜎𝜈). (4) 985

Doing this can help test the self-consistency of the model. It is not guaranteed (or even 986

expected) that the single-trial values of 𝐴𝑖, 𝑘𝑖, 𝜈𝑖 should match the distributions of 𝐴, 𝑘, 𝜈, but they 987

should be similar, and they should fit the data well (Figure S4E, S4F and S4G). We find that the 988

values for 𝐴𝑖, 𝑘𝑖 , 𝜈𝑖 explain the data very well (𝑅2 = .989) (Figure S4D). 989

990

Analysis of the sensitivity matrix 991

To perform PCA on the sensitivity matrix, we first transformed the 𝐸𝐶50 values to the 992

𝑙𝑛(1/𝐸𝐶50), such that odorant-ORN pairs with a high sensitivity (small 𝐸𝐶50) were now 993

represented by large values, those that were less sensitive (large 𝐸𝐶50) had small values, and 994

the logarithmic scale allowed for coverage of the broad range of 𝐸𝐶50 values spanning several 995

orders of magnitude. The remaining pairs that that did not have an 𝐸𝐶50value (the missing data, 996

represented by black squares in Figure 3B), represent pairs with a much lower sensitivity and 997

were set to zero. Figure 4A shows the percentage of variance explained by each PC once PCA 998

was performed on the ln (1/𝐸𝐶50) matrix. In comparison to a shuffled matrix (in which each row 999

is randomly permuted), we found that only the first PC was significantly different (p < 0.0001 for 1000

1000 instances of shuffled data). 1001

1002

We fit the power law distribution using code from (Clauset et al., 2009). The resulting fitting 1003

index of 0.17 (large values mean better fit to the power law for this metric) is larger than the 1004

threshold (0.1) needed to accept the power law hypothesis. Furthermore, we performed a 1005

likelihood ratio comparison with other heavy tailed distributions (Clauset et al., 2009), using 1006

42

parameters best fit to our data, and found that no other distributions are significantly better 1007

(need p<0.1) than the power-law (Table S2). 1008

1009

Derivation of power law scaling of ORN population responses from sensitivity distribution 1010

Here, we explain analytically the power law relation between odorant concentration and the 1011

population response of ORNs. Under the same Hill equation we used to fit individual dose-1012

response curves (Eq. 1, here we set 𝑦𝑚𝑎𝑥 = 1 for simplicity), assume that (i) sensitivities follow 1013

a power law distribution 𝑃(1/𝐸𝐶50) ∝ (1/𝐸𝐶50)−𝜆−1 (or equivalently an exponential distribution 1014

for 𝑘 = ln (1/𝐸𝐶50): 𝑃(𝑘) = 𝜆e−𝜆(𝑘−𝑘0), 𝑘 ≥ 𝑘0) (ii) the Hill coefficient 𝑛 for all odorant-ORN pairs 1015

are the same and greater than 𝜆 (satisfied in the data as 1.42 vs. 0.42). If so, the population 1016

response follows an approximate power law form 𝑟(𝑐) ∝ 𝑐𝜆 for concentrations 𝑐 ≤ 𝑒−𝑘0 (which 1017

means the weakest response pair in the population has not reached the half level). For 1018

convenience, we use the log scale of concentration and 𝐸𝐶50: 𝑥 = ln(𝑐) , 𝑘 = ln (1/𝐸𝐶50) and the 1019

logistic function in place of the Hill equation: 𝑦(𝑥) =1

1+exp (−𝑛(𝑥+𝑘)). 1020

1021

This result can be intuitively obtained by considering the limiting case where the logistic function 1022

is infinitely steep (large Hill coefficient) and is thus replaced by a step function. The population 1023

response combining a large number of odorant-ORN pairs can be expressed as an integral: 1024

𝑟(𝑥) = ∫ 𝑦(𝑥, 𝑘)𝜆𝑒−𝜆(𝑘−𝑘0)𝑑𝑘∞

𝑘0, 𝑦(𝑥, 𝑘) =

1

1+exp (−𝑛(𝑥+𝑘)) is the log-concentration. When 𝑦(𝑥, 𝑘) is 1025

a step function, the integral becomes 𝑃(𝑘 ≥ −𝑥), which is essentially the cumulative density 1026

function for 𝑘. Given the distribution of 𝑘, this is exactly an exponential function 𝑟(𝑥) = 𝑒𝜆(𝑥+𝑘0) 1027

(or a power law function of 𝑐) for 𝑥 ≤ −𝑘0, and saturates at larger concentrations. 1028

