Analyzing response patterns of olfactory bulb glomeruli at low odorant concentration regimes

Sensory input to the olfactory system begins with the binding of an odorant molecule to an olfactory receptor (OR) contained on the surface of an olfactory sensory neuron (OSN). Olfactory bulb (OB) glomeruli receive information from OSNs containing the same OR type and are thought to help process sensory input directed further upstream. Therefore, characterizing the responses and characteristics of ORs is essential to understand further odor processing. To explore these questions, the responses of dorsal mouse OB glomeruli to 185 odorants at near-threshold concentrations were recorded in vivo. Analysis of these data revealed that OR/OSN responses in these concentration regimes are extremely sparse and specific, and that ORs are narrowly tuned to one or a small range of chemically similar odors. In addition, certain glomeruli have been found to be identifiable across animals solely based on their approximate location and the odor to which they are most responsive. Glomeruli that are maximally sensitive to similar odors also cluster together across the dorsal OB surface, supporting the hypothesis of a chemotopic arrangement of glomeruli. We also characterized the response spectrum of the OR defined glomerulus olfr1377, which has yet to be characterized in vivo. Olfr1377 was found to be responsive to a range of acetophenone derivatives, including 4-methoxyacetophenone to which it was especially sensitive to at extremely low concentrations.

ABSTRACT Sensory input to the olfactory system begins with the binding of an odorant molecule to an olfactory receptor (OR) contained on the surface of an olfactory sensory neuron (OSN). Olfactory bulb (OB) glomeruli receive information from OSNs containing the same OR type and are thought to help process sensory input directed further upstream. Therefore, characterizing the responses and characteristics of ORs is essential to understand further odor processing. To explore these questions, the responses of dorsal mouse OB glomeruli to 185 odorants at near-threshold concentrations were recorded in vivo. Analysis of these data revealed that OR/OSN responses in these concentration regimes are extremely sparse and specific, and that ORs are narrowly tuned to one or a small range of chemically similar odors. In addition, certain glomeruli have been found to be identifiable across animals solely based on their approximate location and the odor to which they are most responsive. Glomeruli that are maximally sensitive to similar odors also cluster together across the dorsal OB surface, supporting the hypothesis of a chemotopic arrangement of glomeruli. We also characterized the response spectrum of the OR defined glomerulus olfr1377, which has yet to be characterized in vivo. Olfr1377 was found to be responsive to a range of acetophenone derivatives, including 4methoxyacetophenone to which it was especially sensitive to at extremely low concentrations. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 METHODS 8 RESULTS 13 DISCUSSION 29 ACKNOWLEDGEMENTS 35 REFERENCES 36 APPENDIX 39 iii 1 INTRODUCTION Overview of the olfactory system From an evolutionary standpoint, olfaction is perhaps the oldest phylogenetic sense, and the structure of the olfactory path is conserved in most animals. In mammals, olfaction begins with a sniff to carry odorant molecules upward towards the olfactory epithelium. The olfactory epithelium contains olfactory sensory neurons (OSNs), whose role is to detect and transmit odorant information to the central nervous system. Each OSN expresses only one of over 1000 different types of G-coupled protein receptors called olfactory receptors (ORs) (L. Buck and Axel 1991; Murthy 2011). As odorant molecules diffuse into the mucus, they bind to ORs on the surface of OSNs, initiating the sensory response. Because each OR has a specific response profile, chemical information from odorants can be represented within the olfactory epithelium as patterns of activity across different OSN/OR types. OSNs project a single dendrite from the olfactory epithelium to the olfactory bulb (OB) glomeruli. Glomeruli are spherical structures on the OB surface where OSN axon projections synapse with the dendrites of output and intrinsic OB neurons. Glomeruli receive input from OSNs, and are also thought to be processing units and play a role in shaping sensory input directed further upstream (Murthy 2011; Hildebrand and Shepherd 1997; Mori, Nagao, and Yoshihara 1999). OSNs expressing the same OR converge their axons to two glomeruli in each OB (one on the dorsal and one on the medial OB surface). Therefore, each individual glomerulus receives input from only one OR type (Ressler, 2 Sullivan, and Buck 1993; Vassar et al. 1994; Mombaerts et al. 1996). Consequently, measuring the neural response of a single glomerulus is synonymous to measuring the response of the associated OR. It is thought that olfactory information is encoded by a combinatorial pattern of glomerular activation (Malnic et al. 1999; Rubin and Katz 1999). Because glomeruli are topographically fixed in space, olfactory information can also be represented topographically through glomerular activation patterns. The spatial activity pattern of glomerular responses to diverse odorants is often referred to as an “odor map”. OSNs in glomeruli converge to provide excitatory inputs to mitral/tufted (M/T) cells. Each M/T cell receives its principal input from a single primary dendrite extended into one glomerulus. M/T cells also interact with inhibitory