The Anatomical Logic of Smell 2 3 Mayra L
Total Page:16
File Type:pdf, Size:1020Kb
bioRxiv preprint doi: https://doi.org/10.1101/2021.06.16.448475; this version posted June 16, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Segura, Abou Moussa et al Running title: 3D transcriptomics of the mouse nose 1 The anatomical logic of smell 2 3 Mayra L. Ruiz Tejada Segura1,2,3,#; Eman Abou Moussa4,#; Elisa Garabello5,6; Thiago S. 4 Nakahara7; Melanie Makhlouf4; Lisa S. Mathew4; Filippo Valle5; Susie S.Y. Huang4; Joel 5 D. Mainland8,9; Michele Caselle5; Matteo Osella5; Stephan Lorenz4,10; Johannes Reisert8; 6 Darren W. Logan10; Bettina Malnic7; Antonio Scialdone1,2,3,†,*; Luis R. Saraiva4,8,11, †,* 7 8 1 Institute of Epigenetics and Stem Cells, Helmholtz Zentrum München; Feodor- 9 Lynen-Strasse 21, 81377 München, Germany 10 2 Institute of Functional Epigenetics, Helmholtz Zentrum München; Ingolstädter 11 Landstraße 1, 85764 Neuherberg, Germany 12 3 Institute of Computational Biology, Helmholtz Zentrum München; Ingolstädter 13 Landstraße 1, 85764 Neuherberg, Germany 14 4 Sidra Medicine; PO Box 26999; Doha, Qatar. 15 5 Physics Department, University of Turin and INFN; via P. Giuria 1, 10125 Turin, 16 Italy 17 6 Department of Civil and Environmental Engineering, Cornell University; Ithaca, NY 18 14853, USA 19 7 Department of Biochemistry, University of São Paulo; São Paulo, Brazil. 20 8 Monell Chemical Senses Center; 3500 Market Street, Philadelphia, PA 19104, USA. 21 9 Department of Neuroscience, University of Pennsylvania; Philadelphia, PA 19104, 22 USA. 23 10 Wellcome Sanger Institute; Wellcome Genome Campus, Hinxton, Cambridge, CB10 24 1SD, UK. 25 11 College of Health and Life Sciences, Hamad Bin Khalifa University; Doha, Qatar. 26 27 # Equal contribution (co-first) 28 † These authors jointly supervised this work 29 *Correspondence and requests for materials should be addressed to A.S. (email: 30 [email protected]) or L.R.S. (email: [email protected]; 31 twitter: @saraivalab) Page 1 of 34 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.16.448475; this version posted June 16, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Segura, Abou Moussa et al Running title: 3D transcriptomics of the mouse nose 32 Abstract 33 The sense of smell helps us navigate the environment, but its anatomical logic remains 34 unknown. The spatial location of odorant receptor genes (Olfrs) in the nose is widely 35 thought to be independent of the structural diversity of the odorants they detect. 36 Using spatial transcriptomics, we created a genome-wide 3D atlas of the mouse 37 olfactory mucosa (OM), and identified key genes differentially expressed in space. 38 Expression maps reveal that Olfrs are distributed in a continuous and overlapping 39 fashion over five broad zones in the OM. The spatial locations of Olfrs correlate with 40 the mucus solubility of the odorants they recognize. Thus, we provide direct evidence 41 for the chromatographic theory of olfaction, and elucidate the basic logic for the 42 peripheral representation of smell. 43 44 One-Sentence Summary 45 Chemical solubility is organized topographically in the mouse nose 46 47 Keywords 48 Spatial transcriptomics; 3D; olfaction; olfactory epithelium; RNA-seq; odorant 49 Page 2 of 34 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.16.448475; this version posted June 16, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Segura, Abou Moussa et al Running title: 3D transcriptomics of the mouse nose 50 Main text 51 The functional logic underlying the topographic organization of primary receptor 52 neurons and their receptive fields is well-known for all sensory systems, except 53 olfaction (1). The mammalian nose is constantly flooded with odorant cocktails. 54 Powered by a sniff, air enters the nasal cavity, until it reaches the olfactory mucosa 55 (OM). There, myriad odorants activate odorant receptors (Olfrs) present in the cilia of 56 olfactory sensory neurons (OSNs), triggering a complex cascade of events that 57 culminate in the brain and result in odor perception (1, 2). In mice, most mature 58 OSNs express a single allele of one out of ~1100 Olfr genes (Olfrs) (3-6). Olfrs 59 employ a combinatorial strategy to detect odorants, which maximizes their detection 60 capacity (4, 7). OSNs expressing the same Olfr share similar odorant response profiles 61 (4, 7), and drive their axons to the same glomeruli in the olfactory bulb (8-10). Thus, 62 Olfrs define functional units in the olfactory system, and function as genetic markers 63 to discriminate between different mature OSN subtypes (6, 11). 