It is known since 1996 that mouse odorant receptors (ORs) are involved in determining the positions of the sites of coalescence of axons of olfactory sensory neurons (OSNs)—the thousands of glomeruli in the olfactory bulb. But the molecular and cellular mechanisms of OR-mediated axonal coalescence into glomeruli remain unclear. A model was proposed in 2006–2009 whereby OR-derived cAMP signals, rather than direct action of OR molecules, determine the target destinations (glomeruli) of OSNs in the bulb. This model hypothesizes that OR-derived cAMP signals determine the expression levels of neuropilin 1 (Nrp1) in OSN axon termini; that levels of Nrp1 in glomeruli form a gradient from anterior-low to posterior-high throughout the bulb; and that these Nrp1 levels mechanistically determine anterior-posterior patterning of glomeruli. Here, we describe the first independent evaluation of the Nrp1 model since it was formulated a decade ago. We tested the model for the well-characterized mouse OR M71 using our gene-targeted mouse strains, which are publicly available. In contradiction to the model, we observed a variety of configurations for the M71 glomeruli in the conditional Nrp1 knockout. We then reassessed the model for the original OR transgene with which the model was developed, using the same publicly available mouse strains. We discovered that glomerular positions do not undergo the simple anterior shift that has been reported in the conditional Nrp1 knockout for this OR transgene. Taken together, our findings do not support the Nrp1 model for the anterior-posterior patterning of glomerular positions in the olfactory bulb.
In the mouse, each olfactory sensory neuron expresses one of ∼1100 odorant receptor genes. The odorant receptor determines to which odorants the neuron responds physiologically and in which glomerulus of the olfactory bulb its axon terminates. A model was proposed 10 years ago whereby intracellular signals derived from the odorant receptor determine the level of neuropilin 1, which in turn determines the position of the glomerulus along the anterior-posterior axis of the bulb. We provide the first test of this model, for a well-characterized odorant receptor and separately with the original mouse strains that led to the formulation of the model. Our results do not support the neuropilin 1 model of anterior-posterior patterning of glomerular positions in the olfactory bulb.
In the mouse, the 1100 odorant receptor (OR) genes (Buck and Axel, 1991) determine to which odorants an olfactory sensory neuron (OSN) responds (Bozza et al., 2002) and in which of the 3600 glomeruli of the olfactory bulb (Richard et al., 2010) the axon of an OSN terminates (Mombaerts, 2006). A gene-targeted replacement of the OR coding region provided the first evidence that ORs are mechanistically involved in determining the positions of glomeruli (Mombaerts et al., 1996). A set of experiments with a transgenic mouse MOR23 promoter defined by Vassalli et al. (2002), but expressing a rat OR (I7) instead of mouse MOR23, led to a model whereby OR-derived cAMP signals, rather than direct action of OR molecules, determine the target destinations (glomeruli) of OSNs (Imai et al., 2006, 2009; Luo, 2015). This model proposes that OR-derived cAMP signals regulate the transcription of genes encoding axon guidance molecules, which then guide positioning of glomeruli along the anterior-posterior axis of the bulb. OSNs expressing rat OR I7 from a transgenic mouse MOR23 promoter form, in the medial aspect of the bulb, a novel, artificial glomerulus, which appears to be homogeneous in that it is not coinnervated by OSNs expressing an endogenous mouse OR (Imai et al., 2006, 2009). Immunostaining in bulb sections for the axon guidance molecule neuropilin 1 (Nrp1) in mice at postnatal day 14 (PD14) revealed, in the medial aspect of the bulb, a gradient of Nrp1 expression in a row of 20 glomeruli that encompass the novel, artificial glomerulus, with low glomerular Nrp1 expression at the anterior end and high expression at the posterior end (Imai et al., 2006). By crossing a conditional Nrp1 knockout (Gu et al., 2003) with a transgene-expressing rat OR I7 from a mouse MOR23 promoter along with Cre recombinase, an anterior glomerular shift was reported at PD14; conversely, when Nrp1 was overexpressed in these OSNs from a transgene, a posterior glomerular shift was reported (Imai et al., 2009). Agonist-independent G protein–coupled receptor activity was later proposed to regulate anterior-posterior targeting of axons of OSNs via Nrp1 (Nakashima et al., 2013).
Curiously, when formulating the Nrp1 model, data were reported only for the projection sites (glomeruli) in the medial aspect of the bulb (Imai et al., 2006, 2009). It is well established that glomeruli for a given OR, with a few exceptions (Strotmann et al., 2000), are found in both medial and lateral aspects of the bulb, including the MOR23 glomeruli (Vassalli et al., 2002; Zapiec and Mombaerts, 2015). Thus, it is not known what impact the conditional Nrp1 knockout has on the lateral glomerulus in mice that express rat OR I7 from a transgenic mouse MOR23 promoter.
Here, we describe the first independent evaluation of the Nrp1 model since it was formulated a decade ago. We tested the model with a genetic strategy based on the well-characterized mouse OR, M71. The M71 glomeruli reside posteriorly and are Nrp1+ (Dibattista and Reisert, 2016). Our experimental design makes use of our publicly available mouse strains that carry gene-targeted mutations. We avoided the use of small transgenes, because these are notorious for their line-to-line variability (Vassalli et al., 2002; Vassalli et al., 2011). Moreover, these gene-targeted strains enable us to study glomeruli that are formed by the coalescence of axons of OSNs that express a mouse OR from its endogenous locus. We also created a mosaic situation whereby Nrp1+ M71+ OSNs coexist in the same mouse with Nrp1– M71+ axons, capitalizing on the unusual modality of monoallelic OR gene expression. We observed substantial variability in the configurations of M71 glomeruli in 88 bulbs examined of 47 conditional Nrp1 knockout mice, with ectopic M71 glomeruli forming anteriorly and dorsally. We then reassessed the Nrp1 model by analyzing the publicly available mouse strains with which it was formulated originally (Imai et al., 2009). Surprisingly, we could not confirm the simple anterior shift of the medial glomerulus that was reported in the conditional Nrp1 knockout. The lateral glomerulus typically undergoes a ventral shift and less of an anterior shift. Both medially and laterally, more than one glomerulus is present in the conditional Nrp1 knockout. The domain of the bulb that these multiple glomeruli occupy can be described as a sector (medially) or belt (laterally). Taken together, our results pose a challenge to the Nrp1 model (Imai et al., 2006, 2009; Luo, 2015). A revision of this model becomes imperative.