1029

For the general case of logistic activation, the integral does not have a simple form expression 1030

but involves hyper-geometric functions. However, we can derive a simple closed form 1031

approximation by approximating the logistic function 𝑓(𝑥) = 1/(1 + 𝑒−𝑛𝑥) using piecewise 1032

exponential functions: 1033

𝑓(𝑥) ≈

{

e𝑛𝑥 −e2𝑛𝑥

2, 𝑥 ≤ 0

1 − e−𝑛𝑥 +𝑒−2𝑛𝑥

2, 𝑥 > 0

1034

Such an approximation becomes asymptotically exact when the steepness 𝑛 goes to infinity, or 1035

when the absolute value of 𝑥 goes to infinity. Substituting 𝑦(𝑥, 𝑘) with this approximation, the 1036

43

integral splits into segments, over which the integrand are sums of exponential functions, and 1037

therefore can be easily integrated. This gives the closed form approximation of 𝑟(𝑥): 1038

𝑟(𝑥) = {(1 +

2𝜆2

𝑛2−𝜆2−

𝜆2

4𝑛2−𝜆2) 𝑒𝜆(𝑥+𝑘0) −

𝜆

𝑛−𝜆𝑒𝑛(𝑥+𝑘0) +

𝜆

2(2𝑛−𝜆)𝑒2𝑛(𝑥+𝑘0), 𝑥 ≤ −𝑘0

1 −𝜆

𝑛+𝜆𝑒−𝑛(𝑥+𝑘0) +

𝜆

2(2𝑛+𝜆)𝑒−2𝑛(𝑥+𝑘0), 𝑥 > −𝑘0

1039

For small concentrations, 𝑥 ≤ −𝑘0, the leading term in the above expression is 𝑒𝜆(𝑥+𝑘0), since 1040

𝜆 < 𝑛. This explains that the population response is approximated by an exponential function 1041

with exponent 𝜆. Furthermore, the theory also predicts the magnitude (vertical shift in the log-log 1042

plot of population response as in Figure 3E), that is, 𝑟(𝑥) ≈ (1 +7

4

𝜆2

𝑛2)𝑒𝜆(𝑥+𝑘0), which explains 1043

how the Hill coefficient affect the population response. 1044

1045

Reverse-correlation analysis 1046

White noise experiments were performed in a manner similar to those described in (Kato et al., 1047

2014). Briefly, we used custom code written in MATLAB to control odorant and water switching 1048

such that it followed an m-sequence. Calcium imaging was performed on the axon terminal of 1049

individual ORNs at ~30 frames per second. Calibration and an example of such a recording is 1050

shown in Video S2. We then used a linear-nonlinear model to compare the m-sequence input to 1051

ORN responses during a 150 second interval (from 60 - 210 sec). An 18 second time window 1052

was used for the linear filter, of which 15 seconds represented stimulus history in order to 1053

ensure extraction of the full filter dynamics (Figure 5B). The raw filter amplitudes across 1054

individual animals varied, possibly due to variations in GCaMP6m expression levels or imaging 1055

quality, but the shape of normalized filters was comparable across animals (Figure S5A). Next, 1056

we applied the linear filter to the data and compared this to the output in order to capture the 1057

nonlinear function. We found that a sigmoidal function fits the nonlinear function well (Figure 1058

S5B). We then applied novel m-sequences on the same animal to validate the linear-nonlinear 1059

model (Figure S5C) and found that they fit the data well. Peak and decay times for each filter 1060

were found by extracting the time points corresponding to the maximum amplitude and half 1061

maximum amplitude of the decay phase, respectively. 454 filters were calculated from the 1062

recording of 138 larvae responding to various m-sequence stimuli. Each of the 31 filters 1063

quantified in Figures 5F-5G and S10, are averaged across 10 animals. 1064

1065

Analysis of previous electrophysiology studies on odorant-ORN responses 1066

To assess the concordance between electrophysiology and calcium imaging techniques, we 1067

applied our analysis to Drosophila ORN electrophysiology data previously described in the 1068

44

literature. We used the following two data sets for comparison, 1) Figure 7B from (Kreher et al., 1069

2008) showing Or42a and Or42b responses to ethyl acetate across seven orders of magnitude 1070

in concentration, 2) Figure 1B from (Kreher et al., 2008) showing ORN responses to 26 1071

odorants. 1072

We fit the first data set of ORN dose-response curves to Hill equations in which all parameters 1073

were free, as this electrophysiology data was densely sampled across concentration (Figure 1074

