sign in sign up
<<
Home , Glomerulus (olfaction), Hylidae

bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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-ND 4.0 International license.

Multi-glomerular projection of single neurons is conserved among

Lukas Weiss1,5,*, Lucas D. Jungblut2, Andrea G. Pozzi2, Barbara S. Zielinski3, Lauren A. O'Connell4, Thomas Hassenklöver¹ and Ivan Manzini1,*

1 Department of Physiology and Molecular Biomedicine, University of Giessen, 35392 Giessen, Germany 2 Departamento de Biodiversidad y Biología Experimental, IBBEA-CONICET, Universidad de Buenos Aires, C1428EGA Buenos Aires, Argentina 3 Department of Integrative Biology, University of Windsor, N9B 3P4 Windsor, Ontario, Canada 4 Department of Biology, Stanford University, 94305 Stanford, California, USA 5 Lead Contact * Correspondence: [email protected] (L.W.), [email protected] (I.M.)

single glomeruli. Contrastingly, axonal projections of the axolotl (salamander) branch extensively before entering up to six distinct glomeruli. Receptor neuron labeled in species (, Bufonidae, Hylidae and Dendrobatidae) predominantly bifurcate before entering a glomerulus and 59% and 50% connect to multiple glomeruli in larval and post-metamorphotic , respectively. Independent of developmental stage, lifestyle and adaptations to specific habitats, it seems to be a common feature of olfactory receptor neuron axons to frequently bifurcate and connect to multiple glomeruli. Our study challenges the unbranched concept as a universal vertebrate feature and it is conceivable that also later diverging vertebrates deviate from it. We propose that this unusual wiring logic evolved around the divergence of the terrestrial lineage from its aquatic ancestors and could be the basis of an alternative Individual receptor neurons in the peripheral way of odor processing. olfactory organ extend long axons into the forming with projection Keywords: olfaction, glomeruli, axonal wiring, neurons in spherical neuropil regions, called evolution, sensory system, fish, frog, salamander glomeruli. Generally, odor map formation and odor processing in all vertebrates is based on the Introduction assumption that receptor neuron axons exclusively connect to a single glomerulus without any axonal Vertebrates are equipped with a sophisticated branching. We comparatively tested this hypothesis to detect relevant chemical in multiple fish and amphibian species by applying sparse cell electroporation to trace single olfactory information about their environment. Throughout receptor neuron axons. Sea lamprey (jawless fish) vertebrate evolution, a progressive segregation and zebrafish (bony fish) support the unbranched into parallel olfactory pathways takes place [1–3]. axon concept, with 94% of axons terminating in While the peripheral olfactory organ of fishes

1 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

consists of a single olfactory surface [4–6, but see many animals have long been known to extend 7], the mammalian system is segregated into multiple into spatially distinct glomeruli several anatomically and functionally distinct [23]. However, it is still widely unclear whether subsystems [2,8]. A first bipartition of the the synaptic input that a single PN in the AOB olfactory periphery into a main olfactory receives from multiple glomeruli contains the epithelium (MOE) and a vomeronasal organ information conveyed by different or by the same (VNO) coincides with the evolution of the first type of vomeronasal receptor [24,25]. , the amphibians [9]. However, primordial structures that could potentially be The basic olfactory wiring principles have long homologous to the VNO have been identified in been assumed to be uniform among vertebrates. earlier diverging vertebrates like lungfish [10] and The first vertebrate species that has been found to lamprey [7]. violate the rule of an unbranched ORN axon innervating a single glomerulus in the main OB Chemical detection in the various sensory was the African clawed frog Xenopus laevis [26]. epithelia is relying on the expression of olfactory The majority of examined axons were shown to receptor proteins in the dendritic cilia or be connecting to more than one glomerulus in microvilli of olfactory receptor neurons (ORNs) larval animals [26] and this alternative pattern [11], with the two major receptor gene families was retained after metamorphosis [27]. It remains being the OR-type olfactory receptor genes and elusive whether this multi-glomerular wiring is a the vomeronasal receptor genes [12,13]. In the specific adaptation of the secondarily aquatic main olfactory system of rodents, each ORN Xenopus or if it is a more conserved evolutionary expresses a single OR-type olfactory receptor feature also present in other vertebrate lineages. [14,15] and sends a single, unbranched axon to the olfactory bulb (OB) via the Here, we report that bifurcating ORN axons and (ON). In the main OB, an axon terminally multi-glomerular innervation are not a particular branches in the confines of a single dense adaptation of Xenopus laevis, but a conserved neuropil structure, a glomerulus. All axons of feature throughout the order Anura ( and ORNs equipped with the same olfactory receptor toads). ORN axon tracings in four ecologically type coalesce onto one or very few glomeruli [16– diverse frog species in pre- and post- 18]. Each glomerulus in the main OB is thus metamorphotic animals showed that this believed to relay the information of a single OR- alternative olfactory wiring scheme is type olfactory receptor to the postsynaptic independent of developmental stage and of projection neurons (PNs), distinguished as mitral habitat. We could also show that multi-glomerular and tufted cells in rodents [1]. This constitutes the innervation of single ORN axons are the idea of the chemotopic organization of the rodent predominant pattern in the axolotl salamander, main OB [19]. While mammalian PNs extend which suggests that this feature might be present their single primary into one sole in all amphibians. Contrastingly, both the main glomerulus, PNs in fish, amphibians and reptiles olfactory system of the sea lamprey (jawless fish) often bear several primary dendrites connecting to as well as the olfactory system of zebrafish multiple glomeruli [for review see 20]. (teleost fish) follow the unbranched ORN axon paradigm with a single ORN axon only arborizing In contrast to the wiring logic employed by the within the confines of a single glomerulus. We main olfactory system, all vomeronasal receptor propose that the unusual wiring logic found in neurons (VRN) in the VNO expressing the same amphibians evolved around the divergence of the type of vomeronasal receptor converge onto ~15- terrestrial tetrapod lineage from its aquatic 30 glomeruli in the mammalian accessory ancestors and forms the basis of an alternative olfactory bulb (AOB) [21,22]. PNs in the AOB of way of odor processing.

2 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

Results common in the olfactory system of other aquatic vertebrates. We traced single ORNs from the ORN axons in fish have less branching points of the fully aquatic post- than amphibian axons larval sea lamprey (P. marinus, jawless fish), the zebrafish (D. rerio, bony fish), the axolotl (A. The morphology of an ORN axon in vertebrates is mexicanum, urodela) and the larval clawed frog generally described as an unbranched projection (X. tropicalis, anura; phylogenetic overview in terminating in fine arborizations within a single Figure 1B) to their axon terminals in the glomerulus of the OB [28]. It was already glomeruli of the OB. No experiments were reported that this principle does not apply to the conducted in the accessory olfactory system. wiring scheme in the secondarily aquatic African clawed frog [26,27]. We investigated whether this Axon tracings of the four species differed alternative projection pattern could be more substantially in their general branching structure

Figure 1: ORN axons in the OB of different aquatic vertebrates. A) Single ORN axons of the juvenile sea lamprey (P. marinus; grey), zebrafish (D. rerio; blue), axolotl (A. mexicanum; magenta) and larval clawed frog (X. tropicalis; green) show different levels of branching complexity. Sea lamprey and zebrafish axons are unbranched until they reach their terminals and only have a limited number of subbranches. Axolotl and frog axons branch in proximity to the ON and exhibit more subbranches. Axon tracings are shown from the transition between ON and OB until their terminals (upper panel). Dotted white line indicates the outline of the OB. The lower panel shows representative 3D reconstructions of two axons for each species. The first reconstruction of each species depicts the ORN axon shown in the upper panel. B) The four examined species cover a broad evolutionary period from the divergence of the jawed vertebrates from their jawless ancestors to the emergence of the first tetrapods. All four species lead a fully aquatic lifestyle. OB olfactory bulb, ON, olfactory nerve, ORN olfactory receptor neuron, P posterior, A anterior, L lateral, M medial.