interneurons and granule cells. These connections are thought to provide a method for lateral inhibition (Chen, Xiong, and Shepherd 2000; Isaacson and Strowbridge 1998). The axons from mitral and tufted cells come together and form a structure called the olfactory tract, which runs on each side of the olfactory bulb. Odor information is then further transmitted to the olfactory cortex in which olfactory perception is constructed (Chen, Xiong, and Shepherd 2000; Mori et al. 2006). Characterizing ORs/glomeruli An olfactory response is initiated by the binding of an odorant molecule to an OR, so knowledge of OR response properties is crucial to understanding how sensory inputs are processed by the brain. Previous studies have found that ORs each have a unique response spectrum and respond to several different odorants. Likewise, each odorant is 3 recognized by distinct combinations of ORs (Murthy 2011). Because of this, it has been suggested that odorant information is encoded through a combinatorial code of different OR types. In addition, earlier research has found that there exist ORs that are both broadly and narrowly tuned, many of which respond to odorants with a common chemical quality (Nara et al. 2011). However there still remain many unanswered questions concerning OR specificity. In particular, it is not known how broadly or narrowly tuned ORs influence combinatorial coding. OR specificity has also been shown to vary with odorant concentrations. For example, ORs are often able to respond to multiple odorants at high odorant concentrations (thus displaying broader tuning specificity). Therefore, to understand the olfactory code, it is essential to compare the unique response spectrum of individual OR types, especially at a range of concentrations. By using transgenic and gene-targeting techniques it has been possible for earlier studies to describe OR response spectra via OR-defined glomeruli. ORs and OR-defined glomeruli are referred to by the name of the specific OR gene that is expressed in each. Previous research has characterized the response spectra of defined OR/glomerular models including those expressing the olfr151 (M71), olfr160 (M72), olfr2 (mI7), and olfr73 (mOR-EG) OR receptors. However, the number of characterized ORs is small relative to the large number of OR types, and very few of these have been characterized in vivo (J. Zhang et al. 2012; Bozza et al. 2002; Oka et al. 2006; Tan et al. 2010; Araneda, Kini, and Firestein 2000; Peterlin, Firestein, and Rogers 2014). Olfr1377 is another model OR which has been characterized in vitro by inducing OR expression in heterologous cells (Saito et al. 2009) but has not yet been characterized in vivo. Due to 4 differences between in vitro and in vivo OR responses, knowledge of the in vivo response spectrum of olfr1377 is necessary in order to fully characterize its response properties. One of the goals of this study is to measure olfr1377 responses to a range of odorants in vivo. Olfactory perception and odor space The relationship between odorant/chemical space and olfactory perception is complex, a fact that is often illustrated by comparing the olfactory system to vision or hearing. Namely, at the sensory receptor level, the olfactory system is much more complicated than each of these. To compare, auditory sensory information is organized along only one dimension: sound frequency. Within the cochlea, a high frequency produces a vibration closer to the oval window whereas a low frequency produces a vibration near the back. In this way, frequency is mapped spatially within the cochlea. In vision, each type of cone cell (of which humans have 3) responds to a broad but different range of light wavelengths. If more than one frequency is sensed, the brain interprets this as an intermediate value, producing intermediary colors (as opposed to auditory processing in which multiple frequencies are sensed independently) (Chittka and Brockmann 2005). In contrast, the olfactory system is immensely more complicated on the receptor level than that of both vision and hearing. In mammals, there exists over 1,000 types of OR proteins. Each OSN expresses only one OR type, indicating that there may be over 1,000 types of olfactory sensory neurons. (L. Buck and Axel 1991). Olfactory perceptual space could in theory contain as many dimensions as there are receptors, though this is likely 5 not the case. It is not obvious which chemical components are most represented in this new space; variations in odors are not as easily identified or mappable as that of sound frequency or light wavelength. One way to address this problem is to compare how odorant information is represented in both chemical/odor space and a perceptual space. The translation of odorant information between these two spaces may offer insight as to which odorant qualities are most significant when constructing an odor sensory response. Chemotopy of glomeruli Another important question is if olfactory stimulus information is organized spatially within the OB. If so, the structure of topological organization could hint as to which odorant properties are most represented within olfactory processing. Due to the diversity of odorants and the large number of ORs, this organization is not immediately obvious. One prevailing hypothesis is that glomeruli are organized “chemotopically” with respect to odorant physiochemical characteristics. However, this hypothesis is currently not resolved: various studies have arrived at conflicting conclusions (Bozza et al. 2004; Chae et al. 2019; Farahbod et al. 2006; Johnson et al. 2002; Ma et