Abstract
The axon initial segment (AIS), nodes of Ranvier, and the oligodendrocyte-derived myelin sheath have significant influence on the firing patterns of neurons and the faithful, coordinated transmission of action potentials (APs) to downstream brain regions. In the olfactory bulb (OB), olfactory discrimination tasks lead to adaptive changes in cell firing patterns, and the output signals must reliably travel large distances to other brain regions along highly myelinated tracts. Whether myelinated axons adapt to facilitate olfactory sensory processing is unknown. Here, we investigate the morphology and physiology of mitral cell (MC) axons in the olfactory system of adult male and female mice and show that unilateral sensory deprivation causes system-wide adaptations in axonal morphology and myelin thickness. MC spiking patterns and APs also adapted to sensory deprivation. Strikingly, myelination and MC physiology were altered on both the deprived and nondeprived sides, indicating system level adaptations to reduced sensory input. Our work demonstrates a previously unstudied mechanism of plasticity in the olfactory system.
SIGNIFICANCE STATEMENT Successful transmission of information from the olfactory bulb (OB) to piriform cortex through the lateral olfactory tract (LOT) relies on synchronized arrival of action potentials (APs). The coincident arrival of APs is dependent on reliable generation of APs in the axon initial segment (AIS) and fast conduction mediated by axon myelination. Here, we studied changes in mitral cell (MC) firing and AIS structure as well as changes in myelination of the LOT on unilateral olfactory deprivation in the adult mouse. Strikingly, myelination and MC physiology were altered on both the deprived and nondeprived sides, indicating system level adaptations to reduced sensory input. Our work demonstrates a previously unstudied mechanism of plasticity in the olfactory system.
Introduction
In the olfactory bulb (OB), the firing rates and synchronized firing patterns of mitral cells (MCs) and tufted cells encode vital olfactory information such as odor identity and odor salience (whether an odor deserves attention; Friedrich et al., 2004; Doucette et al., 2011; Gire et al., 2013; Lepousez and Lledo, 2013). The precise signals generated in the mouse OB must faithfully travel several millimeters along myelinated axonal tracts before reaching the piriform cortex (Schwob and Price, 1984; Walz et al., 2006; Nagayama et al., 2010; Chon et al., 2020; Imamura et al., 2020). Downstream neurons in the piriform cortex are sensitive to synchronized input, often failing to fire if the latency of incoming action potentials (APs) is >10 ms (Franks and Isaacson, 2006; Luna and Schoppa, 2008; Bolding and Franks, 2018). Despite the importance of reliable, synchronized olfactory signals and the large distances these signals must travel, little is known about the structure and function of the myelinated axons that govern signal reliability and timing in the olfactory system.
Excitable axonal domains such as the axon initial segment (AIS) and nodes of Ranvier, as well as the insulating myelin sheath, regulate cellular excitability, firing patterns, and conduction speed (Kole, 2011; Evans et al., 2015; Castelfranco and Hartline, 2016). The AIS is the portion of the axon close to the soma defined by a high density of voltage-gated sodium channels necessary for initiating APs and defining kinetics and shape (Palmer and Stuart, 2006; Kole et al., 2007, 2008). On myelinated axons, the myelin sheath and axon form specialized domains called nodes of Ranvier. Nodes of Ranvier allow for fast saltatory conduction by regenerating APs as they travel along axons (Susuki et al., 2013; Castelfranco and Hartline, 2016). The myelin sheath itself is produced by oligodendrocytes in the central nervous system and provides electrical insulation on stretches of axon between nodes of Ranvier. The myelin in these stretches of insulated axon, vital for fast saltatory conduction, also provides trophic and metabolic support to axons (Fünfschilling et al., 2012; Castelfranco and Hartline, 2016; Meyer et al., 2018).
Importantly, the organization of myelinated axons is not static. Both in vitro and in vivo, the AIS changes position, ion channel composition, and length in response to activity deprivation and stimulation paradigms, with significant consequences for cellular excitability (Grubb and Burrone, 2010; Kuba et al., 2010; Evans et al., 2015; Jamann et al., 2021). Nodes of Ranvier develop specific sizes and spacing to optimize AP timing in the auditory brainstem (Ford et al., 2015), while the myelin sheath and oligodendrocyte lineage cells respond to neuronal activity and change throughout life (Gibson et al., 2014; Hill et al., 2018; Hughes et al., 2018). Computational modeling and experimental studies highlight the importance of myelinated axons for synchronized firing, AP reliability, and optimizing AP conduction speed (Kim et al., 2013; Pajevic et al., 2014; Arancibia-Cárcamo et al., 2017). Whether myelinated axons adapt to facilitate olfactory sensory processing is unknown.
Here, we investigated the AIS, nodes of Ranvier, and myelin sheaths (together referred to as myelinated axons) in the mouse OB and lateral olfactory tract (LOT). To determine whether these structures adapt to changing sensory input in vivo, we used adult unilateral naris occlusion (UNO; Baker et al., 1993; Coppola, 2012; Kass et al., 2013) to suppress olfactory input and evaluated changes in myelinated axons and oligodendrocyte lineage cells. We found morphologic adaptations in the AIS, myelin sheath, nodes of Ranvier, and physiological changes in AP generation in MCs. In contrast to developmental UNO (Collins et al., 2018), we found no changes in oligodendrocyte lineage cell density. Whole-cell patch-clamp experiments on MCs revealed increased AP amplitude and width following UNO, as well as altered cell firing. Our findings raise the possibility that excitable axonal domains may adapt to facilitate olfactory sensory processing in the adult.
Materials and Methods
Reagents and antibodies
See Table 1. The research resource identifier (RRID), chemical abstracts number (CAS), national drug code (NDC), or catalog number (CAT) for the chemical, drug, or antibody is given. Not applicable (NA) is given if this information is not available or is custom made.
Reagent and antibody information
Source code and imaging data
Source code used for statistics, figure generation, and analysis can be found in the following repositories: https://github.com/nkicg6/excitable-axonal-domains-figures; https://github.com/nkicg6/excitable-axonal-domains-physiology; https://github.com/Macklin-Lab/imagej-microscopy-scripts.
Source imaging data can be found in the following Open Science Framework (OSF) project: https://osf.io/ez3qt/.
Statistics
All statistics were performed using the R programming language (R Core Team, 2019), MATLAB 2021a, and several packages from the tidyverse family (Wickham et al., 2019a). Plots were made using the R package ggplot2 (Wickham, 2016), using Cairo (Urbanek and Horner, 2020) for PDF export, or the Python package matplotlib (Hunter, 2007). The R packages dplyr (Wickham et al., 2019b), readr (Wickham et al., 2018), tidyr (Wickham and Henry, 2019), and cowplot (Wilke, 2019) were used for data processing and analysis. Results are given as mean ± SD, unless otherwise noted in the text. Values were compared with the Welch two sample t test using R. In the case of multiple comparisons, values were corrected using false discovery rate (FDR; Benjamini and Hochberg, 1995). Statistical significance was set at α = 0.05 (p ≤ 0.05).