3 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

(Figure 1A). ORN axons in the OB of the sea the zebrafish axons have even shorter and fewer lamprey and the zebrafish showed similarity with terminal arborizations than the lamprey axons. In the pattern reported for rodents [28]. A long, contrast, the ORN axons of the two amphibian unbranched axon projects towards the glomerular species bifurcate shortly after entering the OB, layer of the OB where it terminally arborizes. projecting several sub-branches into the While both species follow this common feature, glomerular layer, where each sub-branch

Figure 2: Number of axonal branching points and relative position of the first axonal bifurcation differ in fishes and amphibians. A) Branching points of individual ORN axons and their relative position between the transition from ON to OB (0%) and their most distal axon terminal in the glomerular layer of the OB (100%) are shown. All dots on a vertical line depict the positions of all branching points of a single reconstructed axon. ORN axons of sea lamprey (n = 6; grey), zebrafish (n = 10; blue), axolotl (n = 10, magenta) and of the clawed frog (n = 10; green) are shown. B) Quantitative comparison of the total amount of ORN axonal branching points in each species. Each dot represents a single ORN axon. The black line indicates the median, the white dotted line the mean amount of branching points for axons of each species. Axolotl ORN axons have significantly more branching points than lamprey (p = 0.0058) and zebrafish axons (p = 0.00001). The ORN axons of the western clawed frog are significantly more branched than the zebrafish axons (p = 0.0013). C) Species comparison of the first axonal bifurcation of ORN axons. The relative position of the first bifurcation between the transition from ON to OB (0%) and their most distal axon terminal (100%) are shown. The relative position of the first branching point is closer to the axon terminals in the glomerular layer in the OB of both fish species. The axolotl axons branch in immediate proximity of the ON, significantly different from the fish axons (lamprey, p = 0.014 and zebrafish p = 0.000001). Frog axons also branch closer to the ON, significantly different from the zebrafish (p = 0.014). A representative axonal reconstruction is shown for each species. Statistical significance was tested using Kruskal-Wallis rank sum test followed by Dunn’s multiple comparison post-hoc test with Holm-Bonferroni correction. OB olfactory bulb, ON olfactory nerve, ORN olfactory receptor neuron, GLOM glomerular layer. 4 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

arborizes again. Quantifying the bifurcations of from the ones in lamprey or axolotl axons, but each single axon along the distance from its entry different from zebrafish (p = 0.014, Figure 2C). point in the OB to the axon terminals, we found There are substantial differences in the general significant differences between fishes and ORN axon architecture between the four amphibians (Figure 2A and B). Sea lamprey examined species. Our results clearly indicate that axons have on average 6.5 ± 4.5 terminal ORN axon bifurcations before reaching the arborizations (n = 6). This pattern is not glomerular layer cannot be attributed to an aquatic significantly different from the zebrafish axons habitat, since this feature is absent both in the sea that show even fewer arborizations (3 ± 1.4, n = lamprey and the zebrafish. In contrast, our 10). Zebrafish axons were found to have the least tracings suggest that this alternative branching complex structure, with the highest amount of pattern of bifurcations rostral to the glomerular arborizations for a single axon being five. The layer could be linked more specifically to the axolotl displays a significantly different pattern amphibian lineage, since it was found to be from both the lamprey and zebrafish axons (p = present in both a salamander and a frog species. 0.0058 and 0.00001 respectively, Figure 2B), the ten examined axons bifurcate 29.7 ± 7.7 times on Multi-glomerular ORN axons are present only average, with one axon even branching 42 times. in amphibians Axonal tracings obtained from the of the clawed frog (23 ± 8.1, n = 10) also showed a While it was reported for the rodent main significantly higher degree of branching when olfactory system that single ORN axons project compared to the zebrafish (p = 0.0013). They did into a single glomerulus [17,18], 86% of ORNs in not show significantly different branching points the main OB of larval X. laevis were connected to than lamprey (p = 0.08) or axolotl axons. more than one glomerulus [26]. So far, comparative data of other vertebrate species on In addition to the amount, also the spatial this matter are missing. To classify uni- and multi- distribution of branching points seems to be glomerular ORN axons in the various animal differently organized in fishes and amphibians species, we used an algorithm identifying (Figure 2C). The point of origin for distance glomerular clusters based on the spatial density of measurements was set as the transition between branching and endpoints of the reconstructed ON and OB. While most fish ORN axonal axonal structures. projections only arborize close to their terminals in the glomerular layer, amphibian axons start to We found that out of all fish axons (lamprey: n = bifurcate much closer to the ON and in the nerve 6, zebrafish: n = 10), only one zebrafish axon was layer (Figure 2C). The first lamprey axon classified as multi-glomerular, while the bifurcation happens around 359 ± 139 µm after remaining axons projected into a single target entering the OB, which measures 61 ± 12% of the structure and are thus considered uni-glomerular distance from the origin to the furthest terminal (Figure 3A and B). Only three fish axons (one point. The average unbranched axon segment of lamprey axon and two zebrafish axons) had a the zebrafish tracings was 184 ± 39 µm (81 ± single extra-glomerular branching point (average 12%), axolotl axons first branch at 85 ± 33 µm extra-glomerular branching points lamprey: 0.2 ± (18 ± 5%) and the larval clawed frog axons at 94 0.4, zebrafish: 0.2 ± 0.4), while all other axons ± 40 µm (45 ± 18%). There is a significant only started to arborize within the target difference between the relative position of the glomerular cluster (Figure 3C). The average first bifurcation in axolotl axons and axons of distance from the transition between ON and OB both fish species (lamprey: p = 0.014, zebrafish: p to the glomeruli measured 401 ± 111 µm in sea = 0.000001). The position of the first bifurcations lamprey and 194 ± 41 µm in zebrafish axons. In in Xenopus axons are not statistically different the axolotl, all except for one axon (n = 10)

