CD20 is a mammalian odorant receptor expressed in a subset of olfactory sensory neurons that mediates innate avoidance of predators

Summary: The mammalian olfactory system detects and discriminates between millions of odorants to elicit appropriate behavioral responses. While much has been learned about how olfactory sensory neurons detect odorants and signal their presence, how specific innate, unlearned behaviors are initiated in response to ethologically relevant odors remains poorly understood. Here, we show that the 4-transmembrane protein CD20, also known as MS4A1, is expressed in a previously uncharacterized subpopulation of olfactory sensory neurons in the main olfactory epithelium of the murine nasal cavity and functions as a mammalian odorant receptor that recognizes compounds produced by mouse predators. While wild-type mice avoid these predator odorants, mice genetically deleted of CD20 do not appropriately respond. Together, this work reveals a novel CD20-mediated odor-sensing mechanism in the mammalian olfactory system that triggers innate behaviors critical for organismal survival.

To survive, animals must accurately detect, correctly interpret, and appropriately 50 respond to sensory stimuli in their environment. For most non-primate mammals, the 51 richest source of this information is the immense variety of small molecules present in 52 their external surroundings, which may signify the presence of predators, food, or mates 53 (Brennan and Zufall, 2006). These chemicals are primarily detected by odorant receptors 54 (ORs) expressed at the sensory endings of peripheral olfactory sensory neurons (OSNs), 55 which are coupled to the higher brain circuits tasked with mediating odor perception and 56 initiating olfactory-driven behavior (Bargmann, 2006; Buck and Axel, 1991; Leinwand and 57 Chalasani, 2011; Su et al., 2009). However, how mammals detect and process different 58 classes of olfactory stimuli to initiate distinct behaviors is still not well understood (i.e., 59 how does a mouse know to avoid a cat but to actively seek out a piece of cheese?). 60 61 One emerging hypothesis is that distinct subpopulations of OSNs might be 62 responsible for different behaviors. The olfactory system can be subdivided into multiple 2009; Shinoda et al., 1989). A handful of these olfactory subsystems have been 68 extensively studied, which has led to significant insight into their role in olfactory 69 perception and odor-driven behaviors. Particularly important for elucidating the role of 70 these olfactory subsystems has been the identification of the ORs that they express. Each 71 subsystem expresses different types of ORs, which enable them to detect subsets of 72 chemical space and mediate specific behaviors. For instance, the largest subdivision in 73 the mouse, the main olfactory system, owing to its immense receptor repertoire of 74 approximately 1000 distinct ORs (Buck and Axel, 1991), is able to detect essentially all 75 volatile odorants and therefore plays a key role in odor discrimination and odorant- Nonetheless, despite progress in elucidating the function of a few of these olfactory 86 subsystems, the specific roles of others remain poorly understood. One of the least 87 understood is the olfactory necklace subsystem, which seems to mediate seemingly 88 opposing behaviors for both feeding and innate avoidance of noxious stimuli (Hu et al., preventing a rigorous assessment of whether MS4A proteins participate in odor detection 100 in vivo. Indeed, because MS4A proteins do not resemble any previously described 101 odorant receptors -they are four-transmembrane spanning proteins rather than seven-102 transmembrane GPCRs, there remains some skepticism about whether MS4A proteins 103 function as ORs in vivo (Zimmerman and Munger, 2021). Here, we use newly generated 104 Ms4a knockout mice to show that MS4A proteins function as bona fide ORs in vivo. 105 Moreover, we show that the MS4A family member MS4A1, (better known as CD20, a 106 protein previously identified as a co-receptor for the B cell receptor in lymphocytes), is not 107 expressed in the necklace, but is instead expressed in a novel subset of OSNs outside 108 the necklace. genes (hereafter, referred to as Ms4a cluster knockout mice) (Figures 1A, S1A, and S1B). 124 Ms4a cluster knockout mice are viable, fertile, produced at Mendelian frequency, and are 125 overtly indistinguishable from their wild-type littermates, enabling us to assess olfactory 126 performance in these Ms4a-deficient animals ( Figure S1C).

