Calcium-activated chloride channels clamp odor-evoked spike activity in olfactory receptor neurons

The calcium-activated chloride channel anoctamin-2 (Ano2) is thought to amplify transduction currents in olfactory receptor neurons (ORNs), a hypothesis supported by previous studies in dissociated neurons from Ano2−/− mice. Paradoxically, despite a reduction in transduction currents in Ano2−/− ORNs, their spike output for odor stimuli may be higher. We examined the role of Ano2 in ORNs in their native environment in freely breathing mice by imaging activity in ORN axons as they arrive in the olfactory bulb glomeruli. Odor-evoked responses in ORN axons of Ano2−/− animals were consistently larger for a variety of odorants and concentrations. In an open arena, Ano2−/− animals took longer to approach a localized odor source than Ano2+/+ animals, revealing clear olfactory behavioral deficits. Our studies provide the first in vivo evidence toward an alternative or additional role for Ano2 in the olfactory transduction cascade, where it may serve as a feedback mechanism to clamp ORN spike output.

Our data provide evidence that loss of Ano2 results in enhanced ORN input to OB glomeruli. We also observed that the number of glomeruli responding to an odor was similar in Ano2 +/+ and Ano2 −/− animals. These results are consistent with a scaling mechanism, where Ano2 may function as a negative feedback on ORN excitability and limit the number of action potentials generated in response to odor stimulation.

Loss of Ano2 does not impact respiration.
It is possible that the larger responses to odorants in Ano2 −/− animals is due to faster respiration rates and temporal summation of responses 29 . Other ANO isoforms are expressed in smooth muscle tissue and may regulate the excitability of the diaphragm and the airway [30][31][32][33] , thereby altering normal breathing rhythms in Ano2 −/− animals. We recorded the respiration rate of Ano2 +/+ , Ano2 −/− , and C57BL/6 J mice using a thermocouple placed in front of the animal's nose, under anesthesia conditions consistent with our previous experiments. We first validated the reliability of the external thermocouple for respiration tracking by comparing it to a well-established method, intranasal cannulas (Supplemental Fig. 2). Upon validation, we chose to record respiration using the non-invasive thermocouple to mitigate any effects of damaging the nasal cavity through cannula implantation.
We conclude that anesthetized Ano2 −/− animals do not breathe with increased frequency and that the larger Ca 2+ signals we observed are not due to enhanced respiration rate.
Multiphoton imaging in Ano2 −/− animals. Due to the low resting fluorescence of GCaMP3 we were unable to identify glomeruli that did not respond to at least one of the seven odors in our panel using an epifluorescence microscope. We used multiphoton microscopy to overcome this limitation and were able to visualize all glomeruli, independent of their responsiveness (Fig. 4A). We also expanded our odor panel size to 15 odors to activate a wider range of glomeruli. The optical sectioning facilitated by multiphoton microscopy also allowed us to exclude any effects arising from the activity of en passant axons that could be detected using our epifluorescence setup. As a result of significant out-of-focus fluorescence in wide-field imaging, signal contamination could arise from Ca 2+ activity in axons that pass above inactive glomeruli.