5 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

Figure 3: Lamprey and zebrafish ORN axons connect to a single glomerulus, amphibian ORN axons are multi-glomerular. A) The representative ORN axonal reconstructions of the lamprey (grey) and the zebrafish (blue) both connect to a single glomerulus (dotted circle) and have no bifurcations prior to entering the glomerulus. The representative axolotl ORN axon (magenta) branches four times (black stars) before connecting to three glomeruli, the frog axon (green) has a single extra-glomerular branching point and innervates two separate glomeruli. B) Population analysis of all examined axons, each stacked bar represents the percentual share of uni- and multi-glomerular axons (two to six glomeruli) of one of the four species. All lamprey ORN axons (n = 6) and 90% (n = 10) of zebrafish axons were classified as uni- glomerular by the DBSCAN algorithm. On the contrary, most amphibian axons were classified as multi-glomerular (axolotl 90%, clawed frog 70%), with a varying number of innervated glomeruli. C) Quantitative comparison of the amount of extra-glomerular branching points of all axons of each species. Each dot represents a single ORN axon. The black line indicates the median, the white dotted line the mean amount of extra-glomerular branching points. Axolotl axons have a significantly higher amount of extra- glomerular branching points than lamprey (p = 0.0002) and zebrafish (p = 0.00001). Statistical significance was tested using Kruskal-Wallis rank sum test followed by Dunn’s multiple comparison post-hoc test with Holm-Bonferroni correction. OB olfactory bulb, ON, olfactory nerve, ORN olfactory receptor neuron. followed a multi-glomerular output pattern (Figure 3A and B), with four axons even The anuran olfactory system (represented by the innervating six distinct glomerular structures. The clawed frog), showed a more heterogeneous average axonal distance from the entry point in wiring pattern. Three out of ten tracings were the OB to the glomeruli was 336 ± 109 µm. classified as uni-glomerular, while seven axons Axolotl axons bifurcate 5.1 ± 1.9 times on were multi-glomerular. In comparison with the average before reaching their target glomeruli axolotl, Xenopus axons projected into maximally (Figure 3C), which is significantly different from three glomeruli, while the majority (n = 6) both lamprey and zebrafish (p = 0.0002 and displayed a bi-glomerular wiring pattern (Figure 0.00001 respectively). 3A and B). The mean axonal distance from the nerve to the glomeruli was 162 ± 43 µm, the

6 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

structures branched 1.6 ± 1.0 times before ending tracing experiments in anurans with more or less in glomerular clusters (Figure 3C). The branching water-independent adult lifestyles to test whether density inside a single glomerular cluster was not the alternative olfactory wiring is an olfactory significantly different between lamprey (6.3 ± adaptation of aquatic frogs and tadpoles or habitat 4.7), zebrafish (2.6 ± 1.1) and axolotl (8.5 ± 7.5). independent. To account for the diverse ecology Xenopus glomeruli were more densely packed of the anurans, we examined the following with sub-branches (12.8 ± 4.8) compared to species in addition to the aquatic X. tropicalis. zebrafish (p = 0.00009), which displayed the Larval Rhinella arenarum are vegetarian grazers lowest bifurcation density, often with only one and adults are terrestrial. Both granulatus branching point within a glomerulus. Our results and Ranitomeya imitator have support the hypothesis that multi-glomerular carnivorous/omnivorous tadpoles, their adults are innervation of single ORN axons is not a wiring arboreal and terrestrial, respectively. The major feature exclusive to X. laevis but seems to be difference between these two species is that common among other aquatic amphibian species Ranitomeya provides extensive parental care and of both salamanders and frogs. On the other hand, the number of tadpoles per parent pair is much uni-glomerular wiring is the prevalent - almost lower than in Scinax, Rhinella and Xenopus exclusive - pattern in fishes, which strongly (Figure 4) [29]. resembles the rodent wiring logic. The tracings of ORN axons in R. arenarum, S. The alternative wiring pattern in anurans is granulatus and R. imitator tadpoles showed independent of developmental stage and similar wiring and branching properties to those ecology found in X. tropicalis tadpoles. In Figure 5A two representative axonal reconstructions per species The alternative wiring pattern of multi-glomerular are shown. White stars indicate extra-glomerular ORN axons has so far only been described in bifurcations and white dotted circles highlight amphibians that live a water-bound lifestyle [27]. glomerular clusters. The number of branching While all amphibian larvae are dependent on an points prior to entering the glomeruli was similar aquatic habitat, most adult frogs leave the water in all four species (X. tropicalis: 1.6 ± 1.0, n = 10, after metamorphosis [29]. We conducted ORN R. arenarum: 2.4 ± 1.8, n = 8, S. granulatus: 1.4 ±

Figure 4: Overview about ecology and phylogeny of the four anuran species examined in this study. The clawed frog Xenopus tropicalis belongs to the Pipidae family and is classified as a Mesobatrachian, an evolutionarily more basal frog species. The other three species belong to the evolutionarily more 'modern' frogs, the Neobatrachians, and to the families of Bufonidae, Hylidae and Dendrobatidae. The diverse ecology of both the adult frogs (above) and their larval offspring (below) are summarized next to the drawings. The tree is pruned from the anuran tree in [48], which originally included 3309 species.

7 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

Figure 5: Comparison of the ORN axon morphologies in different larval and juvenile frog species with different lifestyles. A) Two representative ORN axon reconstructions are shown for tadpoles of each of the anuran species X. tropicalis, R. arenarum, R. imitator and S. granulatus. Each axon was reconstructed starting from the transition between ON and main OB to the most distal axon terminal. Multiple axonal branches innervate one or two glomeruli (white dotted circles), and generally have one or two extra-glomerular branching sites (white stars). B) Two representative ORN axon reconstructions are shown for post-metamorphotic terrestrial juveniles of R. arenarum and arboreal S. granulatus. Axonal structures are comparable to their larval counterparts shown in A, one or two glomeruli are innervated by two or more axonal subbranches. C) ORN axons were classified into five branching types (A and B uni-glomerular, C - E multi-glomerular). Types are schematically shown below, black stars indicate extra-glomerular branching points. The number of axons belonging to each

8 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

branching type is shown in the stacked bar plot above. We included all ORN axons of tadpoles (X. tropicalis, green, n = 10; S. granulatus, light yellow, n = 8; R. arenarum, light brown, n = 8; R. imitator, red, n = 6) and juveniles (S. granulatus, dark yellow, n = 7; R. arenarum, dark brown, n = 5). The counts of axons per type for each species and developmental stage are marked according to the color legend below. D) Relative length of the longer and the shorter branch of individual ORN axons innervating the same glomerulus (left plot) or two different glomeruli (right plot). Each pair of dots refers to a single axon, where branch 1 is the longer branch and branch 2 is the shorter one. The thick black line visualizes the mean length relation between the two branches. In uni-glomerular axons, the shorter branch measures 76% of the length of the longer branch (n = 27, ORN axons of all species and developmental stages included). In multi-glomerular axons, the shorter branch measures on average 53% the length of the longer branch innervating a different glomerulus (n = 25, ORN axons of all species and developmental stages included). MOB main olfactory bulb, ON, olfactory nerve, ORN olfactory receptor neuron.