128
To begin to test the role of Ms4a genes in olfactory function, we initially exposed 129 freely behaving Ms4a-deficient mice or their wild-type littermates to 2,5-dimethylpyrazine 130 (2,5-DMP), oleic acid (OA), or alpha-linolenic acid (ALA), previously ( Figures 1D and 1E). In addition, deletion of Ms4as did not affect other non-odor mediated 179 avoidance behaviors such as the amount of time spent in open arms in an elevated plus 180 maze assay ( Figure S1E). Taken together, these results indicate that Ms4a genes encode 181 ORs that mediate specific odor-driven avoidance responses in mammals. were overtly indistinguishable from wild type mice in other ways -they exhibited similar 226

Ms4a6c detects DMP in necklace OSNs
locomotive behaviors and behaved similarly to wild-type mice in assays of anxiety (such 227 as the elevated plus maze) ( Figures S3B and S3C), strongly suggesting that the failure 228 to respond to 2,5-DMP was a specific defect in this particular odor-driven behavior and 229 not a sign of more general nervous system dysfunction.

231
The observation that MS4A1 is required for a mouse to avoid the predator-derived 232 compound 2,5-DMP was surprising since the only previously ascribed function of MS4A1 233 is as a co-receptor for the B-cell receptor in circulating mature lymphocytes, where it is 234 known as CD20 (Tedder and Engel, 1994;Tedder et al., 1988). Although it seemed 235 unlikely that lymphocytes would play a critical role in mediating this olfactory-driven 236 behavior, we assessed the ability of Rag-1-deficent mice, which lack all mature 237 lymphocytes (Mombaerts et al., 1992), to avoid 2,5-DMP. Rag-1 knockout mice avoided 238 DMP to a similar extent as wild-type mice, indicating that mature lymphocyte function was 239 not required for avoidance of 2,5-DMP and further suggesting that CD20 might act in cells 240 outside of the immune system to mediate avoidance of this odor ( Figures S3D and S3E). 241 242 CD20 is expressed in a previously unidentified subpopulation of OSNs 243 244 To identify cells in the olfactory system in which Ms4a1 might be expressed, we 245 stained coronal sections of the mouse olfactory epithelium with an antibody specific for 246 MS4A1. A relatively sparse population of MS4A1-expressing cells that did not express 247 lymphoid markers was found, whose cell bodies reside within the epithelial layer of the 248 main olfactory epithelium (MOE) ( Figure 4A). To verify this unexpected observation, we 249 stained coronal sections of the mouse olfactory epithelium with two additional anti-MS4A1 250 antibodies (raised in different species and recognizing different MS4A1 epitopes). These 251 three anti-MS4A1 antibodies all co-labeled the same cells in the MOE ( Figure 4B). These 252 antibodies did not stain any cells in olfactory epithelial sections obtained from Ms4a1 253 knockout mice, confirming their specificity ( Figure 4C). Moreover, combined fluorescent 254 in situ hybridization and immunohistochemistry experiments detected Ms4a1 mRNA and 255 MS4A1 protein in the same cells indicating that Ms4a1 is expressed in non-lymphoid cells 256 of the mouse olfactory system ( Figure 4D).

258
The ( Figures 4G and 4H), or OSNs of the vomeronasal organ ( Figure 4H). Nonetheless, using 282 a combination of iDISCO tissue clearing and light sheet microscopy, we found that like 283 other OSN populations, MS4A1-expressing neurons also extended their axons into 284 glomeruli within the mouse olfactory bulb suggesting that MS4A1 is expressed in a 285 previously uncharacterized population of olfactory sensory neurons in the olfactory 286 epithelium and that like other members of the MS4A family, MS4A1 might also function 287 as an olfactory chemoreceptor ( Figure 4I). 288 289