We next compared the kinetics of the Ca 2+ responses in Ano2 +/+ and Ano2 −/− animals. Due to the improved signal to noise ratio and optical sectioning of multiphoton microscopy, we were able to compare the responses of individual trials rather than the means of replicates from the same glomerulus. This procedure reduced the influence of respiration, which dictates the response onset. We first characterized the rise time. On average, Ano2 −/− animals responded with a slightly faster rise time than Ano2 +/+ animals (2.35 ± 0.05 s in Ano2 +/+ , n = 2614, and 2.25 ± 0.05 s in Ano2 −/− , n = 3393, p < 0.001 Kolmogorov-Smirnov test). We also found that the odor responses in Ano2 −/− animals decayed at a slightly faster rate than Ano2 +/+ animals (decay time constant = 4.74 ± 0.08 s in Ano2 +/+ , n = 2376, and 4.51 ± 0.06 s in Ano2 −/− , n = 3155, p = 0.004 Kolmogorov-Smirnov test). Our data  indicate that Ano2 deletion may result in spike generation in ORNs within a narrower window following stimulus delivery, resulting in both a quicker rise and decay in Ca 2+ responses. However, the slow timescale of GCaMP3 and our population imaging approach leave open the possibility that more dramatic differences may be observed on the single cell level. Lastly, we compared sparsity of glomerular responses. We calculated the population sparseness (see Methods) to compare the fraction of activated glomeruli across all animals for each odor. Population sparseness is related to the fraction of glomeruli that are activated by a given odor stimulus, with values near one indicating uniform activity across all glomeruli and values near zero indicating highly selective responses across all observed glomeruli. We found that a larger fraction of glomeruli responded in Ano2 −/− animals (population sparseness measure: 0.09 ± 0.02 in Ano2 +/+ and 0.13 ± 0.02 in Ano2 −/− , p = 0.002, Wilcoxon sign-rank test, Fig. 4G). We found no differences between Ano2 +/+ and Ano2 −/− animals in lifetime sparseness, which quantifies the extent to which a given glomerulus responds to different odor stimuli (lifetime sparseness measure: 0.28 ± 0.01 in Ano2 +/+ and 0.28 ± 0.01 in Ano2 −/− , p = 0.89, Wilcoxon rank-sum test, Fig. 4H). If all odors activate the observed glomerulus uniformly, the lifetime sparseness measure will be close to one, and if a glomerular response is specific to only a small number of odors, the measure will be close to zero. Together these results indicate that ORNs in Ano2 −/− animals are indeed more sensitive to odor stimulation, but the breadth of their odor tuning is unchanged. Furthermore, the fact that we observed no difference in odor tuning further argues against the possibility that glomeruli in Ano2 −/− animals receive heterogeneous innervation from multiple ORN subtypes. We also found no evidence for heterogeneous responses within individual glomerular regions of interest.
ORNs are more strongly excited by odors across a range of concentrations. What accounts for the larger fraction of activated glomeruli in Ano2 −/− animals? One possibility is that the signal-to-noise ratio afforded by multiphoton microscopy allowed us to identify weak responses arising from odor-receptor binding in Ano2 −/− animals that are sub-threshold for Ca 2+ signal generation in Ano2 +/+ animals. Conversely, given our previous results, another explanation for the increased Ca 2+ signal magnitude in Ano2 −/− animals is that in response to high odor concentrations, ORNs are able to maintain firing due to a reduction in depolarization induced Na + inactivation driven by the amplifying current through Ano2. However, at low odor concentrations, ORNs in Ano2 −/− animals may have weaker responses than ORNs in Ano2 +/+ animals, since it has been shown that current amplification through Ano2 is most potent close to detection threshold 24 . We next investigated whether ORNs in Ano2 −/− animals are more responsive to odors at different concentrations. We used air dilution to alter odor concentration over four orders of magnitude for two odors, Ethyl valerate and Allyl butyrate, and decreased the odor delivery time to two seconds to prevent saturation of ORN responses at the highest concentrations. The relative concentration of each odor experienced by the animal was verified using a photoionization detector and odor concentrations were normalized to the lowest dilution (see Methods; Supplemental Fig. 3). At the strongest odor concentrations, glomerular ORN responses were again enhanced in Ano2 −/− animals for both odors, as well as a mixture of the two ( Fig. 5A-D, Wilcoxon rank-sum test). However, somewhat surprisingly, at low concentrations our analysis revealed no pair-wise differences. For low concentrations, the largest responses typically occurred in Ano2 −/− animals, but the vast majority of response were typically within the range of those observed in Ano2 +/+ animals, thereby limiting our ability to statistically differentiate the two groups (Supplemental Fig. 4). We were unable to compare the response amplitude of ORNs at their individual detection thresholds due to an inability to identify specific subtypes of ORNs (expressing a particular odorant receptor) across animals. We also note that Ca 2+ signals in ORN axons report spike output rather