al. 2012; Meister and Bonhoeffer 2001; Soucy et al. 2009; Wachowiak and Cohen 2001). Because each glomerulus receives input from only one OR type, a chemotopic arrangement of glomeruli would indicate the presence of some type of interaction between OR type and glomerular positioning during development. Indeed it has been found that ORs and OSNs expressing a given OR are involved in OSN axon guidance and influence glomerular position (Mombaerts et al. 1996; Bozza et al. 2009). Therefore, 6 the degree to which the olfactory bulb is organized chemotopically is determined by OR axon guidance functions. Goals of this study Though many previous studies have characterized OR and glomerular responses, several key questions remain unknown. These include the breadth of OR tuning (especially at lower odorant concentrations), the impact of sparse OR activation on OB chemotopy, and the in-vivo response characterization of olfr1377. To explore these questions, we optimally activated glomeruli in-vivo by presenting odorants at just above responsethreshold concentrations. Because each OR is uniquely tuned, and OSNs expressing the same OR project to one or a few fixed glomeruli, each glomerulus is activated only by the odorants its associated OR is most sensitive to. Next, we further characterized glomerular responses by analyzing the sparseness patterns of this activation, in order to describe the sensitivity of OR activation to odorants in these low concentration regimes. We show that glomerular responses are extremely sparse to low odorant concentrations indicating that each glomerulus is associated with only a few odorants and each odorant is associated with only a few glomeruli. We also show that the sparse glomerular activation to odorants of which each is most sensitive supports the hypothesis that the OB is organized chemotopically. Finally, we characterized the response spectrum in-vivo of the OR-defined glomerulus olfr1377 through neural imaging in gene-targeted transgenic mice. 7 The majority of this thesis project is oriented around data analysis. Most of the initial experiments and data collection were performed by postdoctoral fellow Shawn Burton (however many of these procedures are still described in this report to provide context). My primary contribution to this project has been through the analysis of the aforementioned initial data, and in the experimental assistance and subsequent data analysis of more recent olfr1377 neural imaging experiments. 8 METHODS Transgenic mice and obtaining response trace signals For initial epifluorescence neural imaging experiments, mice expressing the calciumsensitive florescence tag GCaMP6s in all OB glomeruli were generated by crossing OMP-IRES-tTA mice with TetO-GCaMP6s expressing mice. Mice with olfr1377 expressing OSNs labeled were also used, but expressed the GCaMP6f reporter rather than GCaMP6s due to more consistent florescent labeling results in these genetic crosses. To label olfr1377 expressing OSNs, a strain of mice was created using gene targeting downstream of the olfr1377 coding sequence in which an internal ribosome entry site (IRES) and the coding sequence for the florescent marker mKate2 were inserted. Mice with the olfr1377 glomerulus tagged were produced by crossing OMP-Cre crossed with Rosa- GCaMP6f, then crossed with the gene targeted olfr1377 mice. These mice were used for further neural imaging experiments to screen odor concentrations 10x higher than those initially presented, and to characterize the response spectrum of olfr1377. Changes in neuron calcium-ion concentration is thus used as a measurement of neuronal activity, computed as the change in florescence. Continuous GCaMP6s and GCaMP6f signals were able to be obtained from OB glomeruli using both epifluorescence and twophoton imaging. Acute imaging in anesthetized mice For anesthesia, mice were injected intraperitoneally with pentobarbital (50 mg/kg), and then subcutaneously with atropine (0.5 mg/kg), and chlorprothixene (12.5 mg/kg). Mouse temperature, oxygen levels, and heart rate were monitored throughout the procedure. A 9 double tracheotomy was performed so that inhalation could be artificially controlled. Mice were allowed to breathe freely via the descending tracheal tube, while artificial inhalation was controlled via the ascending tracheal tube at a rate of 3 Hz throughout all neural imaging experiments. Anesthesia was sustained throughout each experiment by the delivery of ~0.5% isoflurane with pure oxygen via the descending tracheal tube. Mice were head-fixed and the skull above the OB was removed in order to allow access to the OB glomeruli. For widefield epifluorescence imaging, glomeruli were visualized using a 4x air-objective and for two-photon imaging, glomeruli were visualized using a 16x objective. Olfactometer and odorant preparation Odorants were prepared by diluting in triglyceride and administered using a novel olfactometer described in Burton et al, 2019. Odorants were chosen to encompass a wide range of chemical/odor space (Table A1). Odorant concentrations for initial OSN imaging ranged from ~0.1-1000 pM and the odor panel included a range of carboxylic acids, alcohols, aldehydes, esters, ketones, amines, aromatics, furans, pyrazines, thiazolines, thiols, terpenoids, and sulfides. Secondary OSN imaging used the same set of odors at concentrations ten times higher. The odorant panel used for characterization of the olfr1377 response spectrum included a range of aromatics, pyrazines, and ketones at ~0.1-1 nM concentrations (Table A2). 