S4H). The second data set was used to compare correlations in the odorant-ORN response 1075

matrices and thus we applied the simulated annealing method described above, to order ORNs 1076

and odorants for comparison with our data (Figure S4I). 1077

1078

Data and Software Availability 1079

Data, code and software are available at https://github.com/samuellab/Larval-ORN. 1080

45

Supplemental Figures 1081

1082

46

Figure S1, related to Figure 1: Microfluidic device designs, operation, calibration and 1083

stimulus sequence evaluation. 1084

(A) Microfluidic device designs with 8, 16, or 24 channels for odorant stimulus delivery. 1085

(B) Change in fluorescence intensity during delivery of 5 second step pulses of increasing 1086

concentrations of fluorescein dye, each followed by 5 seconds of water. Inset shows zoom-in of 1087

dashed box, indicating stimulus transition time of ~20 ms. 1088

(C) Combination of on/off valve states required to generate the stimulus sequence highlighted in 1089

the shaded region of panel B. Rows indicate the stimulus type that a larva would experience. 1 1090

and 0 indicate whether the valve is open or closed, respectively. CW represents a water channel, 1091

CC1 and CC2 represent control channels 1 and 2 that allow stimulus switching, C7 and C8 1092

represent odorant delivery channels which open prior to and during stimulus delivery. 1093

(D) Images of fluorescein dye, representing an odorant stimulus, in the microfluidic device 1094

during each state shown in panel C (water, stimulus 7, stimulus 8). White cross indicates closed 1095

channels, star marks the location of a larva’s ORN dendrites. Scale bar = 300 µm. 1096

(E) Or35a-ORN responses to 0.5, 1, 2, 5, and 10 second pulse stimuli to a 10-5 dilution of the 3-1097

octanol odorant. The maximum ORN response saturates when the odorant pulse is longer than 1098

5 seconds. 1099

(F) Or35a-ORN responses to increasing (top panel), primarily decreasing (middle panel), and 1100

random (bottom panel) concentration sequences of 3-octanol pulses, delivered at 5 seconds 1101

each. The response amplitude to each concentration level is independent of stimulus history. 1102

47

1103 Figure S2, related to Figure 1: Anatomical map of ORN dendritic organization. 1104

(A) Dendritic bundle location of each ORN in the larva. Larvae expressing OrX>GFP; 1105

Orco>RFP, where OrX is a specific olfactory receptor, were used to identify each ORN’s 1106

dendritic bundle in green, while all ORNs were visible in red. We infer the vacancy in bundle 2 1107

as the Or33a-ORN. No expression of Or2a and Or7a were observed in first instar larvae. 1108

(B) Summary schematic of stereotyped ORN position in each dendritic bundle for right dorsal 1109

organ. 1110

1111

48

1112

49

Figure S3, related to Figure 2: Odorant panel selection and ORN peak and dynamic 1113

responses. 1114

(A) PCA of all odorants listed in the DoOR database (a collection of olfactory studies previously 1115

described in the Drosophila literature), where each odorant was described by the 32 most 1116

relevant molecular descriptors identified in Haddad et al., 2008. Gray dots correspond to the 1117

690 monomolecular odorants catalogued in the DoOR database. Red dots correspond to the 35 1118

odorants selected for this study. Distributions and breadth of selected odorants (red dots) 1119

closely match those previously used in the Drosophila literature (grey dots). 1120

(B) Raw calcium activity traces for each of 21 ORNs in response to stimulus pulses of five 1121

concentrations of the odorant myrtenal followed by a panel of odorants that were used to 1122

identify each ORN. 1123

(C) Heatmap of peak responses of four ORNs to four alcohol odorants, across four 1124

concentrations of each odorant. Neural images label the four ORNs in the dorsal organ ganglion 1125

during calcium imaging at the highest odorant concentrations. 1126

(D) PCA of ORN population response over a 10~25 second period (during the 5 seconds of 1127

stimulus delivery and 5~20 seconds after stimulus offset) of two odorants, benzaldehyde and 1128

ethyl butyrate. Points are connected in temporal order (indicated by the arrows), forming distinct 1129

trajectories for each odorant. Dashed circle marks baseline starting position at onset of stimulus 1130

delivery. 1131

50

1132

51

Figure S4, related to Figure 3: Hill curve fits to ORN dose-response data. 1133

(A) Dose-response data for each of the 36 odorant-ORN pairs that reached saturation within the 1134

tested concentration range. All responses are fit to a Hill function with a common Hill coefficient. 1135