The results of statistical tests are presented with the test statistic, degrees of freedom, and p value, if applicable. For example, a t test is presented as: t test, t(degrees of freedom) = t statistic, p = p value. Test statistics and p values are rounded to three significant figures in the text.
Resampling
The comparison between control and naris occluded for the mean lengths of the nodes of Ranvier per animal did not differ statistically for a t test (see Results). In order to avoid false positives because of differences in the number of samples per animal that could arise using the Kolmogorov–Smirnov (KS) test, we tested the statistical difference for the distribution of nodes using a t test of the means of repeated samples with replacement (“resampling t test”). In this case we resampled with replacement the length of 40 nodes per animal and computed the mean node length per resample. A t test was then run with the resulting mean node lengths to estimate significance. In order to ensure that this method decreased false positives we ran a simulation where the distribution of the length of nodes was equal to the mean node length for all animals with the SD of all nodes for all animals with the number of nodes per animals set to the actual sampled number.
Animal care
Adult postnatal day (P)60–P90 wild-type male and female mice (strain C57BL/6J, The Jackson Laboratory #000664) were used for all experiments. Mice were bred and housed in the University of Colorado Anschutz Medical campus vivarium.
Animals were always housed in single-sex cages of two to five individuals with a 14/10 h light/dark cycle. Mouse chow and water were available ad libitum. Experimental protocols and animal care were performed in accordance with the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus.
UNO
Adult (∼P60) wild-type mice (males and females) were anesthetized with an intraperitoneal injection of 100 mg/kg ketamine, 10 mg/kg of xylazine. When unresponsive to a toe pinch, animals were given a local application of 2% lidocaine to the external right naris, and the right naris was briefly cauterized with a Bovie high temperature cautery (Bovie Medical Corporation) and a small amount of super glue was applied to seal the naris. We applied gentamycin ophthalmic ointment to the eyes during surgery to maintain hydration. After the surgery, animals were given 0.4 ml of sterile saline subcutaneously, and subcutaneous carprofen (10 mg/kg) the day of surgery and the day after. We monitored the animals as they recovered from anesthesia on a heating pad before returning them to the vivarium in accordance with the University of Colorado Anschutz Medical Campus Institutional Animal Care and Use Committee.
Before the animals were killed for immunohistochemistry, electron microscopy (EM), or physiology, we confirmed naris occlusion by placing 0.1% Triton X-100 on the Occluded naris and ensuring that no bubbles formed (bubbles would indicate incomplete occlusion). A subset of animals was fixed and stained for tyrosine hydroxylase (TH) to confirm effective UNO (See Results).
Control animals (also called sham) were cage mates of Occluded animals and underwent the exact same protocol sans cauterization and closing the naris.
Immunohistochemistry sample preparation
Mice were anesthetized with Fatal-Plus (Vortech Pharmaceuticals) and transcardially perfused with 20 ml of 0.01 m phosphate buffered saline (PBS) (1× PBS) followed by 20 ml of 4% paraformaldehyde [diluted from a 32% aqueous paraformaldehyde solution with PBS; Electron Microscopy Sciences (EMS)] at a flow rate of 10–14 ml/min. The brains were carefully removed and postfixed in 4% paraformaldehyde for 1–2 h at 4°C. Following postfix, brains were placed in a 30% sucrose-PBS solution for 48 h for cryoprotection. Brains were then embedded in molds in optimal cutting temperature (OCT; EMS) and frozen at −80°C until sectioning.
Slices were serially sectioned in the horizontal plane at 30–40 µm thick using a Leica CM1950 (Leica Biosystems) and collected as free-floating sections in PBS in 24-well plates. We performed immunohistochemistry within 7–14 d of sectioning.
Immunohistochemistry
Section sampling
We followed the principles of unbiased stereology for cell number and AIS/node of Ranvier length quantification (Mouton, 2002; Mouton et al., 2017). We defined the anatomic quantification area to be the appropriate regions of OB and LOT [mitral cell layer (MCL) for AISs, granule cell layer (GCL), and LOT for oligodendrocyte lineage cell quantification, and GL for TH quantification] between approximately −2.04 to −5.64 mm ventral to bregma suture (Capra, 2003; Franklin and Paxinos, 2013). We used a systematic random sampling scheme to ensure unbiased cell number and AIS/node of Ranvier length quantification (Mouton, 2002). Briefly, we counted the number of sections collected for each animal in the target anatomic region, and used the Python programming language (Python 3.7.0) “random.choice” function to choose a random starting point for each set. We then chose every nth section, where n = the number of slices in the anatomic area divided by the target sampling number of slices (typically four to five sections per animal).
Section labeling
We performed immunohistochemical labeling using a free-floating slices protocol modified from previous studies (Ahrendsen et al., 2018; Gould et al., 2018). Slices were processed in batches to reduce variability and facilitate comparison.
Slices were washed three times for 5 min each in 1× PBS, then placed in 10 mm sodium citrate 0.05% Tween 20 buffer (pH 6) for 1 min to equilibrate before microwave-based antigen retrieval in a PELCO BioWave Pro microwave (550 W for 5 min; Ted Pella). Following antigen retrieval, slices were washed three times for 5 min each in 1× PBS, and permeabilized slices with 0.1–0.3% Triton X-100 PBS solution for 20 min. Slices were then blocked for 1 h in a solution of 5% normal goat or donkey serum (depending on the antibodies) in 0.3% Triton X-100 PBS at room temperature. Slices were then incubated overnight on a rocker at room temperature in the blocking solution + primary antibodies (see Table 2 for antibody concentration information and Table 1 for vendor information). Note the PLP antibody source (Yamamura et al., 1991). Incompatible antibodies [e.g., TH and contactin-associated protein (Caspr)] were not used on the same slices.
Primary antibody information
The following day, slices were washed four times for 10 min each in 1× PBS, then incubated for 2 h in blocking solution and secondary antibodies on a rocker at room temperature (see Table 3 for secondary antibody concentrations and Table 1 for vendor information). Following secondary antibody incubation, slices were washed four times for 5 min in 1× PBS, then incubated with the nuclear label Hoechst (diluted 1:5000 in 1× PBS) for 2 min on a rocker at room temperature. If slices were labeled with red fluorescent Nissl (NeuroTrace, Thermo Fisher Scientific), we incubated them in a PBS-Nissl solution in the dark for 1 h, then washed four times for 5 min in 1× PBS before performing the Hoechst label.