1.2, n = 8, R. imitator: 1.5 ± 1.6, n = 6), as was and classified them into five categories (type A to the distance of the first branching point relative to E; Figure 5C and D). Among the uni-glomerular the entire length of the axon from the transition axons, we distinguished two main types. Type A is ON-OB to the glomeruli (X. tropicalis: 45 ± 18%, characterized by a single, unbranched axon R. arenarum: 43 ± 26%, S. granulatus: 41 ± 28%, terminating in a single glomerular cluster. This is R. imitator: 45 ± 19). the prevailing type reported in rodents and also found in the fish species we examined here. Type To exclude that the alternative multi-glomerular B also terminates in a single glomerulus, but has wiring pattern is linked to the larval stages and/or at least two separate branches projecting into the their aquatic lifestyle, we conducted sparse cell same glomerular structure. Of all the axons electroporation in juveniles of the terrestrial R. reconstructed from larval and juvenile anurans (n arenarum and the arboreal S. granulatus. Figure = 44), only 11% belonged to Type A. Type B was 5B shows reconstructions of single ORN axons of more frequent, amounting to 32% of the axons. juveniles of the two species. Their morphology 57% of the axons were classified as multi- does not significantly differ from the morphology glomerular, with 50% of all axons innervating two of conspecific tadpoles nor from the other larval glomeruli and only three axons (7%) innervating axons examined. Before reaching the glomeruli, more than two glomerular end-structures (Type axons of juvenile R. arenarum and S. granulatus E). In 23% of all axons two glomerular clusters bifurcate 2.2 ± 1.5 (n = 5) and 1.6 ± 1.1 (n = 7), were innervated by a single branch each (Type C), respectively. The first bifurcations occur at 48 ± in 27 % of the tracings, at least one of the two 29% (R. arenarum) and 28 ± 16% (S. granulatus) glomeruli was innervated by more than one sub- of the total distance between the ON-OB branch (Type D, Figure 5C). Among the axons transition and the axon terminals. The only traced in the various species, all species displayed parameter that slightly differs from axons of their at least four out of the five different types. larval conspecifics (other than an approx. 1.5-fold increase in total axonal length from tadpoles to We additionally measured the differences in the juveniles) is the amount of branching points length of branches entering the same or different inside the glomerular clusters (R. arenarum: glomeruli of a single axonal structures for all the larval 5.7 ± 1.4, juvenile 3.8 ± 0.9, S. granulatus: anuran species. We found 27 cases (20 in larvae, 7 larval 11.7 ± 3.1, juvenile 5.5 ± 4). in juveniles) where a single glomerulus was innervated by at least two separate axonal All four anuran species and ontogenetic stages branches (Figure 5D, left plot). By subtracting the were heterogeneous with regard to the number of shortest innervating branch from the longest, we glomerular clusters that are innervated by a single measured a branch length difference of 46 ± 71 axon and the proportion between uni- and multi- µm in tadpoles and 56 ± 27 µm in juveniles. On glomerular axons. We categorized the axonal average, the shortest branch had 76 ± 16% the structures based on recurring branching patterns length of the longer one. This value was

9 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

consistent between tadpoles (75 ± 16%) and Emergence of the bifurcation of ORN axon juveniles (80 ± 12%). In all multi-glomerular coincides with the first tetrapods axons (n = 25, larvae 19, juveniles 6), we measured the branch length difference between In an attempt to put our findings into an the branches innervating the nearest and the evolutionary context, we compared the wiring furthest glomerulus (Figure 5D, right plot). In properties of the fishes and urodela to the anuran larval anurans, the difference was 58 ± 46 µm, in juveniles, to give an overview about the juveniles 154 ± 214 µm. The distance to the presumably mature system of the animals. Since glomerulus closer to the nerve layer was 53 ± 30 the axolotl is a neotenic salamander, the main % of the distance to the glomerulus most distant olfactory system of the one to two month old from the nerve layer (tadpoles: 48 ± 31%; larva is assumed to be a mostly developed system juveniles: 67 ± 25%). that does not undergo drastic changes until sexual maturity is reached. The majority of the fish The collected data indicates that ORN axonal axons (75%) are uni-glomerular and unbranched projections in larval and juvenile amphibians are prior to entering the glomeruli (Figure 6, blue much more heterogeneous than what has been bars), which is in accordance with the prevailing reported in rodents and what we found in fishes. vertebrate wiring principle. The juvenile anurans In amphibians, multi-glomerular innervation is are the most heterogeneous group. Only the retained throughout their developmental stages minority of axons follow the unbranched axon and does not seem to be linked to a specific principle (8%). Most axons branch at least once lifestyle or habitat. before terminating in one or more glomeruli. 42% of the axons innervate one glomerulus with more

Figure 6: Multi-glomerular innervation pattern is a conserved feature of amphibians and could have emerged with the evolution of the first tetrapods. A) ORN axons of fish species (lamprey and zebrafish, blue bars, n = 16), juvenile frogs (R. arenarum and S. granulatus, yellow bars, n = 12) and the axolotl salamander (magenta bars, n = 10) were classified into five branching types, (A and B uni-glomerular, C-E multi- glomerular). Types are schematically shown, black stars indicate extra- glomerular branching points, dotted circles indicate glomeruli. The relative number of axons belonging to each branching type for each group is shown in the bar plots. The majority of fish axons follow the unbranched axon pattern (type A) and the majority of salamander ORN axons innervate three or more glomeruli (type E). Frog axons are more heterogeneous with the most represented types being bi-glomerular (type C) and uni- glomerular with at least one extra-glomerular branching point (type B). B) One possible evolutionary scenario for the emergence of the multi-glomerular ORN projections could be that the unbranched, uni-glomerular ORN axon is a basic vertebrate trait (*1) and has secondarily evolved into a multi-glomerular alternative wiring logic in the amphibian lineage (*2) and independently also in the accessory system of mammals (*3). C) In a second scenario, the basic unbranched ORN wiring logic (*4) changed to the alternative bifurcating axon logic in vertebrates at the transition from an aquatic to a terrestrial habitat (*5). In mammals, this alternative logic then segregated to the accessory system, while the main olfactory system again followed the unbranched axon wiring logic (*6). The phylogenetic tree was modified from [9].