Materials and Methods
Generation of a gene-targeted mouse strain carrying a Nrp1 floxed allele
We constructed a targeting vector by long-range PCR to generate a conditional mutation in the Nrp1 gene by flanking exon 2 with loxP sites. The linearized targeting vector was electroporated in the parental embryonic stem cell line E14 of the 129P2/OlaHsd genetic background. G418-resistant clones were screened for homologous recombination by Southern blot hybridization of genomic DNA using an external probe. The FRT-flanked neo selectable marker was excised by Flp-mediated recombination. Our Nrp1 floxed strain is publicly available from The Jackson Laboratory (Bar Harbor, ME) as #6707, official strain name STOCK Nrp1<tm1.1Mom>/MomJ. The design of this targeted mutation is similar to the published Nrp1 floxed strain (Gu et al., 2003); in both strains, exon 2 of Nrp1 is excised upon Cre recombination.
Mouse strains from other laboratories
The Cre reporter strain R26-tauGFP41, carrying the gene-targeted mutation ROSA26-CAGS-τGFP (Mayer et al., 2010) with official allele name Gt(ROSA)26Sor<tm1(CAG-Mapt/GFP)Uboe>, was a generous gift from Uli Boehm (University of Saarland, Homburg, Germany). The gene-targeted Nrp1 floxed strain (Gu et al., 2003) was obtained from The Jackson Laboratory as #5247, official strain name B6.129(SJL)-Nrp1<tm2Ddg>/J. The transgenic strain I7-Cre-YFP Tg, also referred to as I7-ires-Cre, I7(WT)-Cre, or I7-ires-Cre-ires-gap-YFP (Imai et al., 2009), was obtained from the RIKEN BRC (National Bio-Resource Project of the MEXT, Japan) as #RBRC02932, official strain name C57BL/6-Tg(Olfr16-Olr226,-Cre,-EYFP)1Hsak.
Mouse husbandry and experimentation
Mice were maintained in specific pathogen–free conditions in individually ventilated cages of the Tecniplast green line. Mice received ad libitum γ-irradiated ssniff V1124-727 feed (ssniff, Soest, Germany). Nesting, bedding, and enrichment were provided as nestpak, Datesand Grade 6 (Datesand, Manchester, UK). Mouse experiments were carried out in accordance with guidelines of the National Institutes of Health and the German Animal Welfare Act, European Communities Council Directive 2010/63/EU, and the institutional ethical and animal welfare guidelines of the Max Planck Institute of Biophysics and the Max Planck Research Unit for Neurogenetics. Approval came from the IACUC of The Rockefeller University (New York, NY); the Regierungspräsidium Darmstadt (Germany); and the Veterinäramt of the City of Frankfurt (Germany).
Immunohistochemistry on sections
Mice (females and males) were anesthetized by intraperitoneal injection of ketamine HCl and xylazine (210 and 10 mg/kg body weight, respectively), and perfused intracardially with ice-cold PBS followed by 4% paraformaldehyde in PBS. The brain was immersed in 15% sucrose in PBS and 30% sucrose in PBS, each overnight at 4ºC on a shaker. After cryoprotection, the brain was trimmed and frozen in optimal cutting temperature compound (Tissue-Tek) on dry ice in ethanol. Serial horizontal sections encompassing the olfactory bulbs were generated with a Leica CM3050 S cryostat, set at 12-µm thickness. The sections were washed with 1× PBS and blocked with 10% normal donkey serum and 0.3% Triton X-100 in 1× PBS for 1 h at room temperature. After the blocking step, sections were incubated in 3% bovine serum albumin and 0.3% Triton X-100 in 1× PBS overnight at 4˚C with primary antibodies: rabbit anti-Adcy3 (#sc-588, 1:1000; Santa Cruz Biotechnology, Dallas, TX) and goat anti-Nrp1 (#AF566, 1:100; R&D Systems, Minneapolis, MN). Sections were then incubated for 1.5 h at room temperature with secondary antibodies: donkey anti-rabbit IgG Alexa Fluor 647 (#711-606-152, 1:500; Jackson ImmunoResearch Laboratories, West Grove, PA) for rabbit anti-Adcy3, and donkey anti-goat IgG Alexa Fluor 546 (#A11056, 1:1000; Thermo Fischer Scientific, Waltham, MA) for goat anti-Nrp1. Nuclear staining was done with 4′,6-diamidino-2-phenylindole (DAPI; #D1306, 1:10,000; Thermo Fisher Scientific) after the washing steps. Sections were imaged with a Zeiss LSM 710 confocal microscope (Oberkochen, Germany).
Immunolabeling of whole mounts
Samples were processed according to the iDISCO protocol (Renier et al., 2014). Primary antibodies were goat anti-Nrp1 (#AF566; R&D Systems) and guinea pig anti-VGLUT2 (#135 404; Synaptic Systems, Göttingen, Germany). Secondary antibodies were donkey anti-goat Alexa Fluor 488 (#A11055; Thermo Fisher Scientific) and donkey anti-guinea pig RRX (#706-295-148; Jackson ImmunoResearch Laboratories).
Serial block-face two-photon tomography
Serial block-face two-photon tomography was performed with a TissueCyte 1000 scanner (TissueVision, Cambridge, MA) equipped with a Zeiss 20× 1.0-NA objective and a Ti:Sapphire laser (Mai Tai HP DeepSee; Spectra-Physics, Santa Clara, CA). Methods were as described in Zapiec and Mombaerts (2015). Briefly, mechanical section thickness was set at 100 µm. Optical z-stacks were captured at an x,y resolution of 1.02 µm per pixel and 5 µm per z-plane, resulting in voxels of ∼5 µm3. Stacks were captured starting 50 µm below the cutting plane until a depth of 150 µm, thus spanning 100 µm in the z-axis. Stacks of images were processed using custom Python, Matlab, and ImageJ scripts provided by TissueVision. The 3D reconstruction and measurements were performed with Amira 6 (FEI, Hillsboro, OR).