than transduction current amplitudes. Therefore, our results indicate that while Ano2 may play a role in shunting ORN excitation following strong odor stimulation, it may have less of an effect on ORN activity following weak odor input. Ano2 deletion increases latency to odor localization. Does increased ORN excitability alter odor detection capabilities of Ano2 −/− animals? Recent studies provide evidence that Ano2 −/− animals exhibit a greater latency to uncover a hidden food-object 9,10 , while another study was unable to find any difference in odor detection and discrimination using a learned behavior 8 . We decided to study innate odor-driven investigation to assess whether Ano2 −/− animals displayed any sensory deficit independent of learned behavior. We investigated the latency to explore odors as an indicator of how easily animals can detect odors 9,10 . To minimize experimenter-induced biases, our automated experimental apparatus consisted of a 56 cm diameter circular arena with four air inlets equally spaced around its circumference, as well as a vacuum in its center to balance air inflow and outflow. Under infrared illumination, mice were allowed to explore the arena space for 10 minutes, after which odorized air was delivered through one of the inlets. We then measured the latency and path taken by each animal to investigate the odor source, as determined by the animal approaching the odorized air inlet within 1 cm (Fig. 6A).
We used the known appetitive odor peanut oil 35 diluted to the same concentration as we used in our imaging experiments (1% in mineral oil). Consistent with previous reports 9,10 , across 8 Ano2 +/+ and 10 Ano2 −/− animals, we found that Ano2 −/− animals required significantly more time to locate the odor source (Ano2 +/+ = 29.28 ± 5.75 s, Ano2 −/− = 94.71 ± 16.47 s, p = 0.003, Wilcoxon rank-sum test; Fig. 6B). Ano2 −/− animals also traveled a significantly longer distance before ultimately arriving at the odor source (Ano2 +/+ = 178.41 ± 28.46 cm, Ano2 −/− = 822.43 ± 175.95 cm, p = 0.002, Wilcoxon rank-sum test; Fig. 6C) Because the odor onset occurred independently of the animal location in the arena, we calculated the initial starting distance from the odor source and observed no differences in their mean initial positions (Ano2 +/+ = 30.71 ± 4.68 cm, Ano2 −/− = 37.78 ± 3.58 cm, p = 0.002 Fig. 6D, p > 0.05, Wilcoxon rank-sum test). At the same time, we observed no differences in the locomotor activity of Ano2 −/− animals as measured by their mean velocity both prior to and following odor delivery (Fig. 6E, p > 0.05, Wilcoxon rank-sum test). These behavioral data suggest a puzzling dissociation between the increased responses to odorants in Ano2 −/− animals and the longer latency to locate the source of an appetitive odor.

Discussion
Our study presents direct evidence in freely-breathing mice that Ano2, despite its role in amplifying transduction currents in ORNs, limits their overall excitation and input to the OB in vivo. Our results are in agreement with recent in vitro measurements of ORN spike output in Ano2 −/− animals 9 and further point towards a dual functionality of Ano2 in ORN excitability whereby it both amplifies transduction currents and limits spike output. Glomerular maps and respiration. Loss of Ano2 in mice could lead to more general changes that might be confounding factors that undermine conclusions about sensory transduction and coding. First, absence of Ano2 might alter the anatomical organization of glomerular maps in the OB. In particular, spontaneous activity in ORNs is known to play an important role in ORN fasciculation and glomerular emergence [21][22][23] . Any differences in spontaneous activity between Ano2 +/+ and Ano2 −/− animals might lead to disorganized glomerular organization and odor representation. Our results argue against a broad topographical reorganization of ORN inputs to the OB in Ano2 −/− animals based on two factors. First, we found that the number of dorsal glomeruli responding above threshold to a given odor was unchanged and second, we found no difference in the lifetime sparseness of individual glomeruli from Ano2 +/+ and Ano2 −/− animals. Furthermore, glomeruli in Ano2 −/− animals do not appear to receive heterogeneous ORN innervation since responding glomeruli were invariably homogeneous. Our in vivo imaging methods restricted us to imaging the dorsal portion of the OB, and we cannot rule out the possibility that ORN targeting is disrupted in the medial, lateral or ventral OB in Ano2 −/− mice. A second factor that might affect the data on glomerular imaging is the respiration rate. Faster respiration may lead to larger Ca 2+ signals because of slow time course of axonal Ca 2+ as well as indicator kinetics. Direct measurement of respiration, however, dispelled this concern -we found no significant change in respiration rate in Ano2 −/− animals. On a methodological note, we also demonstrated that an externally-placed, non-invasive thermocouple is a reliable method to record and measure breathing responses in anesthetized mice. While this method could be valuable for experiments in anesthetized animals, we note that it rapidly loses fidelity when breathing rate increases, as in awake animals (Supplemental Fig. 2E,F).