10 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

than two separate branches, 50% innervate more between ORN axons and PN dendrites within the than two glomeruli (Figure 6, yellow bars). Of the glomeruli is established and whether there are salamander axons, 80% innervate more than three extensive pruning mechanisms taking place glomerular structures (Figure 6, magenta bars). during development. Experiments in newborn None of the axolotl axons followed the rabbits [30] and mice [31,32] have shown that unbranched one axon-one glomerulus principle. exuberant ORN axon growth in the main OB during early development is subsequently pruned. Our results strongly imply an alternative principle Contrastingly, other studies conducted in neonatal of odor processing on the level of the OB and that rats [28] and zebrafish embryos [33] show that the emergence of this principle coincides with the ORN axons arborize to their final morphology divergence of the first terrestrial tetrapods from without erroneous targeting. Tenne-Brown and their aquatic ancestors. Even though data from the colleagues demonstrate the occurrence (15% of smallest order of modern amphibians (Caecilians) all axons) of extra-glomerular bifurcations and is still elusive, axonal bifurcations and multi- multi-glomerular connections of single ORN glomerular innervation seem to be conserved in axons in neonatal mice, but only until two weeks all amphibians regardless of their developmental after birth [32]. Of the above mentioned studies, stage and habitat. only Klenoff and colleagues report the occurrence of axons arborizing in two multiple glomeruli in Discussion rats after the initial pruning phase is over, yet only as a very rare exceptions (<0.1% of axons) [28]. Multi-glomerular innervation is absent in fishes and present in all developmental stages In contrast, we found that multi-glomerular of amphibians wiring is preserved even after metamorphosis is finished in all examined amphibians. 50% of the The consensual hypothesis regarding olfactory axons analyzed in juveniles of R. arenarum and S. wiring in the vertebrate main olfactory system is granulatus innervated more than one glomerulus established on the idea that all ORNs expressing and 92% of the axons bifurcated before entering one allele of the olfactory receptor gene repertoire the glomerular structures. A study conducted in transmit information into only one or two the main OB and the AOB of larval and post- glomeruli in the main OB via an unbranched metamorphotic Xenopus laevis [27] has yielded axon. This leads to the formation of a precise very similar results, showing that neither pre- odotopic map, where each glomerulus is part of a glomerular bifurcations nor multi-glomerular unique olfactory unit that encodes the information innervation can be solely attributed to the larval detected by a specific receptor type [16–18]. In stages. However, in accordance with the results this study we show that ORN axons of post-larval shown by Marcucci and colleagues [31], we sea lamprey and zebrafish clearly follow the noticed a reduction of the total number of unbranched axon principle postulated for all branching points from larval to post- vertebrates [16–18]. Fish axons solely arborize metamorphotic animals. Even though the number within their target-glomerulus and only one axon of extra-glomerular bifurcations (larval 1.7 ± 1.4, out of 16 has been found to innervate two distinct n = 32; juvenile 1.8 ± 1.3, n = 12) remained glomeruli in the zebrafish. In contrast, 57% of constant, the arborizations inside a single ORN axons labeled in anurans show connections glomerulus decreased in the juveniles (larval 10.1 to multiple glomeruli (90% in the salamander ± 5.1; juvenile 4.8 ± 3.1). While the reduction of species examined). branches inside a glomerulus can be explained by pruning mechanisms during development, multi- It is still under debate how the precise wiring glomerular innervation in the amphibian main OB is not strictly linked to an immature larval stage.

11 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

The alternative wiring logic in amphibians has VNO is expressing more recently diverging V2R parallels to the rodent AOB genes [1,35,36].

In the rodent accessory olfactory system, the In line with this heterogeneous receptor gene axons of VRNs expressing the same vomeronasal expression in the amphibian MOE, we found receptor type converge onto ~15-30 glomeruli, quite a heterogeneity in ORN axon branching constituting a more vague spatial code than in the patterns. Different branching types could be main OB [13,21,22]. In contrast to the main connected to different receptors. The mix of ORs system, the postsynaptic partners of the ORN and VRs in the MOE of amphibians could explain axons, the PNs, extend several dendrites into the heterogeneity of branching patterns, while in multiple glomeruli. It is still unclear whether they the more segregated rodent system, axonal integrate between input of VRNs expressing bifurcations and a more vague spatial glomerular different or the same vomeronasal receptors code are only found in the accessory system. It [21,22,24,25]. Several attempts have been made has already been shown that olfactory receptors to unravel the AOB wiring logic from VRN axons are involved in axon guidance and the formation to PN dendrites in rodents. A study investigating of the glomerular map [19,37]. Still, the genetically labeled receptor neurons expressing a expression of vomeronasal receptors in the MOE single type of V1R or V2R in mice gives evidence of amphibians is unlikely to be the only cause for for a homotypic connectivity model. In this axonal bifurcations and the alternative wiring model, a single PN extends its dendrites into pattern. V1Rs and V2Rs are already expressed in multiple glomeruli innervated exclusively by the the sensory epithelium of fishes, but we could same vomeronasal receptor type [24]. Another show that ORN axon bifurcations are absent in study supported a selective heterotypic both lampreys and zebrafish. It is more plausible connectivity model in which a single PN receives that the mechanisms by which the receptors information from glomeruli that get sensory input influence axonal guidance could have changed not from a single vomeronasal receptor type, but over evolutionary time and that the new wiring from closely related receptors within a receptor principle has only emerged after the divergence of subfamily [25]. the first tetrapods from the aquatic ancestors.

The discrepancies between the wiring principles Multi-glomerular ORN innervation is in the rodent main OB and AOB suggest that mirrored by PN morphology in vertebrates there might be different aspects of odor and invertebrates information extracted by the respective subsystems. It has additionally been reported that Just like the PNs in the rodent AOB, PNs in the approx. 10% of single VRN axons in the nerve main OB of amphibians and reptiles extend layer of the AOB split into several sub-branches, multiple dendrites into multiple glomeruli, where reaching out to multiple glomeruli [34]. The they terminate in dendritic tufts [20]. In the case wiring pattern of amphibians described in our of amphibians, the multi-glomerular PN study therefore resembles the rodent accessory morphology seems to be mirroring the multi- system rather than the main system. While the glomerular ORN axons described in this study. In rodent system shows a clear separation into the the sea lamprey, the morphology of the single OR-type receptor expressing MOE and the V1R ORN axons we found show a net like branching and V2R expressing VNO, this segregation is structure within single glomeruli. This incomplete in amphibians. Among the OR-type morphology is also mirrored in the uni-glomerular receptors and other receptor gene families, the arborizations of the lamprey PNs [38]. In many MOE of Xenopus was shown to express V1Rs as teleost fish species, it is known that PNs extend well as early diverging V2Rs, while the Xenopus many primary dendrites into multiple glomeruli.