Secondary antibody information
Slices were then washed two times for 5 min each in 1× PBS and transferred to 0.01 m phosphate buffer for mounting. Slices were mounted on uncharged Gold Seal Rite-on glass slides (Thermo Fisher Scientific, CAT #3050) using Fluoromount-G mounting media (SouthernBiotech) and #1.5 coverslips (Thermo Fisher Scientific). Slides were then stored in the dark at 4°C until imaging.
Physiology section labeling
Following whole-cell patch-clamp experiments, a subset of sections were fixed in 4% paraformaldehyde for 2 h, washed in 1× PBS, and labeled for biocytin, AnkyrinG (AnkG), and nuclei. After fixation, slices were washed three times for 5 min in 1× PBS, then permeabilized in 0.3% Triton X-100 PBS for 20 min. Slices were then blocked for 1 h in a solution of 5% normal goat serum in 0.3% Triton X-100 PBS. Primary antibodies against AnkG were diluted in blocking solution (see Table 2), and slices were left at 4°C on a rocker for ∼72 h. Next, slices were washed four times for 10 min in 1× PBS, before secondary antibodies against AnkG (see Table 3) and Alexa Fluor 594-conjugated streptavidin (1-µl streptavidin to 500-µl block) were diluted in blocking solution and added to the slices. Slices were incubated in the dark in secondary antibody at 4°C for 48 h. Following secondary antibody incubation, slices were washed two times for 5 min in 1× PBS, and incubated in Hoechst diluted in PBS (see Table 3) for 5 min. Next, slices were washed three times for 30 min in 1× PBS, and mounted on uncharged Gold Seal Rite-on glass slides (Thermo Fisher Scientific, CAT #3050) using ProLong Gold Antifade mounting media (Thermo Fisher Scientific, CAT #P10144) and #1.5 coverslips (Thermo Fisher Scientific). Slices were cured in ProLong Gold on a flat surface for 48 h in the dark before imaging.
Electron microscopy fixation and sample preparation
Mice were anesthetized with Fatal-Plus (Vortech Pharmaceuticals) and transcardially perfused with 10 ml of 1× PBS followed by 30 ml of EM fixative (2.5% paraformaldehyde, 2.5% glutaraldehyde, 2 mm calcium chloride, and 0.1 m sodium cacodylate buffer). The PBS and EM fixative were kept on ice at 4°C. After perfusion, the brain was removed and postfixed for ∼12 h at 4°C in the EM fixative. Following postfix, we stored the brains in 0.1 m sodium cacodylate buffer at 4°C until sectioning. For sectioning, brains were embedded in 4% low-melt agarose for stability and cut in 1× PBS on a vibratome (Ted Pella) into 300-µm-thick coronal sections encompassing the region 2.45–3.05 mm anterior to the bregma suture. Further processing was performed as previously described (Ahrendsen et al., 2018). Briefly, using a PELCO Biowave Pro Tissue Processor (Ted Pella), the tissue was rinsed in 100 mm cacodylate buffer and then postfixed in a reduced osmium mixture consisting of 1% osmium tetroxide and 1.5% potassium ferrocyanide followed by 1% osmium tetroxide alone. Dehydration was performed in a graded series of acetone dilutions (50%, 70%, 90%, and 100%) containing 2% uranyl acetate for en bloc staining. Finally, tissue was infiltrated and embedded in Embed 812 (EMS) and cured for 48 h at 60°C. Tissue was oriented so that sections could be cut in the coronal plane to visualize the LOT. Ultrathin sections (65 nm) were mounted on copper slot grids and viewed at 80 kV on a Tecnai G2 transmission electron microscope (FEI). EMs were obtained in consistent regions in the lateral portion of LOT.
Confocal and electron microscopy imaging and quantification
All image analysis (EM and confocal) was performed using the freely available Fiji distribution of ImageJ (Schindelin et al., 2012). EM and confocal images were blinded using a custom Fiji script called blind-files, provided as part of the Lab-utility-plugins update site (see above, Source code and imaging data).
AIS quantification
For AIS quantification, images were taken on a Nikon A1R resonance scanning confocal microscope (Nikon) with a Nikon Plan Fluor 40× oil immersion objective (numerical aperture = 1.3). We acquired 3D confocal stacks 318 × 318 × 15 µm3 (X Y Z) with a voxel size of 0.31 × 0.31 × 0.225 µm3 (X Y Z). We acquired Hoechst (405-nm excitation), AnkG (488-nm excitation), Nissl (561-nm excitation), and Plp (640-nm excitation) images with a line average of 4. We took images from lateral and medial MCL of each bulb for analysis. Images were blinded and analyzed in 3D using the semi-automated tracing tool Simple Neurite Tracer (SNT; Arshadi et al., 2021); a Fiji plugin. We traced from the origin of the AnkG signal at the base of the Nissl+ soma to the termination of the AnkG signal (typically ending abruptly in a Plp+ myelin sheath; see Fig. 1). We then used the Fit Paths option in SNT to automatically optimize path fits using 3D intensity around each traced node (Arshadi et al., 2021) before exporting length measurements as comma separated value (CSV) spreadsheets for analysis using R. AIS length analysis was performed for treatment group (Control vs Naris Occlusion), within group (Left vs Right for Control, Open vs Occluded for Naris Occlusion), and between group (Control vs Open vs Occluded). When AISs were grouped by animal, we calculated the mean AIS length per section and animal.
TH quantification
Horizontal sections (encompassing both OBs) were labeled for TH and nuclei (Hoechst) as described above. We took tiled images of the whole sections (both bulbs) using a Zeiss Axio Imager.M2 widefield microscope with a Zeiss 20× Plan-Apochromat (numerical aperture = 0.8) objective and an HXP 120 metal halide lamp (Carl Zeiss Microscopy). The XY pixel size was 0.65 µm/pixel. We then cropped the tiled images in Fiji so only one bulb was present in each image and labeled them appropriately. Images were then blinded for analysis. We used the polygon tool in Fiji to trace the glomerular layer (GL) (using the Hoechst-labeled nuclei of glomeruli as a guide), then used a custom Fiji script to extract fluorescence intensity measurements into CSV format for further analysis using R. We calculated a mean fluorescence intensity (sum pixel intensity/area traced) per side and animal for the analysis. Intensity is presented in arbitrary units (a.u.) and either compared directly within animal, or as a relative intensity between Control and Naris Occlusion (Right/Left or Occluded/Open; see Fig. 3).