MS4A1 detects nitrogenous heterocyclic compounds in vitro 290 291
To begin to test this hypothesis, and to explore what types of odors MS4A1 might 292 detect, we examined whether heterologously expressed MS4A1 might respond to 293 additional extracellular chemicals to mediate a calcium influx in HEK293 cells co-294 expressing GCaMP6s ( Figure 5A). Expression of MS4A1 did not increase the baseline 295 rate of calcium transients in HEK293 cells ( Figure S5A) but increased intracellular calcium 296 spikes upon presentation of specific chemicals ( Figures 5A and 5B). This was true for 297 both human and mouse MS4A1 proteins ( Figures 5A and S5B). MS4A1 responses were 298 tuned to nitrogenous heterocyclic compounds, including 2,3-DMP, 2,5-DMP and 2,6-299 DMP, and to a lesser extent indole and quinoline ( Figures 5B and S5C). However, not all 300 nitrogenous heterocyclic compounds induced calcium transients in MS4A1-expressing 301 cells, nor did non-nitrogenous compounds like isoamyl acetate and vanillin, indicating 302 some ligand specificity ( Figure 5B). Dose response curves revealed nanomolar and low 303 micromolar EC50s for two specific MS4A1-ligand pairs, which is well within the range of 304 what has been observed for other mammalian odorant receptor/ligand relationships 305 ( Figure 5C). Moreover, depletion of extracellular calcium completely eliminated all 306 calcium transients observed in response to ligand presentation ( Figure 5D). immunostaining experiments with knockout mice were performed with littermate wild-type 421 control mice. The following primer sequences were used for genotyping: 1) Ms4a cluster 422 knockout, common primer (5'-GACAAATGAACTAACCTTGCTTGG-3'), wild type specific  knockout, and Ms4a6c knockout mice that were exposed to water or odors as described 505 above. The staining protocol followed the guidelines provided in the Advanced Cell 506 Diagnostics RNAscope wash buffer (2 minutes twice at RT) between each step following probe 527 hybridization. The mouse target probes used in this study were as follows: Ms4a1-c2, 528 #318671-C2 and Car2-c2, #313781-C2. 529 For the pos-RNAscope immunostaining, sections were blocked with blocking 530 buffer (0.1% Triton X-100 (Sigma-Aldrich, #X100) 5% Normal Donkey serum (Jackson 531 ImmunoResearch, #017-000-121), 3% Bovine Serum Albumin (VWR, #97061-416) in 532 PBS) for 30 minutes at RT. Sections were then incubated with anti-pS6 antibodies (1:100) 533 in blocking buffer overnight at 4°C. On the following day, the slides were washed three 534 times with PBS (5 minutes each at RT) and then incubated with the secondary antibody 535 (1:300) in blocking solution for 45 minutes at RT. Afterwards, the slides were washed 536 three times with PBS (5 minutes each at RT) and mounted using Vectashield antifade 537 mounting media with DAPI (Vector Laboratories, #H-1200-10). To secure the coverslips, 538 nail polish was applied, and the slides were imaged using confocal microscopy, following 539 the procedures described below. 540 541 Conventional FISH 542 For Gucy1b2, Trpc2, Trpm5, Gucy2d, and V1rb1 RNA detection, conventional 543 FISH was performed on nasal epithelial sections from C57/BL6 wild-type mice, following 544 a modified protocol (Ishii et al., 2004). The RNA probes for these genes were previously were generated through in vitro transcription from fully linearized and purified template 548 plasmids (described in Plasmids) containing target gene sequences, using equilibrated 549 phenol (Sigma-Aldrich, #P9346) -chloroform/isoamyl alcohol (Sigma-Aldrich, #25666). 550 Frozen sections were air-dried, fixed with 4% PFA/1X PBS for 10 minutes, and then 551 acetylated with a mixture of 0.1M triethanolamine (Sigma-Aldrich, #90279) and 0.25% 552 acetic anhydride (Sigma-Aldrich, #320102) for 15 minutes at RT. Pre-hybridization was 553 performed in a prehybridization solution (10 mM Tris, pH 7.5, 600 mM NaCl, 1 mM EDTA, 554 0.25% SDS, 1X Denhardt's (Sigma, #D-2532), 50% formamide (Roche, #1814320), 555 300µg/ml yeast tRNA (Sigma-Aldrich, #R6750)) for 5 hours at 65 °C. Following pre-556 hybridization, the sections were hybridized overnight at 60 °C with FITC-labeled RNA 557 probes (1 µg/ml) in a hybridization solution (10 mM Tris, pH 7.5, 200 mM NaCl, 5 mM 558 EDTA, 0.25% SDS, 1X Denhardt's, 50% formamide, 300 µg/ml yeast tRNA, 10% dextran 559 sulfate (Bio Basic, #DB0160), 5 mM NaH2PO4, 5 mM Na2HPO4). 560 On the subsequent day, the slides were sequentially washed with the following On the third day, the slides were washed three times with PBST (1X PBS/0.1% 572 Tween-20), incubated with diluted TSA plus fluorescein (1:50) for 5-10 minutes at RT, 573 and washed five times with PBST. Finally, the sections were immunostained with anti-574 MS4A1 antibodies and imaged using confocal microscopy, following the procedures 575 described below. 576 577 Immunostaining 578 Sections were incubated in a blocking solution containing 5% normal donkey 579 serum, 0.1% Triton-X100, and 1X Tris-buffered saline (TBS) for 1 hour at RT. 580 Subsequently, sections were incubated overnight at 4°C with primary antibodies diluted 581 in blocking solution. On the following day, slides were washed three times with TBST 582 (0.1% Triton-X100 in TBS) and then incubated with secondary antibodies in blocking 583 solution for 1 hour at RT. Afterwards, the slides were washed three times with TBST, 584 counterstained, and mounted using Vectashield antifade mounting media with DAPI. To 585 secure the coverslips, nail polish was applied, and the slides were imaged using confocal 586 microscopy, following the procedures described below. 587 588 Confocal microscopy 589 Slides were imaged using an LSM 900 Airyscan2 confocal microscope (Zeiss) 590 equipped with various objective lenses, including 10X/0.45 M27, 20X/0.8 M27, 40X/1.1 591 water Corr M72, and 63X/1.4 oil DIC. To enhance image quality, acquired digital images 592 were processed by applying a median filter to remove debris that was significantly smaller 593 than the structures being analyzed. Additionally, multi-channel Z-stacks were projected 594 into two dimensions using Zen blue 3.1 software (Zeiss). 595 596 Quantification of phospho-S6 (pS6) positive cells 597 Car2 or Ms4a1 positive cells were selected using imageJ sorftware. pS6 intensity 598 was determined following subtraction of background signal from cells in the olfactory 599 epithelium lacking Car2 or Ms4a1 signal. For necklace cells, 12 sections from the 600 posterior olfactory epithelium were collected, and three Car2-positive regions were 601 randomly selected from each section to perform quantification. For Ms4a1-positive cells, 602 24 sections equally spaced throughout the anterior to posterior axis of the olfactory 603 epithelium were collected, and all the Ms4a1-positive cells from these sections were 604 analyzed. Analysis was performed blinded to genotype and stimulus to ensure unbiased 605 quantification. 606 607 iDISCO 608 iDISCO was performed on the olfactory bulbs of 8-12 week old C57/BL6 wild-type 609 mice following the protocol described by Renier  images were acquired using InspectorPro software (LaVision BioTec), and three-637 dimensional reconstruction and analysis were performed using Imaris x64 software 638 (v.8.0.1, Bitplane). 639