Glomeruli formed by the coalescence of axons of OSNs that express the OR gene M71
The subject of our first study with a conditional Nrp1 knockout is the population of OSNs that express the OR gene M71 (M71+ OSNs), also known as Olfr151. The choice of this OR gene is motivated by extensive knowledge that has been collected over the past 15 years, including the characterization of odorous ligands for M71 (Bozza et al., 2002). We have reported 39 strains with distinct gene-targeted mutations at the M71 locus. Importantly, the strain M71-IRES-Cre (Li et al., 2004) is the only gene-targeted Cre driver strain for a mouse OR gene that is publicly available.
M71 is expressed in a few thousand OSNs at PD21 (Bressel et al., 2016). In mice that carry the gene-targeted M71-IRES-tauGFP mutation (Potter et al., 2001), M71+ OSNs express tauGFP along with M71 from bicistronic transcripts. The green fluorescent protein–positive (GFP+) axons typically coalesce into a single glomerulus (sometimes into two or three glomeruli) posteriorly each in the medial and lateral aspects of the olfactory bulb at PD21 (Fig. 1A). The positions of the M71 glomeruli are conserved. Crossing the M71-IRES-Cre strain to the Cre reporter strain R26-tauGFP41 (Mayer et al., 2010; Wen et al., 2011) phenocopies the configuration of labeled glomeruli seen in the M71-IRES-tauGFP strain (Fig. 1B). Consistent with their posterior position in the bulb, M71 glomeruli are Nrp1+ (Dibattista and Reisert, 2016).
Thus, the population of M71+ OSNs represents a convenient and robust model system served by numerous publicly available gene-targeted strains, to assess the role of Nrp1 in determining the positioning of M71 glomeruli along an anterior-posterior axis.
Ectopic anterior and ectopic dorsal glomeruli in triple-mutant n5247MCZ mice
We determined the effects of a conditional Nrp1 knockout in M71+ OSNs using the same Nrp1 floxed allele (Gu et al., 2003) that was used in Imai et al. (2009). We refer to this allele as n5247, reflecting the strain number in the catalog of The Jackson Laboratory, from which this strain is publicly available. By repeated crossing, we generated triple-mutant n5247MCZ mice. These mice are homozygous for the Nrp1 floxed allele (n5247), homozygous for the gene-targeted M71-IRES-Cre allele (MC) (Li et al., 2004), and hemizygous or homozygous for the widely used Cre reporter Z/EG transgene (Z) Novak et al., 2000).
In triple-mutant n5247MCZ mice, M71+ OSNs express Cre and become devoid of Nrp1; moreover, they become GFP+ by virtue of Cre-mediated excision of a loxP-flanked segment of the reporter transgene that results in permanent GFP expression. Examples of whole mounts of bulbs of triple-mutant n5247MCZ mice are shown for PD14 and PD21 (Fig. 1C, D): a single small GFP+ glomerulus is present anteriorly, and a single large GFP+ glomerulus, dorsally and posteriorly. Henceforward we refer to these glomeruli as “ectopic anterior” and “ectopic dorsal.” This configuration deviates drastically from the typical configuration of one or a few M71 glomeruli posteriorly in each of the medial and lateral aspects of the bulb (Fig. 1A, B).
There was substantial variability, however, in the configurations of sites of coalescence (glomeruli) of GFP+ axons when we examined a total of 45 bulbs of 23 triple-mutant n5247MCZ mice at five ages (Fig. 2A): 10 bulbs at PD0 (when the glomeruli are beginning to form), eight at PD14, 21 at PD21, two at PD45, and four at PD105. In an attempt to discern distinct categories, we identified at least four configurations of GFP+ glomeruli, which we refer to as I, II, III, and IV. Configuration I (22/45 bulbs, 49%) is the most common: it consists of a single small ectopic anterior glomerulus and a single large ectopic dorsal glomerulus and is exemplified in the bulbs shown in Fig. 1C, D. Configuration II (16 bulbs, 36%) comprises a single small ectopic anterior glomerulus and two glomeruli posteriorly. Configuration III (four bulbs, 9%) consists of a single large ectopic dorsal glomerulus but no ectopic anterior glomerulus. Configuration IV (three bulbs, 7%) has two dorsal glomeruli posteriorly but no ectopic anterior glomerulus. Bulbs with an ectopic anterior glomerulus (or a developing ectopic anterior glomerulus, at PD0) thus occur in 84% of bulbs (I + II, 38/45), and bulbs with a single large ectopic dorsal glomerulus in 58% of bulbs (I + III, 26/45). An ectopic anterior glomerulus is present in 80% of bulbs (8/10) at PD0. At PD105, the oldest age examined, there are no ectopic anterior glomeruli in any bulb (4/4). For comparison with Imai et al. (2009), who report data about PD14 mice, the configuration that we observe most frequently at PD14 is II (6/8, 75%): a single small ectopic anterior glomerulus and two posterior glomeruli.
We have described a linear relationship between the number of OSNs that express a given OR gene and the total volume of the corresponding glomeruli in the bulbs (Bressel et al., 2016). We found that in n5247MCZ mice at PD22, the ectopic anterior glomeruli contribute 20.0% (SD ± 6.5%) to the total glomerular volume per mouse, using methods outlined in Bressel et al. (2016). It follows that ∼20% of Nrp1– M71+ OSNs innervate these ectopic anterior glomeruli.
Thus, the multiple configurations of M71 glomeruli in a conditional Nrp1 knockout and the substantial variability are not consistent with the Nrp1 model for anterior-posterior patterning of glomeruli (Imai et al., 2006, 2009). This model, which was formulated based on results from mice expressing a rat OR from a transgenic mouse MOR23 promoter, does not apply to the first tested population of OSNs that express a mouse OR from its endogenous locus, M71.