Node of Ranvier quantification
Node of Ranvier images were taken on a Nikon A1R resonance scanning confocal microscope (Nikon) with a Nikon Plan Fluor 40× oil immersion objective (numerical aperture = 1.3) using a 3× optical zoom. We acquired 3D confocal stacks 106 × 106 × 15 µm3 (X Y Z) with a voxel size of 0.106 × 0.106 × 0.25 µm3 (X Y Z). We acquired Caspr (488-nm excitation), and Nav1.6 (561-nm excitation) and Plp (640-nm excitation) with a line average of 4. We took images from the LOT of both bulbs on each slice for analysis. Images were blinded before we manually traced nodes. We traced a random subset of 10–25 nodes per image in 3D using the Fiji segmented line tool. We only traced nodes where the Nav1.6 signal was approximately contained in a single optical section to control for out of plane errors. After tracing, the ROI files of all traced nodes were saved, and we used a custom script to extract the Nav1.6 fluorescence signal and fit a Gaussian to the signal using Fiji's curve fitting tools. We extracted the fit parameters and calculated the full width at half maximum of the Gaussian fit to determine node length. To control for poor fitting, we excluded all nodes whose Gaussian fit R2 value were <0.9. When node lengths were grouped by animal, we calculated the mean node length per section and animal.
To quantify the density of the nodes in the LOT we used a custom script in Fiji to measure the volume of the LOT and the number of nodes of Ranvier in the 3D confocal stacks. The background fluorescence of the Nav1.6 signal was used to measure the tissue area per z-plane, and the volume was calculated as the sum of areas in all z-planes × number of z-planes × voxel depth. The 3D Objects Counter plugin of Fiji was used to determine the number of objects positive for Nav1.6 and the number of objects positive for Caspr. The 3D MultiColoc plugin was then used to remove the objects positive for Nav1.6 that do not colocalize with objects positive for Caspr. The nodes of Ranvier density is expressed as number of objects positive for Nav1.6 per mm3.
Oligodendrocyte lineage cell quantification
Oligodendrocyte lineage cell images were taken on a Nikon A1R resonance scanning confocal microscope (Nikon) with a Nikon Plan Fluor 40× oil immersion objective (numerical aperture = 1.3). We acquired 3D confocal stacks 318 × 318 × 10 µm3 (X Y Z) with a voxel size of 0.31 × 0.31 × 0.25µm3 (X Y Z). We acquired Hoechst (405-nm excitation), Olig2 (488-nm excitation), PDGFRα (561-nm excitation), and CC1 (640-nm excitation) with a line average of 4. We took images from the GCL and LOT of each bulb for analysis. Images were blinded and analyzed in 3D using the Fiji Cell Counter plugin. Only cells completely contained within the imaging field were counted. Cell Counter plugin results were saved as CSV for analysis in R.
Whole-cell patch-clamp cell morphology quantification
We took images of a subset of cells filled with biocytin and stained for AnkG on a Nikon A1R resonance scanning confocal microscope (Nikon) with a Nikon Plan Fluor 40× oil immersion objective (numerical aperture = 1.3). Some images were tiled to create a representative image of the axon and primary dendrite. Voxel size was 0.62 × 0.62 × 1.1 µm3 (X Y Z), which was sufficient for a faithful representation of the cells. Images were maximum intensity projected in Z for quantification. We manually checked whether a filled cell had a primary dendrite extending to the glomerulus, and whether it had a visible axon with an AnkG+ AIS.
Electron microscopy quantification
EM images were blinded then automatically segmented and measured using AxonDeepSeg, a deep learning-based EM segmentation program (Zaimi et al., 2018). We performed segmentation using the built-in transmission EM (TEM) model, and then manually inspected the blinded images to identify only well-segmented axons for inclusion in the final analysis. We selected measurements from 10–15 well-segmented axons per image. The vast majority of axons in the LOT are myelinated by P30 (Collins et al., 2018), so we only selected myelinated axons. We also excluded any axon with a calculated diameter of <0.4 µm, since axons smaller than this are not often myelinated (Lee et al., 2012).
Electrophysiology
Acute slice preparation
For optimal patch-clamp recordings on older animals (∼P90), we used different modified artificial cerebrospinal fluid (ACSF) solutions for slice preparation and slice incubation (Ting et al., 2014, 2018). Animals were anesthetized with ketamine/xylazine (100 mg/kg ketamine, 10 mg/kg xylazine) and when unresponsive they were transcardially perfused with 25 ml of ice-cold N-methyl-D-glucamine (NMDG)-based ACSF (NMDG-ACSF: 92 mm NMDG, 2.5 mm KCl, 1.2 mm NaH2PO4, 30 mm NaHCO3, 25 mm glucose, 20 mm HEPES, 5 mm Na-ascorbate, 2 mm thiourea, 3 mm Na-pyruvate, 10 mm MgSO4, and 0.5 mm CaCl2, adjusted to pH 7.4 with 5 m HCl, osmolarity 300–310 mmol/kg) bubbled continuously with carbogen (95% oxygen, 5% carbon dioxide). The brain was removed, embedded in 2.5% low melt agarose (diluted in NMDG-ACSF), and cut in 300- to 400-µm horizontal sections using a Compresstome VF-310 (Precisionary Instruments) in carbogen bubbled NMDG-ACSF. Once cut, slices were transferred to incubate at 32°C in carbogen bubbled NMDG-ACSF for 30 min (resting period). During the initial resting incubation period, we performed the sodium spike protocol for three- to six-month-old mice to optimize gigaohm seal formation (Ting et al., 2018). This involved adding set volumes of 2 m sodium chloride at regular intervals to slowly re-equilibrate the slices to sodium ions (described in Ting et al., 2018; three- to six-month-old mice: 250 µl at 5-min resting, 500 µl at 10-min resting, 1000 µl at 15-min resting, 2000 µl at 25-min resting, and transfer at 30 min).
Following the sodium spike in, slices were transferred to a room temperature HEPES-based ACSF solution for 1 h before recording (HEPES-ACSF: 92 mm NaCl, 2.5 mm KCl, 1.2 mm NaH2PO4, 30 mm NaHCO3, 25 mm glucose, 5 mm Na-ascorbate, 2 mm thiourea, 3 mm Na-pyruvate, 2 mm MgSO4, and 2 mm CaCl2, adjusted to pH 7.4 with 5 m NaOH, osmolarity 300 mmol/kg). For all solutions, osmolarity was measured with a VAPRO vapor pressure osmometer (Wescor).
Whole-cell patch-clamp recording
Whole-cell patch clamp was performed with pipettes filled with a potassium gluconate-based internal solution (130 mm K-gluconate, 10 mm HEPES, 10 mm KCl, 0.1 mm EGTA, 10 mm Na2-phosphocreatine, 4 mm Mg-ATP, and 0.3 mm Na2-GTP, adjusted to pH 7.3 with KOH, osmolarity 280 mmol/kg). Some recordings were done with 2 mg/ml of Biocytin, added the day of the experiment, for post hoc cell visualization. Pipettes were pulled from borosilicate glass with filaments, inner diameter 0.86 mm, outer diameter 1.5 mm (item BF-150-86-10, Sutter Instruments) to a tip resistance of 3–4 MΩ with a P-97 Flaming/Brown type micropipette puller (Sutter Instruments).