(B) Dose-response curves for Or13a and Or22c ORNs, each in response to two different 1136

odorants (n>=10). 1137

(C) Comparison of the maximum response amplitude (𝑦𝑚𝑎𝑥) for each ORN to two different 1138

odorants. Each point indicates a separate animal. Colors of each point correspond to legend in 1139

panel B. 1140

(D) Scatter plot of actual versus predicted response for each individual animal with a non-zero 1141

response value in Figure 2A. Dashed line indicates 𝑦 = 𝑥. 𝑅2 of the linear fit is 0.99. 1142

(E-G) Distribution histograms of fitted amplitude, 𝑦𝑚𝑎𝑥 (E), fitted Hill coefficient (F), and fitted 1143

variation in log sensitivity (𝑘 = 𝑙𝑜𝑔10(1/𝐸𝐶50)) (G), across all individuals. 1144

(H) Plot of Or42a and Or42b ORN dose-response electrophysiology data from Kreher et al., 1145

2008. Data from each ORN are fit to a Hill function with a Hill coefficient of 0.72 and 0.71 for 1146

Or42a and Or42b, respectively. 1147

(I) Raw data of ORN firing rates from Kreher et al., 2008 reordered according to the simulated 1148

annealing method used to arrange Figure 2A (see Methods). Color of odorant name indicates 1149

type of functional group it contains (pink, organic acid; light green, terpene; gray, aldehyde; light 1150

orange, ketone; light blue, aromatic; red, alcohol; dark green, ester). 1151

52

1152 Figure S5, related to Figure 5: Characterization of ORN temporal response properties. 1153

(A) Or42a-ORN responses (black curve) to a m-sequence of 10-7 dilution of 3-pentanol (shown 1154

in red). Each of seven traces (shown in black) is from a different animal. The response and 1155

temporal filter amplitudes vary from animal to animal (temporal filters shown to the left of each 1156

animal’s response). Magenta traces show predicted response when the first animal’s temporal 1157

filter and nonlinear function were rescaled to account for amplitude differences and then applied 1158

across all animals. 1159

(B) Non-linear transfer function calculated by comparing measured and predicted responses 1160

using the linear filter of the first trial in panel A. Red curve follows the function of 𝑦 =𝑎

1+𝑒−𝑏(𝑥−𝑐)+1161

𝑑, where 𝑎 = 2.7, 𝑏 = 1.8, 𝑐 = 0.68, 𝑑 = 0.66 . 1162

(C) Validation of the linear-nonlinear (LN) model by comparing predicted and measured 1163

responses to a novel m-sequence stimulus (generated using a different random seed from 1164

panel A. Example of a measured response (black) to the novel m-sequence stimulus (red). 1165

53

Magenta trace shows the LN model predicted response using parameters calculated from the 1166

same animal’s response to the m-sequence stimulus shown in panel A. 1167

(D) Normalized filters of three ORNs (Or1a, Or67b, and Or35a) responding to various odorant 1168

stimuli. 1169

54

Odorant Functional Group(s)

Component of Fruit or

Plant

Behavioral studies in larvae

Physiology studies

acetal acetal fruit Attractive (Mathew et al., 2013)

Mathew et al., 2013

nonane alkane plant oils, tomato

DoOR 2.0

ethyl acetate ester fruit Attractive (Kreher et al, 2008)

DoOR 2.0

3-pentanol alcohol fruit

geranyl acetate ester, prenyl leaf and plant oil

Aversive (Kreher et al, 2008)

Mathew et al., 2013, DoOR 2.0

4-hexen-3-one enone (alkene + ketone)

fruit Attractive (Mathew et al., 2013)

Mathew et al., 2013

ethyl butyrate ester fruit Attractive (Kreher et al, 2008)

Asahina et al., 2009, DoOR 2.0

trans,trans-2,4-nonadienal

enal (alkene + aldehyde)

fruit Attractive (Mathew et al., 2013)

Mathew et al., 2013

isoamyl acetate ester fruit Attractive (Kreher et al, 2008)

Schubert et al., 2014

2-nonanone ketone leaf and seed oil

Attractive (Mathew et al., 2013)

Mathew et al., 2013

butyl acetate ester fruit

Attractive (personal communication Katrin Vogt)

Hoare et al., 2011, DoOR 2.0

hexyl acetate ester fruit Chen et al, 2014, DoOR 2.0

2-heptanone ketone fruit Attractive (Kreher et al, 2008)