Three-dimensional reconstructions of the bulbs of triple-mutant n5247MCZ mice
The variability in configurations of M71 glomeruli in the conditional Nrp1 knockout poses an unexpected experimental challenge. There are variations on a theme, and further examination may reveal additional configurations or subconfigurations. To enable a direct comparison of the positions of M71 glomeruli with and without Nrp1, we imaged bulbs using TissueCyte serial block-face two-photon tomography (Ragan et al., 2012) and generated three-dimensional (3D) reconstructions of bulbs (Zapiec and Mombaerts, 2015). We made use of the intrinsic fluorescence signal in glomeruli in four bulbs of PD21 mice homozygous for the gene-targeted M71-IRES-tauYFP mutation (Nrp1+) and in six bulbs of triple-mutant n5247MCZ mice (Nrp1–) also at PD21. Methods of image acquisition, registration, and image processing have been described in detail (Zapiec and Mombaerts, 2015). Figure 3 and Movie 1 show views of the merged right olfactory bulb, which is composed of a total of 10 left or right olfactory bulbs. The views on this merged bulb are, respectively, dorsal and comparable to Figure 1 (Fig. 3A), and dorsomedial and slightly tilted laterally (Fig. 3B) so that the posterior-medial glomeruli are more visible. Figure 3C provides close-ups. All four M71-IRES-tauYFP bulbs have a single medial glomerulus posteriorly; two have a single lateral glomerulus posteriorly, and two have two lateral glomeruli posteriorly. These configurations are consistent with what has been reported for M71 glomeruli in several gene-targeted strains at or around PD21 (for instance, Potter et al., 2001; Feinstein and Mombaerts, 2004; Feinstein et al., 2004; Zou et al., 2004). The configuration of glomeruli in the 3D reconstructions of bulbs of triple-mutant n5247MCZ mice deviates drastically from that in the M71-IRES-tauYFP mice. Four n5247MCZ bulbs show configuration I (Fig. 2A), with a single ectopic anterior glomerulus and a single ectopic dorsal glomerulus; this configuration is the most frequently seen at PD21. One n5247MCZ bulb is of configuration II (one ectopic anterior glomerulus and two glomeruli posteriorly), and another n5247MCZ bulb is of configuration III, with a single ectopic dorsal glomerulus. New information from these 3D reconstructions is that the ectopic dorsal glomeruli reside roughly halfway between the medial and lateral M71 glomeruli, and at approximately the same anterior-posterior position.
Thus, the 3D reconstructions of bulbs by serial block-face two-photon tomography confirm and extend the observations made by whole-mount epifluorescence. The multiple configurations of M71 glomeruli and the substantial variability are not consistent with the Nrp1 model for anterior-posterior patterning of glomeruli.
Quadruple-mutant NMRCZ mice
OR genes are expressed monoallelically: an OSN that expresses a given OR gene expresses one allele of it (Chess et al., 1994; Strotmann et al., 2000; Ishii et al., 2001). This monoallelic expression is extremely tight (Saraiva et al., 2015). Next, we developed a novel genetic strategy that capitalizes on this unusual modality of gene expression in the mouse. The quadruple-mutant mice NMRCZ enable a direct, mosaic comparison of Nrp1+ and Nrp1– populations of M71+ OSNs within the same mouse, as follows.
We generated a novel floxed allele of Nrp1 by flanking exon 2 with loxP sites, adopting a similar design as for the published Nrp1 floxed allele (Gu et al., 2003). We then produced two triple-mutant strains by repeated crossing: NMCZ and NMRZ. The triple-mutant strain NMCZ is homozygous for our Nrp1 floxed allele (N), homozygous for M71-IRES-Cre (MC), and hemizygous or homozygous for the Z/EG transgene (Z); it is analogous to the triple-mutant n5247MCZ strain but with our Nrp1 floxed allele. As in triple-mutant n5247MCZ mice, ectopic glomeruli also exist in triple-mutant NMCZ mice, both the large ectopic dorsal glomerulus (Fig. 4A) and the smaller ectopic anterior glomerulus (Fig. 4B). The triple-mutant strain NMRZ is homozygous for our Nrp1 floxed allele (N), homozygous for M71-IRES-tauRFP2 (MR) (Li et al., 2004), and hemizygous or homozygous for the Z/EG transgene (Z). By crossing an NMCZ mouse with an NMRZ mouse, quadruple-mutant NMRCZ offspring are obtained that are homozygous for our Nrp1 floxed allele (N), compound heterozygous at the M71 locus (M71-IRES-Cre and M71-IRES-tauRFP2; MRC), and hemizygous or homozygous for the Z/EG reporter (Z). In these quadruple-mutant mice, OSNs that express the M71-IRES-tauRFP2 allele are red fluorescent protein–positive (RFP+) and Nrp1+, and OSNs that express the M71-IRES-Cre allele are GFP+ (via expression of the Z/EG reporter) and Nrp1–.
We analyzed a total of 43 bulbs of 24 quadruple-mutant NMRCZ mice at six ages: three at PD7, eight at PD14, 13 at PD21, 11 at PD35, four at PD45, and four at PD70 (Fig. 2B). There are no bulbs of configuration II, which we had identified earlier in triple-mutant n5247MCZ mice. In addition to the configurations I (four bulbs, 9%), III (two bulbs, 5%), and IV (nine bulbs, 21%), we identified two new configurations in quadruple-mutant NMRCZ mice. In configuration V (Fig. 4C, right, and D), the most frequent configuration (26 bulbs, 60%), there is a single small ectopic anterior glomerulus anteriorly (Fig. 4E, J), a single medial glomerulus (Fig. 4G), and a single large ectopic dorsal glomerulus (Fig. 4H). The medial glomerulus is innervated clearly in a mixed fashion, but to various extents, by RFP+ axons and GFP+ axons. The position of this medial glomerulus is typical for the medial M71 glomerulus. Configuration VI (two bulbs, 4%) has the same glomeruli as configuration V plus a single lateral glomerulus, at a position that is typical for the lateral M71 glomerulus (Fig. 4C, left, and F). Altogether, in 65% of bulbs (V + VI, 28/43), the RFP+ axons (representing the M71-IRES-tauRFP2 allele, without Cre recombination) coalesce into glomeruli at positions that are typical for M71 glomeruli, and they mix in these glomeruli with GFP+ axons (representing the M71-IRES-Cre allele, with Cre recombination; Fig. 4I).
The numbers of labeled OSNs in the triple-mutant and quadruple-mutant mice are comparable to those in M71-IRES-tauGFP mice at PD21 (Fig. 5): n5247MCZ t test p = 0.7635, NMRCZ t test p = 0.0526. These numbers were determined with Abercrombie correction, as detailed in Bressel et al. (2016). We did not find a significant difference between the numbers of labeled OSNs expressing the M71-IRES-Cre or M71-IRES-tauRFP2 allele in NMRCZ mice (t test p = 0.85).