We performed all recordings in the presence of the glutamatergic inhibitors 6,7-dinitroquinoxaline-2,3-dione (DNQX; 10 μm), 2-amino-5-phosphonopentanoic acid (APV; 50 μm), and the GABAergic inhibitor gabazine (5 μm) in ACSF (5 mm HEPES, 125 mm NaCl, 2.5 mm KCl, 1.25 mm NaH2PO4, 24 mm NaHCO3, 12.5 mm glucose, 2 mm MgSO4, and 2 mm CaCl2, adjusted to pH 7.4, osmolarity 300–310 mmol/kg, bubbled continuously with carbogen, called recording ACSF).
During recording, slices were placed in a custom perfusion chamber continuously perfused with carbogen bubbled recording ACSF heated to 33–36°C with a SH-27B in line heater and a TC-324C temperature Controller (Warner Instruments). We performed the experiments using a Zeiss Axioskop 2 FS Plus microscope (Carl Zeiss Microscopy) equipped with differential interference contrast optics and a 40× (numerical aperture = 0.8) Zeiss Achroplan water immersion objective (Carl Zeiss Microscopy). We visualized slices using a CoolSNAP HQ2 camera (Teledyne Photometrics) with Micro-Manager software version 1.4.22 (Edelstein et al., 2014). Patch pipettes were manipulated using a MP-285 manipulator arm driven by a MPC-200 Controller and ROE-200 micromanipulator (Sutter Instruments).
Data were acquired using Clampex software version 10.5.0.9 with an Axopatch 200A amplifier, CV-201A headstage, low pass filtered with a Bessel filter at 2 kHz and digitized with an Axon Digidata 1550A at 20 kHz (Molecular Devices). We did not correct for a junction potential. We performed offline filtering of current clamp traces using a third order Savistky–Golay filter with a 0.5-ms window. Displayed traces were filtered with a 1-ms window for appearance (see Results).
MCs were identified based on their large cell bodies and position in the MCL. A subset of cells was filled with biocytin, fixed with 4% paraformaldehyde, and visualized to confirm the presence of an apical dendrite extending to the glomerular layer (GL; see Results). Access resistance and resting potential were checked shortly after achieving whole-cell configuration, and if it exceeded 40 MΩ, cells were discarded (Fadool et al., 2011). For current clamp experiments, cells were held at –60 mV. We sampled from the first 500 ms of the current clamp experiments, before the current step began, and calculated the mean membrane potential to confirm cells were close to the target holding potential. We noted no differences between cells from Naris Occlusion and Control animals (Naris Occlusion −61.3 mV ± 0.95, n = 20 cells and Control −58.1 ± 2.52, n = 23 cells, t test, t(28) = 1.11, p = 0.28; presented as mean ± SEM). For current step experiments, a series of 1000-ms current steps were applied to evoke APs (0–500 pA, 25-pA steps). If multiple recordings were made from the same cell, we averaged spike counts or AP feature measurements for that cell.
Electrophysiology analysis
Physiology data were analyzed using custom scripts (see above, Source code and imaging data) written in the Python programming language (version 3.7–3.9). We used the pyABF Python module (version 2.2.8) to open axon binary format files (Harden, 2020). Additionally, we used the Python libraries Matplotlib 3.3.2 (Hunter, 2007), numpy 1.19.2 (Harris et al., 2020), and SciPy 1.5.2 (Virtanen et al., 2020).
Results
Quantification of MC AISs in the OB and nodes of Ranvier in the LOT
The coordinated, precise firing of groups of MCs encodes olfactory information in the OB, and firing patterns of MCs change as an animal learns to identify odors in olfactory discrimination tasks (Doucette et al., 2011; Gire et al., 2013; Lepousez and Lledo, 2013; Gschwend et al., 2015; Chu et al., 2016). The AIS is the site of AP initiation, and its morphologic structure and ion channel composition determines AP threshold, AP width and amplitude, and other important firing properties of the cell (Palmer and Stuart, 2006; Kole et al., 2007, 2008; Kole and Stuart, 2012; Jamann et al., 2021).
MCs are known to have identifiable AISs similar to other brain regions (Price and Powell, 1970; Hinds and Ruffett, 1973; Lorincz and Nusser, 2008), but their structure and size distributions in OB are not well understood. We measured the length of MC AISs in 3D volumes using immunohistochemistry and confocal microscopy (Fig. 1). The OB is a highly stratified structure whose layers are clearly delimited using the nuclear label Hoechst (Fig. 1B). The histochemical dye Nissl (NeuroTrace, Thermo Fisher Scientific) broadly labels cells in the OB, and MCs could be clearly identified based on their large Nissl+ somas and their location in the MCL. AnkG is a cytoskeletal scaffolding protein essential for the structure and function of the AIS (Hedstrom et al., 2008). The AnkG+ AIS of MCs extended directly off the MC soma, often immediately into the first myelinated internode (Fig. 1B', white arrowheads mark AIS start, yellow mark the end). AISs were measured in 3D volumes using the ImageJ/Fiji plugin SNT (Arshadi et al., 2021), starting at the AnkG label originating at the base of the Nissl+ cell body and terminating at the distal end of the AnkG label. The mean length of MC AISs in the OB was 25.7 ± 3.81 µm (lengths are given as mean ± SD, n = 687 AISs from 4 animals; Fig. 1C).
Characterization of excitable axonal domains in the olfactory system. A, Diagram of a transverse section of the OB and rostral LOT. Boxes indicate the areas where measurements were performed for the AIS and nodes of Ranvier. Magenta box (A) corresponds to an example region for AIS length measurements (B, B'), and the green boxed region (A) corresponds to an example region for node of Ranvier measurements (D, D'). B, Immunohistochemistry for nuclei (Hoechst, white), the AIS (AnkG, green), cell bodies (Nissl, magenta), and myelin [proteolipid protein (PLP), blue]. The functional layers of the OB are labeled on the edges of the image (EPL = external plexiform layer). Scale bar in B: 50 µm. B', Expanded region highlighting Nissl labeling of MC bodies and AnkG+ AISs in the MCL. White arrowheads label the start of MC AISs and yellow arrowhead show the termination. Scale bar in B': 15 µm. C, Distribution of AIS lengths in the adult OB. Dashed line indicates mean length. D, Nodes of Ranvier in the LOT corresponding to the region highlighted in the green box in A. Nav1.6 (magenta) marks the excitable nodal region, Caspr (green) marks flanking paranodal junctions. Scale bar in D: 10 µm. D', Expanded node of Ranvier from white box in D. Scale bar in D': 2 µm. E, Fluorescence intensity profile of the node of Ranvier in D' (light white line) with a Gaussian fit of Nav1.6 (gray dashed line). The Gaussian fit of Nav1.6 was used to measure node length (see Materials and Methods). F, Distribution of node lengths in the LOT. Dashed line indicates mean length.