Larger odor-evoked responses in Ano2 −/− animals. The major finding of our study is that the magnitude of the ORN Ca 2+ responses following odor stimulation was larger in Ano2 −/− animals, with no observable change in the overall response duration. This was confirmed in two different modes of imaging -widefield microscopy that allowed larger regions to be imaged at lower resolution, and multiphoton microscopy that offered excellent optical sectioning and signal-to-noise ratio. We systematically varied the concentration of two different odorants and found that responses in ORNs from Ano2 −/− animals were consistently larger at most concentrations. Interestingly, at lower concentrations of the two odors, the response amplitudes were similar between both groups. This result suggests that for low odor concentrations, ORN transduction currents may remain sufficiently modest, and limits further amplification through Ano2. In such a scenario, ORN transduction currents may be primarily carried through cyclic nucleotide-gated channels upstream of Ano2, thereby generating small membrane depolarizations without engaging significant Ano2-mediated amplification. Biophysical studies in vitro, however, indicate that Ano2 currents are activated even for weak stimuli 10,36 . In vitro preparations allow for titration of odor concentrations for each neuron, thus allowing careful analysis of transduction currents in different response regimes, including threshold and sub-threshold responses. Perhaps glomerular imaging does not have enough sensitivity to detect responses to low concentrations of odors, and potential differences between Ano2 +/+ and Ano2 −/− animals were missed. In vivo, threshold odor concentrations are likely to activate only a subset ORNs due to their dispersion throughout the nasal epithelium 37 . The population imaging approach used in our studies further decreases the likelihood of detecting low-level signals because any responses arising a small number of ORNs are averaged across all ORNs that terminate at a given glomerulus. An alternative approach may include sparsely labeling ORNs with Ca 2+ indicators to allow for recordings of individual optically isolated axon terminals; however, to date there is no reliable method available for such an approach.
Another key result is that responses saturated at lower amplitudes in Ano2 +/+ than in Ano2 −/− glomeruli, suggesting that the presence of Ano2 had a "clamping" effect, and its absence loosens the clamp to allow greater activity. Our findings and work of others 9 , suggest that in a high odor concentration regime, Ano2 may function as a feedback mechanism to limit the number of spikes generated by ORNs following odor stimulation. The proposed mechanism of action operates through a potent depolarization-induced inactivation of Na + channels following transduction current amplification by Ano2 9 . A potential caveat is that the larger responses we observe in Ano2 −/− animals are the result of lower resting fluorescence due to a decrease in spontaneous ORN activity at some glomeruli 9 , thereby increasing the dynamic range available for Ca 2+ indicator activity. Our experiments used odor concentrations that are generally thought to be sub-saturating for odor evoked responses in ORNs; however, it remains possible that at these concentrations, the largest signals observed in Ano2 +/+ animals exceeded the range of our indicator due to a greater basal Ca 2+ tone. Our results here argue against this possibility since the only observable differences occurred in response to strong odor stimulation -larger responses in Ano2 −/− animals would not be observed if ORN responses in Ano2 +/+ were saturated at these higher concentrations.