Glomeruli in posterior domains of the bulb can be Nrp1+ or Nrp1–
Nrp1 expression in OSNs is dependent on expression of the cAMP-generating enzyme adenylate cyclase Adcy3 (Dal Col et al., 2007). Most if not all mature OSNs express Adcy3. An exception is a population of chemosensory neurons in the mouse main olfactory epithelium that express the cation channel Trpc2 (Liman et al., 1999), the cyclic-nucleotide gated channel subunit Cnga2, and the guanylate cyclase Gucy1b2: they do not express Adcy3 (Omura and Mombaerts, 2014, 2015). Glomeruli of Gucy1b2+ neurons can be examined easily and specifically in mice of the gene-targeted Gucy1b2-IRES-tauGFP strain: neurons that express Gucy1b2 coexpress GFP, and the expression level is high enough to visualize the signal in glomeruli by the intrinsic fluorescence of GFP (Omura and Mombaerts, 2015). Axons of Gucy1b2+ neurons coalesce typically into three glomeruli posteriorly in the olfactory bulb (Omura and Mombaerts, 2015). The absence of Adcy3 expression in Gucy1b2+ neurons and the morphological similarity between Gucy1b2+ glomeruli and canonical glomeruli formed by the coalescence of axons of OR-expressing OSNs predict that Gucy1b2 glomeruli are Nrp1–.
On a horizontal section at an extremely ventral level through the bulb of a homozygous Gucy1b2-IRES-tauGFP mouse at 4 weeks, a single GFP+ glomerulus could be observed at a lateral position (Fig. 6A). This glomerulus was Adcy3– and Nrp1–. Interestingly, numerous glomeruli anterior to this GFP+ glomerulus were Adcy3+ but Nrp1–, thus breaking the causal link between Adcy3 and Nrp1 expression. On a horizontal section at a slightly more dorsal level, another GFP+ glomerulus at a medial position was also Adcy3– and Nrp1–, and adjacent glomeruli were Adcy3+ but either Nrp1+ or Nrp1– (Fig. 6B). At an intermediate dorsal-ventral level of this bulb, a third GFP+ glomerulus, at a lateral position, was Adcy3– and Nrp1–, but the glomeruli just anterior to it were Adcy3+ and Nrp1– or very weakly Nrp1+ (Fig. 6C).
Thus, by examining three posterior domains of a bulb that are defined by Gucy1b2+ glomeruli, we found that there is no absolute link between Adcy3 and Nrp1 expression, and that posterior glomeruli can be either Nrp1+ or Nrp1–.
Nrp1 levels in a 3D reconstruction of a bulb of a wild-type C57BL6/J mouse
It thus became necessary to take a bulb-wide view of Nrp1 levels, and to revisit the proposed anterior-posterior Nrp1 gradient (Imai et al., 2006, 2009), at the level of the entire bulb and in 3D (Zapiec and Mombaerts, 2015).
We performed whole-mount Nrp1 immunofluorescence of wild-type C57BL/6J bulbs at PD14 using the iDISCO protocol (Renier et al., 2014), and imaged and reconstructed the labeled bulbs by applying serial block-face two-photon tomography. Figure 7 shows a representative bulb. The Nrp1 antibody labels glomeruli efficiently, but it also labels OSN axons in the olfactory nerve layer (Fig. 7A), thus obscuring the glomeruli in 3D reconstructions of the bulb. Although we could assess the Nrp1 levels within individual glomeruli when examining the tomography sections, we wanted to obtain a global view of the 3D reconstructed bulb, which required segmenting the glomerular layer from the olfactory nerve layer. This segmentation was accomplished by colabeling with an antibody for VGLUT2, a robust marker of glomeruli that labels OSN axon terminals within glomeruli (Richard et al., 2010). We used the VGLUT2 signal to define the glomeruli by thresholding this signal to perform semi-automated segmentation. By masking the Nrp1 signal to display only the voxels that belong to glomeruli as defined by the VGLUT2 signal (Fig. 7B), we visualized the intensity of the Nrp1 immunofluorescence signal specifically in glomeruli, and thereby obtained clear views of Nrp1 levels in glomeruli across the 3D reconstructed bulb.
A medial view (Fig. 7C) and a lateral view (Fig. 7D) of the 3D reconstruction reveal that, generally, Nrp1low glomeruli reside anteriorly and dorsally, and Nrp1high glomeruli posteriorly. But upon closer examination, the pattern of Nrp1 levels is mosaic and patchy: there is considerable variability in staining intensity among neighboring glomeruli. Nrp1low glomeruli can be found in the immediate vicinity of Nrp1high glomeruli, as we had observed in the three posterior domains defined by Gucy1b2+ glomeruli. The olfactory bulb is often described as comprising two mirror-image maps of glomeruli (two half-bulbs; Nagao et al., 2000) such that glomeruli for a given OR can be found in positions reflected along a mirror plane. But no such symmetry was obvious when we examined the patterns of Nrp1 immunofluorescence signal intensities among the medial and lateral aspects of the bulb.
Further quantitative analyses will be required to settle definitively the hypothesis of a smooth anterior-posterior bulb-wide Nrp1 gradient, and across various possible anterior-posterior “axes.” Several pre- and postnatal ages must also be examined.
Glomeruli formed by the coalescence of axons of OSNs that express rat OR I7 from a transgenic mouse MOR23 promoter with and without Nrp1
We sought to replicate the experiments with the transgenic mice that express the rat OR I7 from a mouse MOR23 promoter along with the Cre recombinase and gap-YFP (Imai et al., 2009). This mouse strain is publicly available from the RIKEN BioResource Center and is here referred to as I7-Cre-YFP Tg. In the gap-YFP marker, the 20 N-terminal amino acid residues of GAP43 are fused to the N terminus of the yellow fluorescent protein (YFP) to target it to the plasma membrane (Moriyoshi et al., 1996). Figure 8A shows epifluorescence whole-mount images of the medial and lateral aspects of a left and a right bulb of two I7-Cre-YFP Tg mice at PD17, visualizing the intrinsic fluorescence from the gap-YFP fusion protein. We confirm and extend the observation (Imai et al., 2009) that axons expressing rat OR I7 from this transgenic mouse MOR23 promoter coalesce into a single glomerulus in the medial aspect of the bulb, and add that a single labeled glomerulus is present in the lateral aspect of the bulb.