12 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

However, this was only shown to be the minority is also present in the axolotl salamander and in in zebrafish [39]. In accordance with these results, anuran tadpoles of four ecologically distinct we also show that single zebrafish ORN axons families (Pipidae, Bufonidae, Hylidae and mostly (90%) terminate in a single glomerulus. It Dendrobatidae) as well as terrestrial post- is intriguing to speculate, whether teleost fishes metamorphotic frogs. From our results, we can equipped with multi-glomerular PNs also display conclude that the multi-glomerular ORN wiring multi-glomerular ORN wiring. pattern is independent of or adult ecology and seems to be a feature derived from the A similar mirror image in the connectivity pattern common ancestor between frogs and salamanders, between ORN axon and postsynaptic PNs occurs since it is also present in axolotl. In both the in the evolution of the in jawless and bony fish species we examined, we orthopteran insects [40]. It was shown that in could not find any clear signs of the presence of more basal orthopterans (e.g. the great green bush this alternative wiring principle. ), single receptor neuron axons are innervating a single glomerulus and a single PN From an evolutionary perspective, our results extends its dendrite into one sole glomerulus, suggest several plausible scenarios, two of which resembling the mammalian main system. will be discussed here. First, the axonal Contrastingly, in later diverging orthopterans bifurcations of ORNs could have developed (locusts and grasshoppers) both single receptor independently in amphibians (*2, Fig 6B), and in neurons as well as PNs connect to multiple the VNO of mammals (*3, Fig 5B), with the glomeruli – a pattern similar to amphibians [40– ancestral vertebrate trait being the one-to-one 42]. The multi-glomerular pattern in locusts is wiring logic (*1, Fig 5B). The presence of multi- linked to the formation of a high number of glomerular wiring structures in the orthoptera microglomeruli (~2500 glomeruli), while the one- supports the idea that this alternative wiring could to-one pattern in basal orthopterans is linked to have a functional advantage in odor processing fewer number of bigger glomeruli (~40) [41]. A and has thus evolved independently multiple similar evolution towards microglomeruli could times [40]. In the second scenario, the multi- have taken place among vertebrates: the sea glomerular pattern could have evolved around the lamprey has very large but few glomeruli (41-65) divergence of the terrestrial tetrapod lineage from [43], the zebrafish has approx. 140 quite its aquatic ancestors (*5, Fig 5B), coinciding with differently sized glomeruli [44] and Xenopus the first occurrence of the VNO [9] and a huge laevis has a larger number of smaller glomeruli, expansion and reshaping of the OR-type gene ~350 in the main OB [45] and ~340 in the AOB family [1,47]. Given that the odorant receptor is [46]. The concept of the branched and multi- directly influencing axon targeting and the glomerular ORN axon has thus developed at least formation of glomeruli [16,37], it could twice independently, however its putative potentially also have an impact on axonal functional implications remain elusive. branching. Newly emerging receptors or axon branching principles after the tetrapod divergence Different evolutionary scenarios for the could be responsible for this alternative wiring emergence of the alternative wiring logic and be more widespread in the MOE and VNO of earlier diverging tetrapods (i.e. amphibians), and Branched receptor neuron axons with multi- more focally expressed in the VNO of mammals glomerular innervation have been shown in about (*6, Fig 5B). 10% of VRNs in the mouse AOB [34] and as a predominant type in the main OB [26] and the Taken together, our results show that the AOB of Xenopus laevis tadpoles, as well as adults prevailing idea of an unbranched ORN axon [27]. In this study we show that this wiring logic arborizing only in a single glomerulus cannot be

13 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

generalized for vertebrates any longer. While Experimental Procedures jawless and bony fishes display the one-to-one wiring logic that is shown for the rodent main 1. Animals olfactory system, our results indicate that bifurcating ORN axons and multi-glomerular Fish species: wiring are a general feature of the amphibian All sea lampreys (Petromyzon marinus) used in olfactory system. This alternative wiring scheme this study were post-larval transformer stages is neither linked to larval stages, nor to a specific (metamorphic stage seven, both sexes) from the habitat or lifestyle of amphibians. It cannot be Connecticut River, Turner Falls, MA. Animals excluded that this feature is even more common were captured and supplied by United States amongst vertebrates and that it constitutes the Geological Survey Conte Anadromous Fish basis of an alternative way for odor processing. Research Laboratory. They were kept in 420 l tanks at 6°C ± 1°C under static renewal Acknowledgements conditions until used. Zebrafish (Danio rerio, both sexes) were kept in oxygenated water tanks We thank all the present and past members of the at room temperature. Manzini laboratory for fruitful discussion and input, especially Thomas Offner and Sara Joy Amphibian species: Hawkins. We thank Anja Schnecko for dedicated Albino larvae of axolotl (Ambystoma mexicanum) animal care and Gianfranco Grande and Eva were obtained from the Ambystoma Genetic Fischer for support to set up experiments. L.W. Stock Center at the University of Kentucky, USA. was granted a travel fellowship (JEBTF-180809) They were kept in oxygenated water tanks (20°C) by the company of Biologists Limited and the and fed with red mosquito larvae. Animals used Journal of Experimental Biology to visit the for this study were both sexes, 5-6 weeks of age. O'Connell lab. This work was supported by DFG Wild type Xenopus tropicalis larvae were bred Grant 4113/4-1 and in part, by Award Number and reared at the Institute of Animal Physiology, S10RR02557401 from the National Center for University of Giessen. They were kept in water Research Resources (NCRR). Its contents are tanks at a water temperature of 25°C and fed with solely the responsibility of the authors and do not algae. Animals used for this study were stages 49- necessarily represent the official views of the 52 after Niewkoop and Faber [49]. NCRR or the National Institutes of Health. Ranitomeya imitator tadpoles of stages 28-31 after Gosner [50] were bred in the laboratory Author Contributions colony at the Biology Department of Stanford University, Palo Alto, CA, USA. Individual Conceptualization, L.W., T.H. and I.M.; tadpoles were kept separately after hatching at a Investigation, Formal Analysis, Visualization and water temperature of 25°C. Writing – Original Draft, L.W.; Writing – Review Rhinella arenarum larvae (Gosner stages 29-31) & Editing, L.W., L.D.J., A.G.P., B.S.Z., L.A.O., and juvenile animals were obtained by in vitro T.H. and I.M.; Funding Acquisition and fertilization from a colony at the Faculdad de Resources, L.D.J., A.G.P., B.S.Z., L.A.O., T.H. Ciencias Exactas y Naturales of the University of and I.M.; Supervision, T.H. and I.M. Buenos Aires. larvae (Gosner stages 31-35) and juveniles were collected from Declaration of Interest the wild (semi- temporary ponds formed in the surroundings of the Campus of the University of The authors declare no competing interests. Buenos Aires). All larval animals were kept in tanks of dechlorinated water at a temperature of

14 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

22°C and fed ad libitum with chard leaves. brainstem. The whole olfactory bulbs were Juvenile animals were kept in glass terraria and dissected out of the tissue. Samples were fed with flies. immediately imaged in Ringer's solution or fixed in 4% PFA in PBS for one hour and imaged later. All experiments performed followed the guidelines of Laboratory animal research of the 4. Image acquisition and processing Institutional Care and Use Committee of the Multi-channel image stacks (z-resolution of 1 µm) University of Windsor (AUPP 14-05), University were acquired using multi-photon microscopy at of Buenos Aires (CD: 316/12, Protocol #22), the an excitation laser-wavelength of 780 nm (upright University of Göttingen (33.9-42502-04- Nikon A1R-MP and upright Leica SP5 12/0779), University of Gießen, (GI 15/7, multiphoton microscopes). Brightness and 932_GP) and Stanford University (APLAC- contrast of the stacks were adjusted using ImageJ 33016). [52]. Since there was no dye introduced into the tissue with a blue-wavelength emission, we used 2. Sparse cell electroporation the blue-wavelength detector (400 – 492 nm) to Animals were anesthetized using 0.02% MS-222 image tissue auto-fluorescence. Pigmentation (ethyl 3-aminobenzoate methanesulfonate; Sigma- derived autofluorescence was mathematically Aldrich) in tap water until completely subtracted from the other emission channels using unresponsive and placed on a wet tissue paper the image-calculator function implemented in under a stereomicroscope. ORNs were stained ImageJ. Image data are presented as maximum using micropipettes pulled from borosilicate glass intensity projections along the z-axis of the virtual capillaries (Warner instruments, resistance 10-15 image stacks. MΩ) filled with fluorophore-coupled dextrans (Alexa dextran 488 and 594, 10 kDa, Life 5. Axonal reconstructions Technologies) diluted at a concentration of 3mM Individual neuronal morphology was in saline Ringer (Amphibian Ringer (mM): 98 reconstructed semi-automatically by defining NaCl, 2 KCl, 1 CaCl2, 2 MgCl2, 5 glucose, 5 Na- branching and endpoints of the axonal structure pyruvate, 10 Hepes, pH 7.8; Lamprey Ringer from the acquired image stacks in Vaa3D [53]. (mM): 130 NaCl, 2.1 KCl, 2.6 CaCl2, 1.8 MgCl2, Only ORN axons that could be traced from the 4 Hepes, 4 dextrose, 1 NaHCO3, pH 7.4, beginning of the nerve layer in the olfactory bulb Zebrafish Ringer (mM): 131 NaCl, 2 KCl, 20 until their terminals in the glomerular layer were NaHCO3, 1.25 KH2PO4, 2.5 CaCl2, 2 MgSO4, 10 reconstructed. The reconstructed neuron-trees dextrose, 5 Na-pyruvate, 10 Hepes, pH 7.2). The were sorted by the sort_swc algorithm pipettes were mounted on the electrode bearing implemented in Vaa3D to define the root of the headstage of an Axoporator 800A (Axon axon as first node of the structure. instruments, Molecular Devices) and inserted into the main nasal cavity of the animals. A 500 ms 6. Structural analysis and identification of train of square voltage pulses (50 V, single pulses glomeruli 300 µs at 200–300 Hz) was triggered to stain Axonal reconstructions were analyzed and neurons [27,51]. This protocol was repeated at quantified using custom written Python scripts in different positions inside the main nasal epithelia. Jupyter notebook. Spatial distribution of The animals were left to recover. branching points and length of branches and sub- branches were assessed based on the 3. Olfactory bulb whole mount preparation reconstructions. All data presented in regard to Animals were anesthetized again (as described branch length is measured in distances along the above) three days after electroporation and killed axonal structure (in µm) or displayed as length- by severing the spinal cord at the level of the ratio in %.