64 Another remarkable feature of the OSN subtypes is their spatial distribution in the 65 OM. Early studies postulated that OSNs expressing different Olfrs are spatially 66 segregated into four broad areas within the OM called ‘zones’ (12, 13), but recent 67 work has suggested up to nine overlapping zones (14-16). The functional relevance of 68 these zones is not fully understood. These studies, combined, have sampled only 69 ~10% of the intact mouse Olfr repertoire. We do not currently understand the full 70 complexity of the OM and, most importantly, lack an unbiased and quantitative 71 definition of zones. In effect, the exact number of zones, their anatomical boundaries, 72 their molecular identity and their potential functional relevance are yet to be 73 determined. 74 One hypothesis is that the topographic distribution of Olfrs/OSN subtypes evolved 75 because it plays a key role in the process of Olfr choice in mature OSNs and/or in 76 OSN axon guidance (17, 18). An alternative hypothesis is that the spatial organization 77 of Olfrs/OSN subtypes is tuned to maximize the detection and discrimination of 78 odorants in the peripheral olfactory system (12). Interestingly, the receptive fields of 79 mouse OSNs vary with their spatial location (19), which in some cases correlate with 80 the patterns of odorant sorption in the mouse OM – this association was proposed as 81 the ‘chromatographic hypothesis’ decades before the discovery of the Olfrs (20), and later 82 rebranded as the ‘sorption hypothesis’ in olfaction (21, 22). While some studies lend 83 support to these hypotheses (reviewed in (23)), others question their validity (24, 25). 84 Thus, the logic underlying the peripheral representation of smell in the peripheral 85 olfactory system still remains unknown, and it is subject of great controversy (23, 26). 86 Spatial transcriptomics, which combines spatial information with high-throughput 87 gene expression profiling, has started to expand our knowledge of complex tissues, 88 organs, or even entire organisms (27-31). In this study, we employed a spatial 89 transcriptomics approach to create an 3D map of gene expression of the mouse nose, Page 3 of 34 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.16.448475; this version posted June 16, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Segura, Abou Moussa et al Running title: 3D transcriptomics of the mouse nose 90 and combined it with machine learning and chemoinformatics to reveal the 91 anatomical logic of smell in mouse. 92 93 A high-resolution spatial transcriptomic map of the mouse olfactory mucosa 94 We adapted the RNA-seq tomography (Tomo-seq) method (30) to create a spatially 95 resolved genome-wide transcriptional atlas of the mouse nose. We obtained 96 cryosections (35 μm) collected along the dorsal-ventral (DV), anterior-posterior (AP), 97 and lateral-medial-lateral (LML) axes (n=3 per axis) of the OM (Fig. 1A), and 98 performed RNA-seq on individual cryosections (gMethods). After quality control (see 99 Methods, fig. S1A-D and table S1), we computationally refined the alignment of the 100 cryosection along each axis, and we observed a high correlation between biological 101 replicates (see Methods and fig. 1B). Hence, we combined the three replicates into a 102 single series of spatial data including 54, 60 and 56 positions along the DV, AP and 103 LML axis, respectively (see Methods and Fig. 1C). On average, we detected >18,000 104 genes per axis, representing a total of 19,249 unique genes for all axes combined (Fig. 105 1D). Molecular markers for all canonical cell types known to populate the mouse OM 106 were detected among all axes (Fig. 1E), and were expressed at the expected levels (6). 107 Next, we verified the presence of a spatial signal with the Moran's I (32) (fig. S1E). 108 Given the left/right symmetry along the LML axis (see Methods; Fig. 1C), the data 109 was centered and averaged on the two sides (see Methods) – henceforth the LML axis 110 will be presented and referred to as the lateral-medial (LM) axis. We could reproduce 111 the expression patterns in the OM for known spatial markers, including the 112 dorsomedial markers Acsm4 and Nqo1 (33, 34), and the complementary ventrolateral 113 markers Ncam2 and Reg3g (35, 36) (Fig. 1F, fig. S1F). 114 115 Spatial differential gene expression analysis identifies cell type-specific 116 expression patterns and functional hotspots in the OM 117 Having established the accuracy of our dataset, we investigated cell-type specific 118 patterns.