It is also possible that Ano2 alters neural excitability in other ways, especially since Ano2 is expressed in ORN terminals 8,38 . For example, in the thalamocortical 39 system Ano2 functions to suppress neural excitability by enhancing the magnitude of action potential after-hyperpolarization. In the hippocampus 40 , Ano2 decreases the duration of individual action potentials by relying on a chloride gradient that favors outward membrane currents close to resting potential 41 . Although the chloride gradient in the nasal epithelium favors inward currents at near resting potentials 16,18 , the presumed chloride gradient at ORN axonal terminals could yield outward currents through Ano2, triggered by inward flux of Ca 2+ during action potentials. A reduction in ORN transduction currents at the olfactory cilia of Ano2 −/− animals could be offset by a reduction in the action potential after-hyperpolarization in the axonal compartments of ORNs.
Independent of the mechanisms involved, Ano2 seems to functionally compress the dynamic range of odor responses in individual ORNs by leaving weak responses unaffected (or enhancing them) and truncating the magnitude of responses to strong odors.

Role of Ano2 in olfactory coding.
Our data suggests that currents through Ano2 may serve as a negative feedback mechanism to prevent excessive activation at higher concentrations. It remains possible that at low to moderate concentrations, Ano2 may act to amplify sensory signals and affect activity in ways undetected by our measurements. For instance, the latency to first spike (from the onset of inhalation) could be shorter in Ano2 +/+ ORNs because of the amplification by Ano2. Such changes in timing could play an important role in odor coding [42][43][44] , but our imaging methods may not have the temporal resolution or sensitivity to detect such differences in latency. It is apparent that simply scaling up the activity in ORNs is insufficient enhance odor detection and may instead have deleterious effects.
Another potential role of Ano2 may arise from differences in expression of Ano2 in ORNs of a common subtype. Through varying expression levels, ORNs projecting to the same glomerulus could de-correlate their firing patterns in response to the same stimulus by shunting their spike output at different levels and thereby increasing their information carrying capacity as a population. Past studies demonstrate that intrinsic biophysical diversity between sister mitral cells functionally reduces correlations in their spike output 45 and these observations are consistent in other systems including ganglion cells 46 and M1 type ganglion cell photoreceptors 47 in the retina. In our mouse line, all ORNs were labeled with GCaMP3 and we were therefore unable to study heterogeneity at the single cell level. However, future studies may seek to record from a small number of ORNs projecting to the same glomerulus to determine whether Ano2 plays a role in de-correlating their spike output. Furthermore, a decreased information carrying capacity of ORNs in Ano2 −/− animals provides a potential explanation for the odor localization deficits observed in this study and others 9,10 . We observed that Ano2 −/− animals, despite having larger odor-evoked ORN responses, require longer to locate a relatively low-concentration odor source. While our result is in agreement with recent behavioral analyses of Ano2 −/− mice 9,10 , it stands in contrast to other reports that Ano2 −/− mice have no detectable deficits in odor discrimination 8 . Notably, our paradigm takes advantage of innate odor-seeking behavior in mice rather than task performance following learning. One possibility is that Ano2 −/− animals are able to overcome olfactory deficits through learning, thereby allowing for comparable performance when the experimental timescale is extended. Additional studies are necessary to explore potential differences in innate vs. learned olfactory behaviors, as well as any compensatory adaptations in Ano2 −/− mice.
Our result suggests that deletion of Ano2 does not simply scale up the sensitivity of ORNs, but rather, results in a fundamental reformatting of how odor information is transmitted to the brain. At present it is not clear how odor information is restructured in ORNs of Ano2 −/− animals, or if dysfunction in odor information processing is further compounded by downstream neurons.