15 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

The number of glomeruli innervated by an axonal separately for each species and ontogenic stage structure was determined using the DBSCAN (in µm: P. marinus 90, D. rerio 20, A. mexicanum algorithm (Density-Based Spatial Clustering of 20, X. tropicalis 11, R. arenarum pre- Applications with Noise) implemented in the metamorphotic 10, post-metamorphotic 15, S. sklearn machine learning package written for granulatus pre-metamorphotic 14, post- Python [54]. The algorithm clusters together metamorphotic 16, R. imitator 10). These values points that are in close spatial proximity to a lot of were chosen based on previously reported neighboring points (glomerular cluster) while it glomerular size (P. marinus: [55]; D. rerio: marks points far away from its closest neighbors [44] or our own glomerular tracing experiments as low-density noise. All branching- and (all amphibian species). endpoints of a neuron-tree structure from the root to the terminals of the axons were used as input 7. Statistical Analysis points. Based on the algorithm, an axon terminal Averaged data are presented as mean ± standard was considered a glomerular cluster if at least deviation. Statistical significance was determined three points were in close spatial proximity. Blunt by Kruskal-Wallis rank sum test followed by axonal endings without terminal bifurcations were Dunn’s multiple comparison post-hoc test, unless marked as noise outliers. To account for the otherwise stated. To control familywise error rate different glomerular size between the various for multiple comparisons, a Holm-Bonferroni animals used for this study, the minimal distance correction was applied. for points to be considered a cluster was chosen

References

1. Bear, D.M., Lassance, J.-M., Hoekstra, H.E., and 8. Breer, H., Fleischer, J., and Strotmann, J. (2006). Datta, S.R. (2016). The evolving neural and genetic The : Multiple olfactory subsystems. architecture of vertebrate olfaction. Curr. Biol. 26, Cell. Mol. Life Sci. 63, 1465–1475. R1039–R1049. 9. Eisthen, H.L. (1992). Phylogeny of the vomeronasal 2. Munger, S.D., Leinders-Zufall, T., and Zufall, F. system and of receptor cell types in the olfactory (2009). Subsystem organization of the mammalian and vomeronasal epithelia of vertebrates. Microsc. sense of smell. Annu. Rev. Physiol. 71, 115–140. Res. Tech. 23, 1–21. 3. Eisthen, H.L. (1997). Evolution of vertebrate 10. González, A. (2010). Lungfishes, like tetrapods, olfactory systems. . Behav. Evol. 50, 222–233. possess a vomeronasal system. Front. Neuroanat. 4, 130. 4. Hansen, A., and Zeiske, E. (1998). The peripheral olfactory organ of the zebrafish, Danio rerio: an 11. Mombaerts, P. (1999). Seven-transmembrane ultrastructural study. Chem. Senses 23, 39–48. proteins as odorant and chemosensory receptors. Science 286, 707–711. 5. Sato, Y. (2005). Mutually exclusive glomerular innervation by two distinct types of olfactory 12. Buck, L., and Axel, R. (1991). A novel multigene sensory neurons revealed in transgenic zebrafish. J. family may encode odorant receptors: A molecular Neurosci. 25, 4889–4897. basis for odor recognition. Cell 65, 175–187. 6. Hamdani, E.H., and Døving, K.B. (2007). The 13. Dulac, C., and Axel, R. (1995). A novel family of functional organization of the fish olfactory system. genes encoding putative pheromone receptors in Prog. Neurobiol. 82, 80–86. mammals. Cell 83, 195–206. 7. Ren, X., Chang, S., Laframboise, A., Green, W., 14. Malnic, B., Hirono, J., Sato, T., and Buck, L.B. Dubuc, R., and Zielinski, B. (2009). Projections (1999). Combinatorial receptor codes for odors. from the accessory olfactory organ into the medial Cell 96, 713–723. region of the olfactory bulb in the sea lamprey (Petromyzon marinus): A novel vertebrate sensory structure? J. Comp. Neurol. 516, 105–116.