We then crossed I7-Cre-YFP-Tg mice to gene-targeted mice carrying the Nrp1 floxed allele n5247 (Gu et al., 2003), thus reproducing the same cross used in Figure 1A of Imai et al. (2009). Figure 8B shows epifluorescence whole-mount images of the medial and lateral aspects of six bulbs of six I7-Cre-YFP Tg littermates at PD14, the same age analyzed in Imai et al. (2009). We made distinct observations for the medial and lateral aspects of these bulbs. Medially, we observed two labeled glomeruli in five of the six bulbs, and four labeled glomeruli in one bulb. One of these glomeruli appeared larger than the others, and was more anterior except in the case of the four labeled glomeruli, of which the largest was the second most anterior glomerulus. A virtual line drawn between the labeled glomeruli in a given bulb yielded various orientations or axes, which were grossly along the anterior-posterior dimension but also had a ventral or dorsal component. Laterally, we observed two to six labeled glomeruli (average 3.5), with one glomerulus larger than the others. Connecting the labeled glomeruli in a given bulb did not reveal a consistent orientation or axis. Because there is no information about the labeled glomeruli in the lateral aspect of the bulb in Imai et al. (2009), a comparison of the data is not possible.
Thus, the positions of the labeled glomeruli and their substantial variability in the I7-Cre-YFP Tg × n5247-flox cross are not consistent with the Nrp1 model for anterior-posterior patterning of glomeruli for this novel, artificial population of OSNs that express rat OR I7 from a transgenic mouse MOR23 promoter.
3D reconstructions of the bulbs of I7-Cre-YFP Tg mice with and without Nrp1
Finally, as we had done for triple-mutant n5247MCZ mice, we imaged labeled bulbs of I7-Cre-YFP Tg mice with and without Nrp1, by applying serial block-face two-photon tomography based on the intrinsic fluorescence of the gap-YFP marker and then generating 3D reconstructions of the bulbs.
Figure 9 shows one example of a reconstructed (right) bulb of a I7-Cre-YFP Tg mouse at PD14; it is the same bulb as shown in the bottom right panels of Fig. 8B. For comparison, the epifluorescence whole-mount images of the same bulb are shown again. The two labeled glomeruli in the medial aspect seen in the epifluorescence whole-mount image are clearly detectable in the 3D reconstruction. In the lateral aspect, a fifth, small labeled glomerulus is detectable in the 3D reconstruction, but is not seen in the epifluorescence whole-mount image.
Figure 10A shows a merged bulb of three I7-Cre-YFP Tg bulbs at PD14, including the bulb shown in Figure 9. The 3D reconstruction reveals a single labeled glomerulus in each the medial and lateral aspects of the bulb. The advantage of the merged bulb approach is that glomerular positions can be compared directly: we found that the three glomeruli were tightly clustered in each of the medial and lateral aspects. In sharp contrast, the merged bulb composed of eight I7-Cre-YFP Tg × n5247-flox bulbs and the three I7-Cre-YFP Tg bulbs at PD14 (thus 11 bulbs in total) revealed a broad scattering of multiple I7-Cre-YFP Tg × n5247-flox glomeruli in both the medial and lateral aspects of the bulb (Fig. 10B). Medially, the scattered glomeruli (mean 2.3, minimum 1, maximum 5) occupied a broad domain of the bulb that can best be characterized as a sector, anchored at the position of the tightly clustered glomeruli of the I7-Cre-YFP Tg bulbs and radiating anteriorly and ventrally. Laterally, the scattered glomeruli (mean 3.0, minimum 2, maximum 5) were centered at the tightly clustered glomeruli of the I7-Cre-YFP Tg bulbs; seven glomeruli underwent a dorsal shift, 14 a ventral shift, and three no shift. A ventral view of this merged bulb (Fig. 10B) revealed that the glomeruli occupied a belt that spanned the ventral extent and connected the medial and lateral domains of the I7-Cre-Tg glomeruli with Nrp1. This glomerular belt was interrupted for a segment defined by the ventral ridge of the bulb. In three I7-Cre-YFP Tg × n5247-flox bulbs at PD21, there were more labeled glomeruli (medially: mean 6, minimum 4, maximum 9; laterally: mean 5.7, minimum 3, maximum 8), and the degree of scattering both medially and laterally was increased (Fig. 10C). The distribution of glomeruli in the PD14 and PD21 bulbs can be evaluated in Movie 2.
Taken together, our observations with the cross I7-Cre-YFP Tg × n5247-flox at the same age of PD14 are not consistent with the data reported in Imai et al. (2009). Our findings do not support the Nrp1 model of anterior-posterior patterning of glomeruli in the bulb for this novel population of OSNs that express rat OR I7 from a transgenic mouse MOR23 promoter.
Naturally, any model that aspires to explain how axons of ∼1100 populations of OSNs, each expressing a distinct OR gene sort and coalesce into ∼3600 glomeruli at conserved positions in the olfactory bulb, needs to be tested for more than a single OR gene. The Nrp1 model (Imai et al., 2006, 2009; Luo, 2015) was developed for one OR (rat OR I7), which was either FLAG- or HA-epitope tagged and expressed from a small transgenic mouse MOR23 promoter, and glomeruli were examined solely for the medial aspect of the bulb. In the intervening decade, the Nrp1 model has not been extended to an endogenous mouse OR, or to any other OR. We have here sought to assess the validity of the Nrp1 model for another population of OSNs, which express mouse OR M71 from the endogenous M71 locus, using our publicly available, gene-targeted mouse strains. The configurations of M71 glomeruli that we discern in the conditional Nrp1 knockouts with two distinct Nrp1 floxed alleles are not consistent with this Nrp1 model. Moreover, we cannot confirm the simple anterior shift in medial projection sites for the population of OSNs expressing rat OR I7 from a transgenic mouse MOR23 promoter; the lateral projection sites, which were not examined (Imai et al., 2006, 2009), typically undergo a ventral shift, and less of an anterior shift.