While precise firing patterns are generated in the OB, MC axons must carry these signals out of the OB to the piriform cortex and other brain regions via the highly myelinated LOT (Schwob and Price, 1984; Walz et al., 2006; Nagayama et al., 2010; Chon et al., 2020; Imamura et al., 2020). Nodes of Ranvier are the axonal gaps between the myelin sheaths along the axon. Nodes of Ranvier regenerate the AP as it travels along an axon in a process called saltatory conduction (Susuki et al., 2013). Tuning of node of Ranvier length or channel composition may be a sensitive way to regulate AP conduction velocity and synchrony (Arancibia-Cárcamo et al., 2017). To assess the structure of nodes of Ranvier in the LOT, we used antibodies directed against Caspr and the voltage-gated sodium channel Nav1.6. Caspr marks the site of myelin sheath to axon adhesion and serves as an important ion diffusion barrier (Bhat et al., 2001; Rios et al., 2003), while Nav1.6 is the primary voltage-gated sodium channel in mature nodes, responsible for regenerating the AP (Caldwell et al., 2000; Boiko et al., 2001; Rasband et al., 2003). The LOT contains a high density of nodes clearly identified by Nav1.6 flanked by Caspr (Fig. 1D). We measured node of Ranvier length manually by fitting a Gaussian to the fluorescence intensity profile of antibody labeled Nav1.6 and measuring the full width at half maximum of that fit (Fig. 1D',E; see Materials and Methods). The mean length of nodes of Ranvier in the LOT was 1.17 ± 0.267 µm (mean ± SD n = 811 nodes from 4 animals; Fig. 1F).
Quantification of oligodendrocyte lineage cells and the myelin sheath
In the central nervous system, the myelin sheath is made by oligodendrocytes. Oligodendrocytes develop from oligodendrocyte progenitor cells (OPCs), a population long known to proliferate, differentiate into oligodendrocytes, and produce myelin in response to neuronal activity (Barres and Raff, 1993; Young et al., 2013; Gibson et al., 2014; McKenzie et al., 2014). In the mouse olfactory system, a significant period of OPC differentiation and myelination begins around P10 and is largely complete by P30 (Philpot et al., 1995; Collins et al., 2018). Little is known about oligodendrocyte lineage cells in the mature olfactory system. The transcription factor Olig2 marks the entire oligodendrocyte lineage, while PDGFRα marks immature OPCs and CC1 marks mature oligodendrocytes. We counted OPCs using antibodies directed against Olig2 and PDGFRα, and oligodendrocytes using antibodies against Olig2 and CC1. We counted oligodendrocytes and OPCs in the GCL and the LOT of adult animals in 3D confocal volumes (Fig. 2A–D). The GCL contained an average of 6426 ± 3173 OPCs/mm3 and 33,616 ± 5534 oligodendrocytes/mm3 (cell counts are given as mean density ± SD). The LOT contained 7486 ± 2098 OPCs/mm3 and 99,810 ± 7979 oligodendrocytes/µm3 (all cell counts were performed with a systematic random sampling scheme using three to five sections per animal, where n animals ≥4; see Materials and Methods).
Characterization of oligodendrocyte lineage cells and myelin sheaths in the OB and LOT. A, Diagram of a transverse section of the OB and LOT with gray boxes indicating the GCL and LOT. B, Oligodendrocyte lineage cells in the GCL. C, Oligodendrocyte lineage cells in LOT (area between dashed white lines). Yellow arrows indicate OPCs, and white arrows indicate oligodendrocytes. Scale bars in B, C: 25 µm. D, Oligodendrocyte lineage cell quantification in GCL and LOT. Points indicate individual animals (n = 4 animals) and bars indicate the group means. E, EM image of myelinated axons in a coronal section of LOT. Scale bar in E: 0.5 µm. F, Diagram demonstrating g-ratio calculations and histogram of g-ratio distribution in LOT. To calculate g-ratio, the perimeter of the axon (PA) was divided by the perimeter of the surrounding myelin sheath (PM) to yield g. Histogram shows distribution of g-ratios in LOT. Dashed line histogram indicates mean g-ratio. G, Distribution of axon diameters in LOT. Dashed line indicates the mean axon diameter.
The myelin sheath provides electrical insulation and vital trophic/metabolic support to axons (Fünfschilling et al., 2012; Castelfranco and Hartline, 2016; Meyer et al., 2018). The g-ratio (presented as axon perimeter/total fiber perimeter), is a measure of myelin thickness relative to axon diameter. The g-ratio has long been used in computational modeling studies as a parameter to assess the conduction velocity along myelinated axons (Rushton, 1951; Chomiak and Hu, 2009). Since axon size and myelin thickness (g-ratio) have significant effects on axonal conduction velocity and AP fidelity (Chomiak and Hu, 2009; Kim et al., 2013; Etxeberria et al., 2016), we measured the g-ratio of myelinated axons in the LOT between 2.45 and 3.05 mm anterior of the bregma suture (Fig. 2D–H). The mean g-ratio was 0.72 ± 0.07, while the mean axon diameter was 0.87 ± 0.31 µm (Fig. 2F,H; n = 265 axons from 5 animals).
UNO causes adaptations in MC AISs and nodes of Ranvier
The OB is a remarkably plastic structure. New neurons are incorporated into functional circuits in the OB throughout life (Whitman and Greer, 2007; Lazarini et al., 2009; Yamaguchi et al., 2013), and the firing patterns of MCs are known to change significantly as an animal learns an olfactory discrimination task (Friedrich et al., 2004; Doucette et al., 2011; Gire et al., 2013; Lepousez and Lledo, 2013; Gschwend et al., 2015; Chu et al., 2016; Losacco et al., 2020). Little is known about whether axonal domains adapt to changing olfactory input in adult animals. To test whether myelinated axons respond to changing olfactory input in adult mice, we performed UNO on P60 mice for 30 d and measured the length of MC AISs and nodes of Ranvier in the LOT.
UNO is a model for sensory deprivation in which one naris is surgically closed to block sensory input. One of the hallmarks of UNO is an activity-dependent decrease in TH mRNA and protein expression in a subset of dopaminergic periglomerular cells in the GL (Baker et al., 1993; Sawada et al., 2011). To assess the efficiency of UNO, we measured the fluorescence intensity of TH in the GL of P60 mice that underwent UNO for 30 d (called Naris Occlusion) and control animals (cage mate sham surgery, called Control; Fig. 3). Within Naris Occlusion animals, we noted a significant decrease (∼45%) of TH fluorescence intensity in the Occluded bulb GL compared with the Open bulb GL (paired t test, t(6) = 9.58, p = 7.4e-5, n = 7 animals). Control animals showed no significant difference in TH intensity between Left and Right bulbs (Fig. 3D, paired t test, t(3) = −0.383, p = 0.727, n = 4 animals). We compared the relative intensity (Occluded side/Open side for Naris Occlusion animals, and Left side/Right side for Control animals) and found that the relative intensity was roughly equal (centers on 1) in Control animals, but was significantly reduced in Naris Occlusion animals, indicating effective UNO (Fig. 3E, t test, t(7.2) = 6.25, p = 0.00038).