Oppliger et al.2000, DoOR 2.0

3-octanol alcohol fruit and plants

Attractive (Kreher et al, 2008)

Mathew et al., 2013, DoOR 2.0

6-methyl-5-hepten-2-ol

alcohol, alkene

fruit Attractive (Mathew et al., 2013)

Mathew et al., 2013

pentyl acetate ester fruit Attractive (Kreher et al, 2008)

Mathew et al., 2013, DoOR 2.0

trans-3-hexen-1-ol alcohol, alkene

leaf oil Attractive (Mathew et al., 2013)

Mathew et al., 2013

1-pentanol alcohol fruit Attractive (Mathew et al., 2013)

Mathew et al., 2013, DoOR 2.0

linalool alcohol, prenyl

fruit and leaf oil

DoOR 2.0

4-methylcyclohexanol

alcohol Oppliger et al.2000, DoOR 2.0

methyl salicylate ester, phenol

leaf odor Aversive (Kreher et al, 2008)

DoOR 2.0

55

benzyl acetate ester, phenyl

fruit

Attractive (personal communication Katrin Vogt)

DoOR 2.0

4-phenyl-2-butanol alcohol fruit DoOR 2.0

benzaldehyde aldehyde plant oils Aversive (Kreher et al, 2008)

Chen et al, 2014, DoOR 2.0

methyl phenyl sulfide

thioether, phenyl

coffee Attractive (Mathew et al., 2013)

Mathew et al., 2013

2,5-dimethylpyrazine

pyrazine coffee Attractive (Mathew et al., 2013)

Mathew et al., 2013, DoOR 2.0

2-acetylpyridine pyridine, ketone

coffee, tea leaf

Attractive (Mathew et al., 2013)

Mathew et al., 2013, DoOR 2.0

anisole ester, phenyl

not in fruit, in plant seed

Attractive (Kreher et al, 2008)

Mathew et al., 2013

4-methyl-5-vinylthiazole

thiazole, alkene

coffee and fruit

Attractive (Mathew et al., 2013)

Mathew et al., 2013

4,5-dimethylthiazole thiazole coffee and fruit

Attractive (Mathew et al., 2013)

Mathew et al., 2013, DoOR 2.0

(1R)-(-)-myrtenal enal plant oil and citrus fruits

DoOR 2.0

2-methoxyphenyl acetate

ester, ether, phenyl

not in fruit or plants

Attractive (Mathew et al., 2013)

Mathew et al., 2013

2-phenyl ethanol alcohol, phenyl

fruit

menthol alcohol plant oil Aversive (personal communication with Katrin Vogt)

DoOR 2.0

pentanoic (valeric) acid

carboxylic acid

fruit DoOR 2.0

Table S1, related to Figure 2: Description of odorants. 1170 List of 35 odorants used in this study, followed by their molecular functional groups, presence in 1171

the natural environment, and examples of their use in the previous literature to study behavior 1172

and physiology in the Drosophila larva. 1173

56

1174 1175 Table S2, related to Figure 3: Comparison of sensitivity fit using different distributions. 1176

Comparison of fitting ORN population sensitivities using power law versus other common 1177

heavy-tailed distributions. Statistically significant p-values are denoted in bold. Other heavy-1178

tailed distributions are not significantly better than the power law. 1179

1180

57

Supplemental Video and Data 1181

Video S1. Dose-dependent activation of ORNs. 1182

Left: Calcium imaging of 21 pairs of larval ORNs in response to increasing concentrations of the 1183

1-pentanol odorant, from 10-7 to 10-3 dilutions. Video starts by scanning through the imaging 1184

volume to identify all ORNs activated by the odorant. Three ORNs are responsive on the left 1185

side and five on the right side. ORN identity was confirmed from the dendrite location and 1186

response to panel of 15 odorants (not shown in the Video). Right: Fluorescence traces of ORN 1187

responses to step pulses of odorant stimuli. Related to Figure 1. 1188

1189

Video S2. ORN responses to pseudorandom white noise stimulus. 1190

Top left, stimulus delivery marked by fluorescence. Bottom left, axon terminal of Or45a-ORN 1191

responding to the white noise stimulus using 10-7 dilution of 2-nonanone. Top and bottom right, 1192

real time plots of the input stimulus and ORN response during an experiment. Related to Figure 1193

5. 1194

1195

Data S1. Raw activity data of 21 ORNs responding to 34 odorants at multiple concentration 1196

levels collected from 238 recordings. Related to Figure 2. 1197