10 ROI selection and data set generation All data generation and analysis were preformed using MATLAB software. Data generation was done using previously developed MATLAB GUIs for use in the Wachowiak lab, while statistical and sparseness analysis was done using novel customwritten MATLAB code. Regions of interest (ROIs) were drawn around each glomerularsized region of activity from neural imaging experiments. ROIs were first selected using a semi-automated algorithm set to detect circular boundaries. Additional ROIs were then manually added as needed, and ROIs were manually checked to ensure accuracy of placement. ROIs size was chosen to fit well within the glomerular boundaries in order minimize overlap and bleed-over from nearby responses. Neural responses were calculated by subtracting the mean resting florescence from 1 to 0 seconds before odor delivery from the mean florescence from 2 to 3 seconds after odor delivery, and then dividing by the resting florescence (i.e., ΔF/F). Responses were calculated as the average ΔF/F value over three odorant trials. In the case of data in which the odorants were presented at 10x higher concentrations, the pre and post odor delivery time frame was 2 to 0 seconds and 0.5 to 4 seconds, respectively, and was thresholded using the z score of the pre-odor baseline signal at z=15. All other data was thresholded manually by visually inspecting ΔF/F response maps for responses. Sparseness measurements In this context, two distinct but related measurements of sparseness are used: lifetime and population sparseness. Here, lifetime sparseness referrers to the sparseness of a single glomerulus across many odorant presentations (i.e. the sparseness of the glomerulus’s 11 responses across its “lifetime”). Population sparseness measures the sparseness of the responses to many glomeruli to a single odorant (i.e. the sparseness of a “population” of glomeruli). The equations for calculating lifetime and population sparseness identical, differing only in the set of responses used for the calculation: For a set of responses to a single glomerulus across n odorants (R={r1, r2, … rn}) lifetime sparseness of the set is defined as follows: 𝐿# = 1 1 (1 − 𝑛) ∗ +1 − 𝑟 (∑2/34 - 𝑛/ 0)1 7 𝑟/1 2 ∑/34 5 6 𝑛 Similarly, for a set of glomerular responses to one particular odorant (G={g1, g2, … gn}) population sparseness of the set is defined as follows: 𝑃# = 1 1 (1 − 𝑛) ∗ +1 − 𝑔 (∑2/34 - 𝑛/ 0)1 𝑔1 ∑2/34 5 / 6 𝑛 7 In the case that there is only one non-zero response in a set of responses (i.e., perfect sparseness), sparseness will take on a value of 1. Suppose that we have a set of responses R={r1, 0, 0, …0n}. Then the sparseness of this set is calculated as follows: 𝑟41 1 1 1 1 1 ∗ +1 − 7= ∗ +1 − 𝑛1 7 = ∗ 51 − 6 = 1 1 1 1 1 𝑛 𝑟 𝑟4 (1 − 𝑛) -1 − 𝑛0 -1 − 𝑛0 ∑2/34 5 / 6 𝑛 𝑛 𝑟 (∑2/34 - 𝑛/ 0)1 12 However, in the case that each response in a set of responses is the same (i.e., perfect non-sparseness), sparseness will take on a value of 0. Suppose that for a set of responses R={r1, r2, … rn}, each ri=a. Then the sparseness of this set is calculated as follows: 1 1 (1 − 𝑛) ∗ +1 − 𝑟 (∑2/34 - 𝑛/ 0)1 𝑟1 ∑2/34 5 / 6 𝑛 7= 1 1 -1 − 𝑛0 ∗ :1 − 𝑎1 1 <= ∗0 =0 1 1 𝑎 -1 − 𝑛0 Therefore, sparseness (both lifetime and population sparseness) will range from 0 to 1 with a value of 0 indicating perfect non-sparseness and a value of 1 indicating perfect sparseness. Individual measurements of Ls across all glomeruli were used to calculate mean lifetime sparseness for each dataset. Likewise, individual measurements of Ps across all odorants were used to calculate mean population sparseness for each dataset. Glomerular tuning Average glomerular tuning curves were generated for both near-threshold odorant data and 10x higher odorant concentration data. First, for both near-threshold and 10x higher data, the response spectrum of each individual glomerulus was normalized by dividing by the highest response value, and then the individual responses were ranked from largest to smallest. An average glomerular response spectrum for both near-threshold and 10x higher data was then calculated by averaging responses across each rank (figure 8). 15 responsiveness and approximate location (Figure 3). To quantifiably identify these glomeruli, the response spectrum of the strongest-activated glomerulus for each odorant (for all odorants with no more than two glomeruli activated above a 50% dmax cutoff and with glomerular activation in at least 6 of 8 OBs) was compared to that of all other glomeruli across each OB. An ‘error ratio’ was calculated as the fraction of possible choices where the maximally activated glomerulus for each odorant is also not the max correlated glomerulus. 20 odorants (corresponding to 10 glomeruli) had an error ratio of zero, meaning that the strongest activated glomerulus also had the highest correlation 100% of the time (average median correlation coefficient of 0.962). Reducing this to an 80% match yields 27 identified glomeruli. The consistency in the response spectrum of each of these identified glomeruli suggests that each corresponds to the same OR across animals, and that these glomeruli can be consistently identified from their response to a single odorant. 20 Figure 6. Lifetime and population sparseness CDFs are shown for 8 OBs (4 mice). A majority of Ls and Ps values are above 0.95, indicating a high level of sparseness. 