UNO experimental design and UNO validation. A, Experimental timeline. P60 animals underwent UNO for 30 d and were analyzed at P90. B, Example bilateral horizontal section of OB showing TH immunolabeling in a P90 Control animal. C, Example bilateral horizontal section of OB showing TH immunolabeling in a P90 animal following 30 d UNO (Naris Occlusion). Scale bars in B, C: 150 µm. D, TH fluorescence intensity measured in the GL is not significantly different between the left and right sides of Control animals (paired t test, t(3) = −0.383, n = 4 animals). There is a significant decrease in TH glomerular fluorescence intensity between the Open and Occluded bulbs of Naris Occlusion animals (paired t test, t(6) = 9.58, n = 7 animals). Bulk TH glomerular fluorescence intensity is given in a.u. E, Plot showing relative intensity of OBs within Control animals or within Naris Occlusion animals. Naris Occlusion animals have significantly lower relative TH intensity compared with Control animals. Dashed line at 1 indicates equal intensity on either side (Right/Left for Control or Occluded/Open for Naris Occlusion). Relative TH intensity was significantly reduced in Naris Occlusion animals compared with Control (t test, t(7.2) = 6.25).
Do MC AISs adapt to changing sensory input following UNO? We measured MC AISs in 3D volumes (as in Fig. 1; see Materials and Methods) in both 30 d. Naris Occlusion animals and Controls. The distribution of AIS lengths was not significantly different between the Left and Right sides of Control animals (KS test, D = 0.0599, p = 0.592, n = 678 AISs from 4 animals), but in Naris Occlusion animals, the Open and Occluded sides were significantly different (KS test, D = 0.128, p = 3.97e-7, n = 1894 AISs from 7 animals; Fig. 4A–C).
MC axonal adaptations following 30 d of UNO. A, Example MC AISs from the left and right OB of Control animals. A', Example MC AISs from the Occluded and Open OB from UNO animals. Scale bar in A', A: 15 µm. B, Empirical distributions comparing the AIS lengths from the left and right OBs of Control animals. P value is from a KS test comparing the distributions (KS test, D = 0.0599, n = 678 AISs from 4 animals). Points along the empirical distribution show the mean length of each OB's AISs mapped to the probability within the respective distribution. Inset shows within-animal comparison of mean AIS lengths. Points in the bar plot insets represent the mean AIS length from an animal's left or right OB (abbreviated L and R, respectively), and the bars represent the mean value for that group (paired t test, t(3) = −0.0548). N = 678 AISs from 4 animals. C, Similar to B but comparing the empirical distribution of AISs from Open and Occluded OBs from UNO animals (KS test, D = 0.128, n = 1894 AISs from 7 animals). Inset shows within-animal comparison of mean AIS lengths in Open versus Occluded bulbs (abbreviated Op and Oc, respectively) from Naris Occlusion animals (as in B, paired t test, t(6) = 3.18). D, Mean AIS lengths calculated on a per-animal OB basis (Control combines the left and right OBs from Control animals, while Naris Occlusion Occluded and Open OBs were kept separate). AIS lengths from Control OBs were not significantly different from AIS lengths from Occluded OBs (t test t(9) = 0.32), but AISs from Open OBs were significantly longer than both Control (t test, t(8.66) = −2.84) and Occluded OBs (t test, t(11.6) = −2.69). FDR corrected p values are shown in plot. E, Example nodes of Ranvier in the LOT from the left and right sides of Control animals. E', Similar to E, but in UNO animals. Scale bar in E', E: 2 µm. F, Empirical distributions comparing node of Ranvier lengths from the left and right LOTs of Control animals. P value is from a KS test comparing node of Ranvier length distributions (KS test, D = 0.0773, n = 811 nodes from 4 animals), displayed similar to B, C. Inset shows within-animal between side comparisons of mean node of Ranvier lengths. Points represent the mean node of Ranvier length from an animal's left or right LOT, and the bars in the inset represent the mean value for that side (paired t test, t(3) = 0.227). G, Similar to F, but comparing node of Ranvier lengths from the Open and Occluded LOTs of Naris Occlusion animals (KS test, D = 0.0508, n = 1409 nodes from 4 animals). Inset similar to F, showing within-animal between side comparison of mean node of Ranvier lengths (paired t test, t(3) = 0.131). H, Comparison of node of Ranvier length distributions for Control and Naris Occlusion. P value is from a KS test comparing the node of Ranvier lengths for LOTs from Control animals or Naris Occlusion animals (KS test, D = 0.137). Points along the distribution show individual animal means mapped to the distribution. Inset shows a per-animal comparison of mean node length [Control (C) vs Naris Occlusion (N)]. The points represent per-animal means and bars represent the group means (t test, t(3.38) = 0.56). I, Cartoon illustrating the ALAC. Regions of interest were taken from the locations specified by the magenta boxes. J, Example tract of the ALAC leading into one of the OBs. The tracts of the ALAC are tightly grouped and organized. Scale bar: 25 µm. K, Similar to H, empirical distributions comparing nodes of Ranvier in the ALAC between UNO animals and Control animals. P value is from a KS test comparing node of Ranvier length distributions (KS test, D = 0.15, n = 4 control animals and 4 UNO animals). Points along the empirical distribution represent the mean node of Ranvier length for a specific animal mapped to the distribution. Inset shows a per-animal comparison of mean node length (Control vs Naris Occlusion). Points represent per-animal means and bars represent the group means (t test, t(4.44) = −2.58).
For each group (Control and Naris Occlusion), we summarized the data within animals and performed paired t tests (Fig. 4B,C, insets). The left and right sides of Control animals were not significantly different (paired t test, t(3) = −0.0548, p = 0.96, n = 4 animals), but the Open and Occluded sides of Naris Occlusion animals were significantly different (paired t test, t(6) = 3.18, p = 0.019, n = 7 animals).