16 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

15. Chess, A., Simon, I., Cedar, H., and Axel, R. (1994). Neurosci. 33, 17247–17252. Allelic inactivation regulates olfactory receptor gene expression. Cell 78, 823–834. 28. Klenoff, J.R., and Greer, C.A. (1998). Postnatal development of olfactory receptor cell axonal 16. Mombaerts, P. (2006). Axonal wiring in the mouse arbors. J. Comp. Neurol. 390, 256–267. olfactory system. Annu. Rev. Cell Dev. Biol. 22, 713–737. 29. Wells, K.D. (2007). The ecology and behavior of amphibians (University of Chicago Press). 17. Ressler, K.J., Sullivan, S.L., and Buck, L.B. (1994). Information coding in the olfactory system: 30. Yilmazer-Hanke, D.M., Hudson, R., and Distel, H. Evidence for a stereotyped and highly organized (2000). Morphology of developing olfactory axons epitope map in the olfactory bulb. Cell 79, 1245– in the olfactory bulb of the rabbit (Oryctolagus 1255. cuniculus): A Golgi study. J. Comp. Neurol. 426, 68–80. 18. Vassar, R., Chao, S.K., Sitcheran, R., Nuñez, J.M., Vosshall, L.B., and Axel, R. (1994). Topographic 31. Marcucci, F., Maier-Balough, E., Zou, D.-J., and organization of sensory projections to the olfactory Firestein, S. (2011). Exuberant growth and bulb. Cell 79, 981–991. formation of olfactory sensory neuron axonal arborizations. J. Comp. Neurol. 519, 3713–3726. 19. Mombaerts, P., Wang, F., Dulac, C., Chao, S.K., Nemes, A., Mendelsohn, M., Edmondson, J., and 32. Tenne-Brown, J., and Key, B. (1999). Errors in Axel, R. (1996). Visualizing an olfactory sensory lamina growth of primary olfactory axons in the rat map. Cell 87, 675–686. and mouse olfactory bulb. J. Comp. Neurol. 410, 20–30. 20. Dryer, L., and Graziadei, P.P.C. (1994). dendrites: a comparative approach. Anat. Embryol. 33. Dynes, J.L., and Ngai, J. (1998). Pathfinding of (Berl). 189, 91–106. olfactory neuron axons to stereotyped glomerular targets revealed by dynamic imaging in living 21. Belluscio, L., Koentges, G., Axel, R., and Dulac, C. zebrafish embryos. Neuron 20, 1081–1091. (1999). A map of pheromone receptor activation in the mammalian brain. Cell 97, 209–220. 34. Larriva-Sahd, J. (2008). The accessory olfactory bulb in the adult rat: A cytological study of its cell 22. Rodriguez, I., Feinstein, P., and Mombaerts, P. types, neuropil, neuronal modules, and interactions (1999). Variable patterns of axonal projections of with the main olfactory system. J. Comp. Neurol. sensory neurons in the mouse vomeronasal system. 510, 309–350. Cell 97, 199–208. 35. Date-Ito, A., Ohara, H., Ichikawa, M., Mori, Y., and 23. Ramón y Cajal, S. (1911). Histologie du systeme Hagino-Yamagishi, K. (2008). Xenopus V1R nerveux de l’homme et des vertebres, Vol. 2. vomeronasal receptor family is expressed in the Maloine, Paris, 887–890. main olfactory system. Chem. Senses 33, 339–346. 24. Del Punta, K., Puche, A., Adams, N.C., Rodriguez, 36. Syed, A.S., Sansone, A., Nadler, W., Manzini, I., I., and Mombaerts, P. (2002). A divergent pattern of and Korsching, S.I. (2013). Ancestral amphibian sensory axonal projections is rendered convergent v2rs are expressed in the main olfactory epithelium. by second-order neurons in the accessory olfactory Proc. Natl. Acad. Sci. U. S. A. 110, 7714–7719. bulb. Neuron 35, 1057–1066. 37. Feinstein, P., and Mombaerts, P. (2004). A 25. Wagner, S., Gresser, A.L., Torello, A.T., and Dulac, contextual model for axonal sorting into glomeruli C. (2006). A multireceptor genetic approach in the mouse olfactory system. Cell 117, 817–831. uncovers an ordered integration of VNO sensory inputs in the accessory olfactory bulb. Neuron 50, 38. Green, W.W., Basilious, A., Dubuc, R., and 697–709. Zielinski, B.S. (2013). The neuroanatomical organization of projection neurons associated with 26. Nezlin, L.P., and Schild, D. (2005). Individual different olfactory bulb pathways in the sea olfactory sensory neurons project into more than lamprey, Petromyzon marinus. PLoS One 8, one glomerulus in Xenopus laevis tadpole olfactory e69525. bulb. J. Comp. Neurol. 481, 233–239. 39. Fuller, C.L., Yettaw, H.K., and Byrd, C.A. (2006). 27. Hassenklöver, T., and Manzini, I. (2013). Olfactory Mitral cells in the olfactory bulb of adult zebrafish wiring logic in amphibians challenges the basic (Danio rerio): morphology and distribution. J. assumptions of the unbranched axon concept. J. Comp. Neurol. 499, 218–230.

17 bioRxiv preprint doi: https://doi.org/10.1101/788133; this version posted October 1, 2019. 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-ND 4.0 International license.

40. Hansson, B.S., and Stensmyr, M.C. (2011). ancient continental vicariance and recent oceanic Evolution of . Neuron 72, 698–711. dispersal in amphibians. Syst. Biol. 63, 779–797. 41. Ignell, R., Anton, S., and Hansson, B.S. (2001). The 49. Nieuwkoop, D.P., and Faber., J. (1994). Normal antennal lobe of orthoptera - Anatomy and Table of Xenopus laevis (Daudin). Garland, New evolution. Brain. Behav. Evol. 57, 1–17. York. 42. Ernst, K.D., Boeckh, J., and Boeckh, V. (1977). A 50. Gosner, K.L. (1960). A simplified table for staging neuroanatomical study on the organization of the anuran embryos and larvae with notes on central antennal pathways in insects - II. identification. Herpetologica 16, 183–190. Deutocerebral connections in Locusta migratoria and Periplaneta americana. Cell Tissue Res. 176, 51. Haas, K., Sin, W.C., Javaherian, A., Li, Z., and 285–308. Cline, H.T. (2001). Single-cell electroporation for gene transfer in vivo. Neuron 29, 583–591. 43. Frontini, A., Zaidi, A.U., Hua, H., Wolak, T.P., Greer, C.A., Kafitz, K.W., Li, W., and Zielinski, 52. Schindelin, J., Arganda-Carreras, I., Frise, E., B.S. (2003). Glomerular territories in the olfactory Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., bulb from the larval stage of the sea lamprey Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Petromyzon marinus. J. Comp. Neurol. 465, 27–37. Fiji: an open source platform for biological image analysis. Nat. Methods 9, 676–682. 44. Braubach, O.R., Fine, A., and Croll, R.P. (2012). Distribution and functional organization of 53. Peng, H., Ruan, Z., Long, F., Simpson, J.H., and glomeruli in the olfactory bulbs of zebrafish (Danio Myers, E.W. (2010). V3D enables real-time 3D rerio). J. Comp. Neurol. 520, 2317–2339. visualization and quantitative analysis of large-scale biological image data sets. Nat. Biotechnol. 28, 45. Manzini, I., Heermann, S., Czesnik, D., Brase, C., 348–353. Schild, D., and Rössler, W. (2007). Presynaptic protein distribution and odour mapping in glomeruli 54. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, of the olfactory bulb of Xenopus laevis tadpoles. V., Thirion, B., Grisel, O., Blondel, M., Eur. J. Neurosci. 26, 925–934. Prettenhofer, P., Weiss, R., Dubourg, V., et al. (2011). Scikit-learn: Machine Learning in Python. J. 46. Nezlin, L.P., and Schild, D. (2000). Structure of the Mach. Learn. Res. 12, 2825–2830. olfactory bulb in tadpoles of Xenopus laevis. Cell Tissue Res. 302, 21–29. 55. Green, W.W., Boyes, K., Mcfadden, C., Daghfous, G., Auclair, F., Zhang, H., Li, W., Dubuc, R., and 47. Niimura, Y., and Nei, M. (2006). Evolutionary Zielinski, B.S. (2017). Odorant organization in the dynamics of olfactory and other chemosensory olfactory bulb of the sea lamprey. J. Exp. Biol. 220, receptor genes in vertebrates. J. Hum. Genet. 51, 1350–1359. 505–517. 48. Pyron, R.A. (2014). Biogeographic analysis reveals

18