22 Figure 7. A comparison of cumulative lifetime and population sparseness CDFs to 185 odors at near-threshold concentrations and a subset of 159 odors at 10x near-threshold concentrations. At 10x higher odorant concentrations sparseness values are still high, with most Ls and Ps values above 0.9. To visually compare glomerular tuning at different odorant concentrations, average glomerular tuning curves were generated for both near-threshold odorant data and twophoton data at 10x higher odorant concentrations (Figure 8). These were generated by 24 The average narrow tuning of glomeruli suggests there may not be much overlap between the response spectrum of each individual glomerulus. This can be quantified by analyzing the dimensionality of these data: low glomerular response overlap should result in a highdimensional response space. To test this, principal component analysis (PCA) was conducted for glomerular responses to near-threshold odorants. Results show that it takes an average of 40-50 principal components to explain ~80% of data variance for each OB (Figure 9), with the first PC only accounting for >10% variance. Therefore, the glomerular response space to low odorant concentrations is high-dimensional, indicating that there is very little overlap between the individual response spectra of each particular glomerulus. Figure 9. PC vs cumulative percent variance explained. In all OMP datasets, a large number of PCs are needed to explain a majority of the data variance, suggesting that the underlying data is highly dimensional. 26 Characterization of the olfr1377 glomerulus While glomeruli tuning can be described generally using broad florescence labeling and imaging across the entire OB (as previously described), transgenic and gene-targeting techniques can be employed to identify and characterize the tuning properties of individual OR-defined glomeruli. Previous research has characterized the response spectra of many defined OR/glomerular models (J. Zhang et al. 2012; Bozza et al. 2002; Oka et al. 2006; Tan et al. 2010; Araneda, Kini, and Firestein 2000; Peterlin, Firestein, and Rogers 2014). However, while the spectrum of the OR-defined glomerulus olfr1377 has been characterized in-vitro it has yet to be characterized in vivo (Saito et al. 2009). Mice with the olfr1377 glomerulus tagged were produced by crossing OMP-Cre crossed with Rosa- GCaMP6f, then crossed with the gene targeted olfr1377 mice, in which all olfr1377 expressing OSNS are fluorescently marked. A panel of 48 odors was presented at a range of ~0.1-1 nM concentrations (with the exception of 4-methoxyacetophenone) (Table A2). Many of these odorants selected have previously been identified as being potential activators of olfr1377 in vitro (Matsunami, personal communication, 2020). Odorants selected included a range of aromatics, pyrazines, and ketones. It was found that at these concentrations, olfr1377 was maximally responsive to 4methyl-3-penten-2-one. It was also responsive to several aromatics including 4methylacetophenone, 2,4-dimethylacetophenone, acetophenone, 2-hydroxyacetophenone, 4-menthylanisole, and 2-methylacetophenone (figure 11). 27 Figure 11. Olfr1377 response spectrum to a range of aromatics, pyrazines and ketones presented at ~0.1-1 nM concentrations (with the exception of 4methoxyacetophenone, presented here at ~.04 pM). Olfr1377 was found to be extremely sensitive to 4-methoxyacetophenone, showing responses at molar concentrations smaller than ~0.04 pM. Suppressive behavior was observed in olfr1377 at higher concentrations of 4-methoxyacetophenone; at concentrations at or above ~40 pM, 4-methoxyacetophenone was observed to cause an initial large response followed by little to no responses for subsequent odorant presentations (4-methoxyacetophenone and otherwise) (figure 12). This suggests that 4methoxyacetophenone exhibits very tight binding to the olfr1377 receptor, which may prevent subsequent odorants from binding. 29 DISCUSSION To explore OSN sensitivity in low-concentration regimes, a panel of 185 odorants was administered at concentrations ranging from ~0.1-1000 pM. In general, this concentration range is much lower than those of previous studies characterizing OSN sensitivity (e.g. Davison and Katz at ~4e-7 M, Ma et al. at ~2.5e-4 M, Chae et al. at ~1e-4 M, and Soucy et al. at ~1e-5 M). Inspection of dorsal OB maps for each of the 185 odorants revealed that certain glomeruli can be identified across animals solely by their approximate position and the odor to which they are maximally responsive. Glomerular responses in these low concentration regimes are extremely sparse, with most lifetime sparseness (glomerular sparseness across odorants) and population sparseness (odorant sparseness across glomeruli) values above 0.9. This is true even for odorants administered at concentrations ten times higher. Therefore, high sparseness is not a product of picking near-threshold concentrations but is a general characteristic of responsiveness at low concentrations. Analysis of these sparse data sets revealed that they are also highdimensional, and that glomeruli are narrowly tuned to one or a few chemically related odorants. Glomeruli that are maximally responsive to chemically similar odors cluster together on the surface of the dorsal OB supporting the hypothesis of a chemotopic map of odorant sensitivity. Finally, the response spectrum of olfr1377 was characterized in vivo to a panel of 48 aromatic, pyrazines, and ketones administered at nanomolar concentrations. 