Odor-driven behavior assay 641
For behavioral experiments, 8-12-week-old Ms4a cluster knockout, Ms4a6c 642 knockout, Ms4a1 knockout and littermate wild-type control mice were group housed in 643 the behavioral assay room and allowed to acclimate for at least one day prior to the start 644 of the experiments. Two hours prior to the behavioral assay, mice were individually fasted 645 in their home cage. During the experiment, mice were placed in single use, disposable 646 cages (Innovive, #M-BTM) with a disposable paper curtain separating the avoidance zone 647 from the odorized zone. Only the odorized zone was enclosed by an acrylic sheet on the 648 top of cage. A clean filter paper was placed in a 35 mm petri dish within the odorized 649 zone. Without any odor stimulus, mice were first allowed to freely explore their 650 surroundings for 30 minutes. Subsequently, the mice were exposed to water (40 µL) 651 applied to the filter paper in the odorized zone for a duration of 3 minutes. After water 652 exposure, the same mouse was then exposed to 40 µL of odorant for 3 minutes. The 653 odorant was delivered onto a fresh filter paper for the experiment. Animal behavior during 654 the entire experiment, including habituation, water exposure, and odor exposure was 655 recorded with a webcam (Logitech, #LOWCC920S). Those videos were then analyzed 656 with ezTrack (Pennington et al., 2019).