Materials and Methods
Animal Care, General Statements. The Anoctamin-2 knock-out (Ano2 −/− ) mouse line was obtained from the PBmice project of Fudan University (see Li et al., for characterization) 24 . C57BL/6 J, Ano2 +/+ , Ano2 −/− , and OMP-GCaMP3 25 mice were used in this study. The age of all animals at the time of the experiments was two to six months. All mice used in this study were housed in an inverted 12-hour light cycle and fed ad libitum. All the experiments were performed in accordance with the guidelines set by the National Institutes of Health and approved by the Institutional Animal Care and Use Committee at Harvard University.
In vivo imaging. Surgery. Adult mice were anesthetized with an intraperitoneal injection of ketamine and xylazine (100 and 10 mg/kg, respectively) and eyes were covered with petroleum jelly. The scalp was shaved and opened. After thorough cleaning and drying, the exposed skull was gently scratched with a blade, and a titanium custom-made headplate was glued on the scratches. The cranial bones over the OBs were then removed using a 3 mm diameter biopsy punch (Integra Miltex). The surface of the brain was cleared of debris and a glass coverslip was glued into the vacated cavity in the skull. Dental cement (Jet Repair, Lang Dental) was used to cover the headplate and form a well around the cranial window. Animals were allowed to recover for at least three days. Prior to each imaging session, animals were anesthetized with a mixture of ketamine and xylazine (90% of dose used for surgery) and body temperature was maintained at 37 °C by a heating pad.
Epifluoresence. Two photo lenses coupled front to front were used to image the OB surface onto the sensor of a CMOS camera (DFK 23GPO31, The Imaging Source GmbH). Images (960 × 600 pixels) were acquired at 8-bit resolution and 8 frames/s. Data from the camera were recorded to the computer via data acquisition hardware (National instruments) and custom software in Labview. A blue LED (CBT-90, Luminus) with a maximum output of 1.65 mW/mm 2 was used for excitation.

Multiphoton.
A custom-built two-photon microscope was used for in vivo imaging. Fluorophores were excited and imaged with a water immersion objective (20× , 0.95 NA, Olympus) at 920 nm using a Ti:Sapphire laser (Mai Tai HP, Spectra-Physics). Images were acquired at 16-bit resolution and 4 frames/s. The pixel size was 1.218 μm, and fields of view were typically 365 × 365 μm. The point-spread function of the microscope was measured to be 0.51 × 0.48 × 2.12 μm. Image acquisition and scanning were controlled by custom-written software in Labview. Emitted light was routed through two dichroic mirrors (680dcxr, Chroma and FF555-Di02, Semrock) and collected by a photomultiplier tube (R3896, Hamamatsu) using filters in the 500-550 nm range (FF01-525/50, Semrock). For imaging glomerular responses to odor concentrations an additional 16 channel olfactometer outfitted with two odors, Ethyl valerate and Allyl butyrate, was used. The initial concentration series for each odor was 80%, 16%, 8%, 1.6%, 0.8%, 0.16%, 0.08% (v/v) in mineral oil and further diluted 16 times with air. Odors were presented for 2 s to prevent adaptation at the strongest concentrations. For all experiments, odors were delivered 3-5 times each.