30 Sparse activation and narrow tuning of olfactory receptors Previous literature suggests that there may exist both broadly and narrowly tuned OR receptors (Nara et al. 2011). However, these studies often test using odorant concentrations much higher than what might be physiologically relevant. OR responsiveness is often tested in vitro, which has been shown to provide somewhat different responses than in vivo. The concentration of odorants used for in vitro studies is often in the range of ~1 mM-100nM (Saito et al. 2009). Here, we evaluated the responsiveness of OR/OSNs in vivo to a broad panel of odors administered at a much lower concentration range of ~0.1-1000 pM. In these concentration regimes, our data show that OR responses are highly sparse and specific. The sparseness of each dataset was evaluated using both lifetime sparseness (Ls) and population sparseness (Ps). Ls and Ps is extremely high with the majority of values calculated to be above ~0.9. This is true even when calculated for responses to odorants administered at concentrations 10x higher (though comparison of these values may be limited by the fact that the secondary 10x responses were measured using GCaMP6f rather than GCaMP6s). Hence, OR/OSN responses at these low concentration regimes are extremely sparse. PCA and analysis of average glomerular tuning suggests that these data are highly dimensional and that glomeruli are narrowly tuned to a small subset of chemically related odorants. These results have important implications regarding odor coding at low odorant concentrations. Previous research has suggested that odor quality is encoded through a 31 combinatorial coding, in part because earlier studies have indicated that ORs can be broadly tuned to a wide range of odorants (Murthy 2011). In a combinatorial odor coding system, odor identity would be encoded by specific combinations of OR responses. However, at low odorant concentrations OR tuning appears to be narrowly tuned and very specific, creating a high-dimensional odor coding space. It may be the case that at low-concentration (and perhaps more behaviorally relevant) odorant regimes that odor coding is non-combinatorial. This means that it may be that activation of individual ORs rather than OR combinations is sufficient to encode odor identity. Interpretation of these results may be limited by the fact that OSN responses were measured in anesthetized mice: responses may vary from awake mice due to artificial sniffing, lack of neuromodulation as a function of behavioral state, or anesthetic interference with neural responses. Future studies may wish to define the breadth of OR tuning in awake mice. Identified glomeruli In these low concentration regimes, certain glomeruli were able to be identified across animals solely based on their responsiveness and approximate location. Analysis of their response spectra showed consistency between that of each identified glomerulus, specifically to the odorant to which the glomerulus is maximally responsive. This suggests that each of these identified glomeruli is the “same” across animals (i.e. each correspond to the same OR). This may provide a potential platform for functionally probing glomeruli. Previous studies have relied on gene-targeting techniques to probe specific glomeruli in vivo, 32 however, these results suggest that certain glomeruli may be able to be functionally identified based off their responsiveness to specific low-concentration odorants and their approximate position. Future studies could expand this platform by identifying the receptors behind each functionally identified glomerulus in order to more robustly identify and describe their response characteristics. Chemotopy Previous literature has shown that similar glomerular odorant activation patterns are observed to specific odorants (Murthy 2011). This has helped to inspire the hypothesis that certain odorant structural characteristics may be spatially represented in the OB, or that there exists a chemotopic map of odorant quality across the OB. Further studies have supported this hypothesis by showing that chemically similar odors tend to evoke similar patterns of activity in stereotypic regions of the OB. However, this hypothesis is currently not resolved: additional studies have refuted the idea of a chemotopic arrangement of the OB (Soucy et al. 2009; Chae et al. 2019). Our results add to this growing body of support. Here, we show that maximum projection maps of glomeruli with maximal sensitivities to similar odorants are spatially clustered together on the OB surface, though these clusters are somewhat fragmented. This clustering suggests that glomeruli that respond to chemically similar odorants cluster together in stereotypic patterns across the surface of the OB. However, these results are observational and lack robust statistical analysis. Future studies may wish to further examine this pattern to analyze the statistical significance of these perceived clusters. 33 Olfr1377 response spectrum The olfr1377 OR has previously been characterized in vitro but has not yet been characterized in vivo (Saito et al. 2009). To characterize this in vivo we presented a panel of 48 aromatics, pyrazines and ketones at a range of ~0.1-1 nM concentrations. Olfr1377 was shown to be responsive to a small selection of these odors including many acetophenone derivatives. This response spectrum is perhaps more narrowly tuned than ones that that have been previously identified for olfr1377, however this may reflect the difference between responses measured in vivo vs. in vitro. However, the range of odors and concentrations tested is somewhat limited, to more robustly characterize the response spectrum of olfr1377 further analysis should test a wider range of odors at different concentrations. Of great interest is that olfr1377 was found to be sensitive to 4-methoxyacetophenone at extremely low concentrations (smaller than ~0.04 pM). Furthermore, after the administration of 4-mehtoxyacetophenone at high concentrations, the response of olfr1377 is suppressed. This suggests that 4-methoxyacetophenone is able to tightly bind to the olfr1377 receptor, and may competitively exclude other odors (including itself) from further binding. This phenomenon, often called inverse agonism, has interesting implications regarding receptor affinity and ligand binding. Some odorants prevent subsequent OR binding due to high OR affinity, inhibiting OSN responses. In addition, perception of odor mixtures may be influenced by inverse agonism due to such odorant- 34 receptor interactions, such that the total odorant mixture response pattern is not necessarily equal to the sum of its parts (L. Xu et al. 2020). Conclusion Analysis of mouse glomerular responses to odorants at near-threshold concentrations has revealed that responses in these concentration regimes are extremely sparse, and that ORs are narrowly tuned to one or a small range of chemically similar odors. Some glomeruli are identifiable across animals solely based on their approximate location and the odorant to which they are most responsive to, and glomeruli that are maximally sensitive to similar odorants cluster together chemotopically across the OB surface. The OR defined glomerulus olfr1377 is responsive to a range of acetophenone derivatives, and extremely sensitive to 4-methoxyacetophenone low concentrations. This work adds to the effort to characterize OR functions and responses. Knowledge of OR function is essential to understanding further upstream neural processing, and ultimately the central olfactory system’s ability to create an odor perception from sensory stimuli. 35 ACKNOWLEDGEMENTS This research was supported by the Undergraduate Research Opportunities Program (UROP) at the University of Utah, and grants from the NSF (IOS-1555919) and the NIH (NS109979). I thank Matt Wachowiak for excellent mentoring and assistance through every step of this project. I also thank Shawn D. Burton for preforming the initial neuralimaging experiments; Alla Borisyuk for data analysis and MATLAB guidance; Shaina M. Short, Isaac A. Youngstrom, and Tom Peiting for helpful discussion; and Rebecca L. Kummer, Gustavo A. Vásquez-Opazo, and Diana Borrego for technical assistance. Finally, I thank the School of Biological Sciences, Department of Neurobiology and Anatomy, and the Honors College at the University of Utah for supporting me in this research and in my undergraduate education. 36 REFERENCES Araneda, R. C., A. D. Kini, and S. Firestein. 2000. “The Molecular Receptive Range of an Odorant Receptor.” Nature Neuroscience 3 (12): 1248–55. https://doi.org/10.1038/81774. Bozza, Thomas, Paul Feinstein, Chen Zheng, and Peter Mombaerts. 2002. “Odorant Receptor Expression Defines Functional Units in the Mouse Olfactory System.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 22 (8): 3033–43. https://doi.org/20026321. Bozza, Thomas, John P. McGann, Peter Mombaerts, and Matt Wachowiak. 2004. “In Vivo Imaging of Neuronal Activity by Targeted Expression of a Genetically Encoded Probe in the Mouse.” Neuron 42 (1): 9–21. https://doi.org/10.1016/s0896-6273(04)00144-8. Bozza, Thomas, Anne Vassalli, Stefan Fuss, Jing-Ji Zhang, Brian Weiland, Rodrigo Pacifico, Paul Feinstein, and Peter Mombaerts. 2009. “Mapping of Class I and Class II Odorant Receptors to Glomerular Domains by Two Distinct Types of Olfactory Sensory Neurons in the Mouse.” Neuron 61 (2): 220–33. https://doi.org/10.1016/j.neuron.2008.11.010. Buck, L., and R. Axel. 1991. “A Novel Multigene Family May Encode Odorant Receptors: A Molecular Basis for Odor Recognition.” Cell 65 (1): 175–87. https://doi.org/10.1016/0092-8674(91)90418-x. Buck, Linda B. 2004. “Olfactory Receptors and Odor Coding in Mammals.” Nutrition Reviews 62 (11 Pt 2): S184-188; discussion S224-241. https://doi.org/10.1111/j.1753-4887.2004.tb00097.x. Burton, Shawn D., Mia Wipfel, Michael Guo, Thomas P. Eiting, and Matt Wachowiak. 2019. “A Novel Olfactometer for Efficient and Flexible Odorant Delivery.” Chemical Senses 44 (3): 173–88. https://doi.org/10.1093/chemse/bjz005. Chae, Honggoo, Daniel R. Kepple, Walter G. Bast, Venkatesh N. Murthy, Alexei A. Koulakov, and Dinu F. Albeanu. 2019. “Mosaic Representations of Odors in the Input and Output Layers of the Mouse Olfactory Bulb.” Nature Neuroscience 22 (8): 1306–17. https://doi.org/10.1038/s41593-019-0442-z. Chen, W. R., W. Xiong, and G. M. Shepherd. 2000. “Analysis of Relations between NMDA Receptors and GABA Release at Olfactory Bulb Reciprocal Synapses.” Neuron 25 (3): 625–33. https://doi.org/10.1016/s0896-6273(00)81065-x. Chittka, Lars, and Axel Brockmann. 2005. “Perception Space—The Final Frontier.” PLoS Biology 3 (4): e137. https://doi.org/10.1371/journal.pbio.0030137. Davison, Ian G., and Lawrence C. Katz. 2007. “Sparse and Selective Odor Coding by Mitral/Tufted Neurons in the Main Olfactory Bulb.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 27 (8): 2091– 2101. https://doi.org/10.1523/JNEUROSCI.3779-06.2007. Farahbod, Haleh, Brett A. Johnson, S. Sakura Minami, and Michael Leon. 2006. “Chemotopic Representations of Aromatic Odorants in the Rat Olfactory Bulb.” The Journal of Comparative Neurology 497 (3): 350–66. https://doi.org/10.1002/cne.20982. 37 Hildebrand, J. G., and G. M. Shepherd. 1997. “Mechanisms of Olfactory Discrimination: Converging Evidence for Common Principles across Phyla.” Annual Review of Neuroscience 20: 595–631. https://doi.org/10.1146/annurev.neuro.20.1.595. Isaacson, J. S., and B. W. Strowbridge. 