658
Elevated plus maze assay 659 The elevated plus maze (EPM) apparatus consisted of plus-shaped (+) apparatus  660 with two open and two enclosed arms. The closed arms are enclosed by black walls 661 (30 × 5 × 15 cm) and the open arms are exposed (30 × 5 × 0.25 cm). The maze was 662 elevated 45 cm above the floor, and a red fluorescent light was positioned 1 meter above 663 the maze as light source. The whole assay was performed in a darkroom. 8-12-week-old 664 Ms4a cluster knockout, Ms4a6c knockout, Ms4a1 knockout, and littermate wild-type 665 control mice were group housed in the darkroom for 1 hour prior to the experiment. 666 Individual mice were placed at the center of the maze, and the mouse was allowed to 667 freely explore the maze for 5 minutes. anti-pS6 antibodies (1:400) in blocking buffer overnight at 4°C. On the following day, the 737 cells were washed three times with PBS and then incubated with secondary antibody 738 (1:300) in blocking solution for 45 minutes at RT. After three washes with PBS, the cells 739 were mounted using Vectashield antifade mounting media with DAPI. 740 741 Calcium imaging 742 24 hours post-transfection, media in the wells were aspirated and washed twice 743 with 1X Ringer's solution supplemented with 1 mM CaCl2 (Ca 2+ -Ringer's solution). The 744 wells were then incubated for 30 minutes in the cell culture incubator with Ca 2+ -Ringer's 745 solution. Following the incubation, the plate was transferred to a Lionheart LX Automated 746 microscope (BioTek), and calcium imaging was performed using Gen5 software (BioTek). 747 Preliminary images were acquired with brightfield, RFP, and GFP filters at 20X 748 magnification prior to each experiment, focusing on an imaging field containing cell 749 numbers between 50 and 200. Subsequently, using the same field of view and fixed z-750 axis, images were captured for 10 minutes (1 FPS). During the kinetic image acquisition, 751 either Ca 2+ -Ringer's solution as a negative control or specific odorants (50 µM for 2,3-752 DMP, 2,5-DMP, 2,6-DMP, indole, quinoline, pyridine, pyrrolidine, vanillin, IAA) solubilized 753 in Ca 2+ -Ringer's solution were pipetted into the upper edge of each well after 360 seconds 754 for a duration of 10 seconds. To determine dose-response curves and calculate the EC50, 755 HEK293 cells co-expressing GCaMP6s and either mCherry or mCherry-MS4A1 were 756 treated with six logarithmic orders of 2,3-DMP or 2,5-DMP (ranging from 10 nM to 1 mM) 757 starting with the lowest concentration. For experiments conducted without extracellular 758 calcium, all solutions were replaced with 1X Ringer's solution supplemented with 1 mM 759 EGTA and 1 mM EDTA to chelate calcium. All acquired images were aligned to the first 760 image of each experiment using Gen5 software, and subsequent images were analyzed 761 using Fiji software.

763
Analysis of calcium imaging data 764 Mp4 videos were converted into a sequence of PNG images with ffmpeg software, 765 the image sequence was then imported into Fiji. GCaMP6s and mCherry positive cells 766 were selected, and their GCaMP6s intensities were calculated across the whole image 767 sequence, which was subsequently analyzed using a customized R script. Briefly, for 768 each selected cell, the average intensity and standard error of GCaMP6s 30 seconds 769 prior to ligand presentation was calculated. 2.5-fold of the standard error above mean 770 intensity was then used as a threshold to determine if the cell responded to the odor.

772
Statistics and reproducibility 773 For quantification of pS6, at least three biological repeats were performed for each 774 odorant treatment. All analyses were conducted blinded to genotype and stimulus to 775 ensure unbiased quantification. One-way ANOVA was performed to calculate the 776 statistical difference of mean intensity of pS6 from all groups. A post-hoc Dunnett's test 777 was used to determine if the mean intensity of pS6 from a given odorant treatment was 778 significantly different from the eugenol control group. 779 For the odor-driven behavior assay, at least five biological repeats were performed 780 for each odorant. The analysis was conducted in an automated manner whenever 781 possible in the absence of human supervision to ensure blinding to genotype and 782 stimulus. For the total distance traveled, an unpaired Welch t-test was performed to 783 calculate statistical differences. For the distance between the mouse's center of mass 784 and the odor, a paired t-test was performed to calculate statistical differences. For 785 comparing the proportion of time spent in the odorized zone and the distance between 786 the mouse's center of mass and odor (unpaired), a one-way ANOVA was performed to 787 calculate statistical differences between all groups. A post-hoc Dunnett's Test was 788 applied to determine if the values from a given odorant treatment are significantly different 789 from those from the water control group. For the avoidance index, an unpaired Welch t-790 test was used to compare each knockout group and their wild-type littermates. A false-791 discovery rate was controlled using a two-stage step-up developed by Benjamini, Krieger, 792 and Yekutieli. 793 For the EPM assay, at least five biological repeats were performed for each 794 genotype. An unpaired Welch t-test was used to compare time spent in the open arm 795 between groups. 796 For calcium imaging, at least nine biological repeats were performed for each 797 odorant. The analysis was conducted blinded to protein expressed and stimulus to ensure 798 unbiased quantification. For identifying 2,5-DMP responsive MS4A receptors, a one-way 799 ANOVA was performed using a post-hoc Dunnett's test. For screening chemicals that