Analysis. Calcium signals were extracted from raw images using custom-written scripts in MATLAB (MathWorks Inc.) and reported as ΔF/F signals, where F represents the average baseline fluorescence. Regions of interest were selected from average fluorescence projections for multiphoton imaging and ΔF/F projections for epifluorescence imaging. Response amplitude was measured from between three and five repeats of each odor as the mean response in the 5 seconds following odor onset. For analysis of response kinetics, measurements of the response rise time and decay time constant were taken from individual trials rather than reported as the mean to capture any intertribal variability. Rise time was measured as the time from when the signal first deviated 3.5 standard deviations from the mean of the baseline period to the peak of the response. Decay constants were obtained by fitting a single exponential to the signal, starting at the peak. Bleaching was corrected by fitting a single exponential to blank odor trials in multiphoton imaging and fitting a single exponential to the baseline period for epifluorescence experiments. For images of ΔF/F signals, the mean of an equal number of median filtered frames in the baseline and odor period was used. Traces of ΔF/F signals were smoothed for display. For figures where a threshold was applied to the data, thresholds were calculated based on the distribution of blank odor trials. An area under the receiver operating curve analysis was performed and the lowest threshold yielding ten responses for every one blank odor response was chosen. Sparseness measures were calculated as previously reported 48,49 . Population sparseness measures the fraction of glomeruli that are activated by a given odor, with Where: m = the number of odors, r j = response of glomerulus A to odor j. All statistical comparisons for imaging experiments were made as described in the text for each figure and values are given as mean + /− standard error of the mean.
Respiration measurements. Surgery. Animals were anesthetized with ketamine/xylazine as described above and a head plate was implanted in the skull as described previously in this article. For some mice, a small craniotomy was also made through the right nasal bone (1 mm anterior from the frontal/nasal fissure, 1 mm lateral from the midline), and a hollow cannula (#C313G; Plastics One Inc.) was lowered into the hole and glued to the skull. Finally, the whole exposed skull was covered with dental cement (Jet Repair, Lang Dental). The animals were given a week after the surgery to recover before any experiment was performed.
Respiration monitoring. Two strategies were used to monitor the breathing: measuring the intranasal pressure through an implanted cannula 50,51 , and measuring the temperature in front of the nose 52 .
For the intranasal pressure strategy, animals previously implanted with a cannula were head-fixed. Then, the cannula was connected to a pressure sensor (24PCEFJ6G; Honeywell International) through a piece of polyethylene tubing. The voltage signal generated by the sensor was amplified 1000×, low-pass filtered at 60 Hz, and digitized at 1000 Hz using custom software written in Labview.
For the temperature measurement, mice were head-fixed, and a thermocouple (5TC-TT-JI-40-1M, Omega Engineering) was placed ~2 mm in front of their nose. The voltage changes generated by the temperature variations were amplified 10000×, low-pass filtered at 60 Hz, and digitized at 1000 Hz using custom software written in Labview.
Analysis. Analysis of breathing signals and statistical tests were performed using custom software written in MATLAB. Two types of statistical tests were used: the Kruskal-Wallis test, and a bootstrap test to compare the means of two distributions (MATLAB function "bootstrp()" repeated 1,000,000 times for each bootstrapped statistics). The MATLAB toolbox CircStat was used to analyze circular data 53 . The critical p-value was set at 5% for all the tests, and Bonferroni correction was applied for multiple comparisons. On the figures, all the values are given as mean + /− standard error of the mean, unless otherwise stated.
Open Field Behavior. The arena consisted of 56 cm diameter circular inner chamber with four air inlets equally spaced around its circumference. The circular inner chamber was housed in light-and sound-proof outer chamber and illuminated using infrared LEDs. Throughout each experiment, airflow was maintained at a constant velocity for each inlet. After 10 minutes baseline exploration, air to one of the inlets was redirected through an odorized chamber while ensuring no change in its velocity. A vacuum was located at the center of the arena and its flow matched to the sum of all air inlets to prevent the accumulation of odor in the arena. Peanut oil was diluted in mineral oil as in imaging experiments. Each animal was only tested once and the order in which they were tested was randomized. After each trial the arena was thoroughly cleaned with ethanol to eliminate the presence of social cues. Images for tracking were acquired at 8 Hz using a USB camera (Grasshopper3, Point Grey Imaging) and custom-written Labview software. Images were processed using custom MATLAB routines to measure location and velocity. Animals with an initial position >10 cm from the odor source were excluded from our analysis. All statistical comparisons for behavior experiments were made with Wilcoxon rank-sum test and values are given as mean + /− standard error of the mean.
Data availability. All datasets in this manuscript are available from the corresponding author upon request.