The Nrp1 model for anterior-posterior positioning of glomeruli does not apply to the native population of OSNs expressing mouse OR M71 from the endogenous M71 locus
We observed multiple configurations and substantial variability in the patterns of labeled glomeruli when Nrp1 was knocked out conditionally in M71+ OSNs. Further examination may reveal additional configurations or subconfigurations.
We discerned four configurations in triple-mutant n5247MRZ mice, in which all M71+ OSNs are Nrp1–. In quadruple-mutant NMRCZ mice, in which half of M71+ OSNs are Nrp1+ and the other half are Nrp1–, we discern five configurations, of which three are shared with n5247MRZ mice. A common observation is the occurrence of novel, ectopic glomeruli, anteriorly and/or dorsally. The ectopic anterior glomeruli are formed by ∼20% of Nrp1– M71+ OSNs. Their far-anterior ectopic location is strikingly reminiscent of the ectopic anterior glomeruli reported by two groups in constitutive Adcy3 knockout mice, for M71 (Zou et al., 2007) and for M72 (Dal Col et al., 2007), an OR that is highly related to M71. The lack of Nrp1 expression in Adcy3 knockout mice (Dal Col et al., 2007) suggests a possible mechanistic link between OR expression, cAMP generation through Adcy3, and Nrp1 expression.
Our experiments on M71 with two Nrp1 floxed alleles, including the one used by Imai et al. (2009), are not consistent with the model, in that we did not observe a simple anterior shift of M71 glomeruli when Nrp1 was knocked out conditionally in M71+ OSNs. The single large ectopic dorsal glomerulus that we observed frequently was not shifted along an anterior-posterior axis at all, but resided halfway between the medial and lateral M71 glomeruli.
The Nrp1 model for anterior-posterior positioning of glomeruli does not apply to the novel population of OSNs expressing rat OR I7 from a transgenic mouse MOR23 promoter
During the course of this project, we realized that we needed to take a step back and seek to reproduce the core original findings with which the Nrp1 model was formulated (Imai et al., 2009). With publicly available strains, we generated the same cross of Cre driver strain and Nrp1 floxed allele, but our observations differed. Instead of a single projection site (glomerulus) shifting anteriorly at PD14, we observed multiple glomeruli formed by the coalescence of axons expressing rat OR I7 from a transgenic mouse MOR23 promoter. When viewed in a merged bulb after 3D construction, together these medial glomeruli occupied a broad domain of the bulb that can best be characterized as a sector that is anchored posteriorly on the tightly clustered medial glomeruli in the transgenic mice with Nrp1. This sector radiates anteriorly and ventrally along various vectors and is not consistent with a simple shift along an anterior-posterior “axis” of the bulb; at best, the vectors have an anterior component. This sector may reflect the domain of the bulb within which these axons, now devoid of Nrp1, are allowed to navigate. We then proceeded to flip these bulbs and examined the labeled glomeruli in the lateral aspect, which were not documented in Imai et al. (2006, 2009). We discovered that the scattering of the glomeruli is even more pronounced in the lateral than in the medial aspect. Together, these lateral glomeruli occupy a domain of the bulb that can be regarded as a belt or a band, centered at the tightly clustered lateral glomeruli in the mice with Nrp1 and extending toward the medial aspect along the ventral ridge. Seven of 24 lateral glomeruli were shifted dorsally compared with the lateral glomeruli in the mice with Nrp1, but most (14/24) were shifted ventrally; the remaining three glomeruli showed no shift. No glomeruli underwent a simple shift along an anterior-posterior “axis” of the bulb.
At PD21, the scattering of glomeruli appears to have increased, with glomeruli radiating further within the sector (medially) and the belt (laterally). We have not examined mice at older ages. There may thus be a phenomenon of dynamic instability when mechanisms of axonal coalescence of OSNs are perturbed by a conditional Nrp1 knockout in one particular population of OSNs. In this context, it must be kept in mind that the glomeruli that are formed by the coalescence of axons expressing rat OR I7 from a transgenic mouse MOR23 promoter are novel and artificial: they do not exist in wild-type, nontransgenic mice.
Possible explanations for the discrepancy in our findings with Imai et al. (2006, 2009), as far as the medial aspect of the bulb is concerned, are a higher quality of our epifluorescence whole-mount imaging and the new method of 3D reconstruction with serial block-face two-photon tomography (Zapiec and Mombaerts, 2015). Unfortunately, the lateral aspect of the bulb was not examined in Imai et al. (2006, 2009).
Assessment of Nrp1 gradients in the bulb
It is intrinsically difficult to provide convincing evidence of a Nrp1 protein gradient across the bulb, extending smoothly from anterior-low to posterior-high, at various dorsal-ventral levels, both in the medial and lateral aspects of the bulb, and at relevant ages. We have here tested the concept of a Nrp1 gradient in two ways, and with the same Nrp1 antibody as used in Imai et al. (2006, 2009). By conventional immunofluorescence, we examined three posterior domains at several dorsal-ventral levels in the vicinity of Gucy1b2+ glomeruli, which are Adcy3– and Nrp1–. We found that there are Nrp1+ and Nrp1– glomeruli in these posterior domains, thus breaking the link between Adcy3 and Nrp1 expression, and constituting numerous exceptions to the notion that posterior glomeruli are Nrp1+. Moreover, there is substantial heterogeneity within these posterior domains: some glomeruli adjacent to Gucy1b2+ glomeruli are Nrp1+ (and Adcy3+), and others are Nrp1– (but also Adcy3+). Serial block-face two-photon tomography (Zapiec and Mombaerts, 2015) is an advanced imaging method, enabling us to quantify Nrp1 immunofluorescence levels in each of the thousands of glomeruli of the bulb. With this method, we were unable to identify consistent, robust, smooth gradients of Nrp1 across glomeruli along any axis, including possible anterior-posterior axes; but the absence of evidence for a bulb-wide gradient in our hands is not evidence for its absence. Our 3D reconstructions reiterate the difficulty in the unambiguous definition of any “axis” in the olfactory bulb (Zapiec and Mombaerts, 2015): there are no landmarks or fiduciary points to determine “anterior” and “posterior” accurately and reproducibly and to the level of precision that is required to assess gradients rigorously and critically. Nonetheless, we found that, generally, anterior glomeruli are Nrp1low, and generally, the posterior bulb has glomeruli that are Nrp1high. But Nrp1 levels do not form a simple gradient from anterior-low to posterior-high throughout the bulb. Instead the Nrp1 pattern is complex and can better be described as mosaic or patchy, or even idiosyncratic, with the posterior bulb hosting glomeruli that are strongly Nrp1+ but intermingled with Nrp1low glomeruli.