1998. “Olfactory Reciprocal Synapses: Dendritic Signaling in the CNS.” Neuron 20 (4): 749–61. https://doi.org/10.1016/s08966273(00)81013-2. Johnson, Brett A., Sabrina L. Ho, Zhe Xu, Joanne S. Yihan, Sallis Yip, Edna E. Hingco, and Michael Leon. 2002. “Functional Mapping of the Rat Olfactory Bulb Using Diverse Odorants Reveals Modular Responses to Functional Groups and Hydrocarbon Structural Features.” The Journal of Comparative Neurology 449 (2): 180–94. https://doi.org/10.1002/cne.10284. Kauer, J. S., and A. R. Cinelli. 1993. “Are There Structural and Functional Modules in the Vertebrate Olfactory Bulb?” Microscopy Research and Technique 24 (2): 157–67. https://doi.org/10.1002/jemt.1070240207. Ma, Limei, Qiang Qiu, Stephen Gradwohl, Aaron Scott, Elden Q. Yu, Richard Alexander, Winfried Wiegraebe, and C. Ron Yu. 2012. “Distributed Representation of Chemical Features and Tunotopic Organization of Glomeruli in the Mouse Olfactory Bulb.” Proceedings of the National Academy of Sciences of the United States of America 109 (14): 5481–86. https://doi.org/10.1073/pnas.1117491109. Malnic, B., J. Hirono, T. Sato, and L. B. Buck. 1999. “Combinatorial Receptor Codes for Odors.” Cell 96 (5): 713–23. https://doi.org/10.1016/s0092-8674(00)80581-4. Meister, M., and T. Bonhoeffer. 2001. “Tuning and Topography in an Odor Map on the Rat Olfactory Bulb.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 21 (4): 1351–60. Mombaerts, P., F. Wang, C. Dulac, S. K. Chao, A. Nemes, M. Mendelsohn, J. Edmondson, and R. Axel. 1996. “Visualizing an Olfactory Sensory Map.” Cell 87 (4): 675–86. https://doi.org/10.1016/s0092-8674(00)81387-2. Mori, Kensaku, Hiroshi Nagao, and Yoshihiro Yoshihara. 1999. “The Olfactory Bulb: Coding and Processing of Odor Molecule Information.” Science 286 (5440): 711– 15. https://doi.org/10.1126/science.286.5440.711. Mori, Kensaku, Yuji K. Takahashi, Kei M. Igarashi, and Masahiro Yamaguchi. 2006. “Maps of Odorant Molecular Features in the Mammalian Olfactory Bulb.” Physiological Reviews 86 (2): 409–33. https://doi.org/10.1152/physrev.00021.2005. Murthy, Venkatesh N. 2011. “Olfactory Maps in the Brain.” Annual Review of Neuroscience 34: 233–58. https://doi.org/10.1146/annurev-neuro-061010-113738. Nara, Kiyomitsu, Luis R. Saraiva, Xiaolan Ye, and Linda B. Buck. 2011. “A Large-Scale Analysis of Odor Coding in the Olfactory Epithelium.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 31 (25): 9179–91. https://doi.org/10.1523/JNEUROSCI.1282-11.2011. Oka, Yuki, Sayako Katada, Masayo Omura, Makiko Suwa, Yoshihiro Yoshihara, and Kazushige Touhara. 2006. “Odorant Receptor Map in the Mouse Olfactory Bulb: In Vivo Sensitivity and Specificity of Receptor-Defined Glomeruli.” Neuron 52 (5): 857–69. https://doi.org/10.1016/j.neuron.2006.10.019. 38 Peterlin, Zita, Stuart Firestein, and Matthew E. Rogers. 2014. “The State of the Art of Odorant Receptor Deorphanization: A Report from the Orphanage.” The Journal of General Physiology 143 (5): 527–42. https://doi.org/10.1085/jgp.201311151. Ressler, K. J., S. L. Sullivan, and L. B. Buck. 1993. “A Zonal Organization of Odorant Receptor Gene Expression in the Olfactory Epithelium.” Cell 73 (3): 597–609. https://doi.org/10.1016/0092-8674(93)90145-g. Rubin, B. D., and L. C. Katz. 1999. “Optical Imaging of Odorant Representations in the Mammalian Olfactory Bulb.” Neuron 23 (3): 499–511. https://doi.org/10.1016/s0896-6273(00)80803-x. Saito, Harumi, Qiuyi Chi, Hanyi Zhuang, Hiroaki Matsunami, and Joel D. Mainland. 2009. “Odor Coding by a Mammalian Receptor Repertoire.” Science Signaling 2 (60): ra9. https://doi.org/10.1126/scisignal.2000016. Schoppa, N. E., J. M. Kinzie, Y. Sahara, T. P. Segerson, and G. L. Westbrook. 1998. “Dendrodendritic Inhibition in the Olfactory Bulb Is Driven by NMDA Receptors.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 18 (17): 6790–6802. Soucy, Edward R., Dinu F. Albeanu, Antoniu L. Fantana, Venkatesh N. Murthy, and Markus Meister. 2009. “Precision and Diversity in an Odor Map on the Olfactory Bulb.” Nature Neuroscience 12 (2): 210–20. https://doi.org/10.1038/nn.2262. Tan, Jie, Agnès Savigner, Minghong Ma, and Minmin Luo. 2010. “Odor Information Processing by the Olfactory Bulb Analyzed in Gene-Targeted Mice.” Neuron 65 (6): 912–26. https://doi.org/10.1016/j.neuron.2010.02.011. Vassar, Robert, Steve K. Chao, Raquel Sitcheran, Jennifer M. Nun˜ez, Leslie B. Vosshall, and Richard Axel. 1994. “Topographic Organization of Sensory Projections to the Olfactory Bulb.” Cell 79 (6): 981–91. https://doi.org/10.1016/0092-8674(94)90029-9. Wachowiak, M., and L. B. Cohen. 2001. “Representation of Odorants by Receptor Neuron Input to the Mouse Olfactory Bulb.” Neuron 32 (4): 723–35. https://doi.org/10.1016/s0896-6273(01)00506-2. Xu, Fuqiang, Nian Liu, Ikuhiro Kida, Douglas L. Rothman, Fahmeed Hyder, and Gordon M. Shepherd. 2003. “Odor Maps of Aldehydes and Esters Revealed by Functional MRI in the Glomerular Layer of the Mouse Olfactory Bulb.” Proceedings of the National Academy of Sciences of the United States of America 100 (19): 11029– 34. https://doi.org/10.1073/pnas.1832864100. Xu, Lu, Wenze Li, Venkatakaushik Voleti, Dong-Jing Zou, Elizabeth M. C. Hillman, and Stuart Firestein. 2020. “Widespread Receptor-Driven Modulation in Peripheral Olfactory Coding.” Science (New York, N.Y.) 368 (6487). https://doi.org/10.1126/science.aaz5390. Zhang, Jingji, Guangzhe Huang, Adam Dewan, Paul Feinstein, and Thomas Bozza. 2012. “Uncoupling Stimulus Specificity and Glomerular Position in the Mouse Olfactory System.” Molecular and Cellular Neurosciences 51 (3–4): 79–88. https://doi.org/10.1016/j.mcn.2012.08.006.