Glomerular Nrp1 levels in the literature
Our observations of mosaic, patchy patterns of Nrp1 levels across glomeruli in sections and in the 3D reconstructed bulbs are consistent with the literature. An early study in rats reported that “[n]europilin-1 immunoreactivity was confined to the rostrolateral and caudal glomeruli of the main olfactory bulb. Moderate neuropilin-1 immunoreactivity was observed in most glomeruli, whereas a small number of glomeruli was either very strongly labeled or unlabeled” (Pasterkamp et al., 1998). Another group reported in adult mice that “NP-1+ olfactory axons projected selectively to glomeruli within two compartments, the lateral band … and the medial band.… NP-1+ glomeruli (which receive NP-1+ olfactory axons) were completely excluded from the anteromedial region … and the ventral region” (Taniguchi et al., 2003). A third group observed in PD5 mice that “Npn-1+ glomeruli in the rostral-most 600 µm of the olfactory bulb were variable in size and intensity of staining” and “[a]t no position along the entire rostrocaudal axis of the olfactory bulb were there detectable glomeruli containing npn-1+ axons near the ventral midline of the olfactory bulb of wild-type mice” (Schwarting et al., 2004). The Nrp1 patterns in mice were studied in detail in an attempt to explain the phenotype of mice with a knockout of Sema3A, a ligand for Nrp1 (Schwarting et al., 2000, 2004; Taniguchi et al., 2003): Sema3A at the ventral midline would guide Nrp1+ axons to regions in the lateral and medial bulb. Dibattista and Reisert (2016) wrote, “we observed a mosaic pattern of Nrp1 expression in sagittal section of the [olfactory bulb]. Glomeruli showed high, low, or even absent levels of Nrp1.” Thus, at least four groups have reported on Nrp1 levels in glomeruli of rat and mouse since 1998, and like ourselves, none have identified a gradient or gradient-like pattern of Nrp1 levels throughout the bulb.
What, then, may Nrp1 do in axonal wiring of OSNs?
It is puzzling that there are no clear commonalities in the phenotypes of the conditional Nrp1 knockout between the two populations of OSNs that we studied, although both populations reside in the dorsal main olfactory epithelium and project their axons to glomeruli in the dorsal bulb. When a Cre driver strain for another OR gene becomes available, it is possible that yet another spectrum of phenotypes will be observed. Regardless how strong the effects of the conditional Nrp1 knockouts are on the number and positions of glomeruli, our observations for these two populations of OSNs cannot be interpreted in terms of Nrp1 levels determining glomerular positions along the anterior-posterior axis. Our findings leave unanswered the broader question as to whether and how OR-specific cAMP signals direct axonal targeting in the olfactory system.
A similar variability for the same Nrp1 floxed allele (Gu et al., 2003) was reported for axons of chemosensory neurons from the Grueneberg ganglion (a specialized olfactory subsystem) that express Gucy2g (Matsuo et al., 2012). One explanation is that in the absence of Nrp1, OSN axons are free to enter into or navigate across otherwise restricted regions of the bulb (such as regions where Sema3A is present), resulting in multiple options for sites where Nrp1– axons can coalesce into glomeruli. We speculate that for OSNs expressing rat OR I7 from a transgenic mouse MOR23 promoter, the sector and belt represent the degree of freedom in the medial and lateral aspects of the bulb that a conditional Nrp1 knockout affords to these axons. In the absence of Nrp1, the positional variability of glomeruli that is observed normally (Zapiec and Mombaerts, 2015) is greatly increased, but without a clear set of rules between medial and lateral aspects, and between the two populations of OSNs that we studied. In any case, Nrp1 contributes to reducing the complexity of the challenging task for axons of OSNs expressing a given OR to coalesce into one or a few glomeruli in highly restricted domains on the bulb.
Taken together, our results with gene-targeted mice expressing mouse OR M71 from the endogenous M71 locus and transgenic mice expressing rat OR I7 from a mouse MOR23 promoter pose a challenge to the Nrp1 model that was formulated by Imai et al. (2006, 2009) and that has made it into a textbook (Luo, 2015). A revision, reformulation, or refinement of this model becomes thus imperative. The lack of public availability of Cre driver strains for other OR genes, in particular gene-targeted strains, precludes additional experiments with conditional Nrp1 knockouts. Our analyses of Nrp1 levels in 3D reconstructed bulbs and a survey of the literature do not provide evidence for smooth, continuous, or fine-grained gradients of Nrp1 along any axis of the bulb, including along anterior-posterior axes. Moreover, it remains difficult to define and to measure “the” anterior-posterior axis of the bulb in a meaningful and comparable manner. We deem it unlikely that Nrp1 determines positioning in a simple, straightforward, and uniform fashion along any axis for the 3600 glomeruli of the mouse olfactory bulb. Nonetheless we hope that our intriguing observations of a sector and belt of scattered glomeruli in I7-Cre-YFP Tg × n5247-flox mice may provide clues about the mechanisms whereby a conditional Nrp1 knockout exerts such strong effects on the axonal coalescence of OSNs that express the same OR.
Note added in proof: while this paper was under review, Assens et al. (PMID 27578798) reported data that are overlapping and consistent with our data, interpretation, and conclusions.
Acknowledgments: This paper is dedicated to the memory of Andreas Walz. We thank Masayo Omura for providing the data of Figure 6. P.M. acknowledges the generous support from the Max Planck Society.
Authors report no conflict of interest.
M.K. and P.M. conceived the project. B.Z., O.C.B, and M.K. performed experiments and contributed to editing the paper. A.W. generated the conditional Nrp1 knockout strain in the laboratory at The Rockefeller University, New York, NY. P.M. supervised the project and wrote the paper.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.