Strikingly, Naris Occlusion animals (Open and Occluded sides combined) had a larger range of AIS lengths than Control animals (7.06–55.1 µm, SD = 5.83 µm for Naris Occlusion, 15.4–39.5 µm, SD = 3.81 µm for Control). Indeed, when we compared the variances of the two main distributions (Control vs Naris Occlusion), they were significantly different (Fligner–Killeen test χ2(1) = 59.5, p = 1.26e-14). Within the Control group (Left vs Right), we found no statistically significant difference in variance (Fligner–Killeen test χ2(1) = 0.0106, p = 0.918), indicating homogeneous variance between left and right sides of Control animals. However, within Naris Occlusion animals (Open vs Occluded), variance was significantly different (Fligner–Killeen test χ2(1) = 9.97, p = 0.00159). A larger variance of AIS lengths in UNO animals could have implications for information encoding within the OB by increasing the diversity of AP shapes and firing frequencies from MCs (see Discussion).
To further investigate the difference in length distributions between Control and Naris Occlusion animals, we calculated the mean AIS length per animal and bulb. Given the consistency of AIS lengths in Control animals, we combined the left and right bulbs into one group. The omnibus ANOVA comparing AIS lengths from Control bulbs, Open bulbs, and Occluded bulbs was significant (Fig. 4D, ANOVA, F(2,15) = 4.72, p = 0.026). Surprisingly, AISs from Open bulbs were significantly longer than Control bulbs (t test, t(8.66) = −2.84, FDR corrected p = 0.03, mean length Open = 26.9 ± 1.02 µm, mean length Control = 25.4 ± 1.04 µm), and Occluded bulbs (t test, t(11.6) = −2.69, FDR p = 0.03, mean length Open = 26.9 ± 1.02 µm, mean length Occluded = 25.2 ± 1.24 µm). Interestingly, Occluded bulbs were not significantly different from Control bulbs (t test, t(9) = 0.322, FDR p = 0.75, mean length Occluded = 25.2 ± 1.24 µm, mean length Control = 25.4 ± 1.04 µm).
Together, our data indicates that 30 d of UNO causes both a significant increase in MC AIS length in the Open bulb compared with Control and Occluded bulbs, and increases the variance of the entire distribution of AIS sizes in UNO animals relative to Controls.
Changes in (or loss of) nodes of Ranvier are predicted to have significant effects on conduction velocity and AP reliability (Kim et al., 2013; Arancibia-Cárcamo et al., 2017; Hamada et al., 2017), both of which are vital for olfactory sensory processing in downstream brain regions such as the piriform cortex (Franks and Isaacson, 2006; Luna and Schoppa, 2008; Bolding and Franks, 2018). To investigate whether nodes of Ranvier adapt to sensory deprivation in the olfactory system, we measured the length of nodes (as described in Fig. 1D',E; see Materials and Methods) in Naris Occlusion animals and Controls. The distribution of node of Ranvier lengths between the left and right sides of Control animals was not significantly different (KS test, D = 0.0773, p = 0.179, n = 811 nodes from 4 animals). The distribution of node lengths between the Open and Occluded sides of Naris Occlusion animals was also not significantly different (KS test, D = 0.0508, p = 0.338, n = 1409 nodes from 4 animals). Similarly, when the data were summarized by animal, we found no significant differences (Fig. 4, insets F,G, Control paired t test, t(3) = 0.227, p = 0.835, Naris Occlusion paired t test, t(3) = 0.131, p = 0.904).
Since between-bulb distributions (Left vs Right and Open vs Occluded) were not different in either group, we next compared the distribution of node of Ranvier lengths from Control animals to Naris Occlusion animals (Fig. 4H). When we directly compared the mean values per animal, values were not significant (t test, t(3.38) = 0.56, p = 0.611; Fig. 4H, inset). However, we found that the distribution of node lengths was significantly different between the Control and Naris Occlusion animals (t test, t(9) = 3.18, p =0.01; we performed bootstrap resampling to accommodate different sample sizes; see Materials and Methods) with node lengths in the Naris Occlusion animals shorter than for the Control animals. In the case of nodes of Ranvier, we found no evidence of differing variances between the left and right sides of Control animals (Fligner–Killeen tests, χ2(1) = 1.15, p = 0.283), the Open and Occluded sides of Naris Occlusion animals (Fligner–Killeen tests, χ2(1) = 0.564, p = 0.453), or Naris Occlusion versus Control (Fligner–Killeen tests, χ2(1) = 2.95, p = 0.086). Finally, we measured the density of nodes and we found no difference between Open and Occluded and no difference between Naris Occluded and Control animals (t test, t(3) = 3.18, p = 0.66; t test, t(11) = 2.2, p = 0.24).
We also surveyed nodes of Ranvier in the anterior limb of the anterior commissure (ALAC). The ALAC carries bilaterally projecting olfactory axons between OBs, from both olfactory cortices, from the anterior olfactory nucleus, and also contains centrifugal projections from other brain regions (Brunjes, 2012; Rabell et al., 2017; Collins et al., 2018). Changes in the ALAC could mediate bilateral adaptations. A previous study found no significant difference in g-ratios in the ALAC following developmental naris occlusion, although they did find a decrease in myelin sheath thickness (Collins et al., 2018). We measured nodes of Ranvier from different portions of the ALAC in Control and Naris Occlusion animals and compared node length distributions (Fig. 4). When we calculated the mean node length for the 4 animals in each group and compared them, we did not find a significant difference (t test, t(4.44) = −2.85, p = 0.055). However, when we considered the node length distributions we found that, opposite to the LOT, node lengths in the ALAC were longer in Naris Occlusion animals compared with Controls [Fig. 4K, resampling t test (see Materials and Methods), t(9) = −9.32, p = 0.000006; compare with Fig. 4H].
Together, these data point to a system-level adaptation in node of Ranvier length following Naris Occlusion. Nodes of Ranvier are formed from axon-myelin interactions (Susuki et al., 2013), so we next investigated whether oligodendrocyte lineage cells or the myelin sheath adapt to adult UNO.
UNO leads to adaptations in myelin
Do oligodendrocyte lineage cells and the myelin sheath respond to UNO in adults? We quantified OPC and oligodendrocyte density in Control and Naris Occlusion animals in the GCL and LOT after 30 d of UNO (Fig. 5). We found no significant difference between the left and right sides of Control animals in the density of GCL oligodendrocytes (paired t test, t(3) = −0.364, p = 0.74, n = 4 animals) or GCL OPCs (paired t test, t(3) = 1.53, p = 0.223, n = 4 animals). We also found no significant difference between the Open and Occluded sides of Naris Occlusion animals in GCL oligodendrocytes (paired t test, t(3) = 2.43, p = 0.0938, n = 4 animals) or GCL OPCs (paired t test, t(3) = −0.516, p = 0.642, n = 4 animals), although three of the four animals had a reduction in oligodendrocytes in Occluded bulbs (Fig. 5A'). There was no significant difference between the main groups (Control and Naris Occlusion) in GCL oligodendrocytes (t test, t(5.41) = 0.192, p = 0.855, n = 8 animals) or GCL OPCs (Fig. 5, row A, t test, t(5.22) = −0.668, p = 0.533, n = 4 animals/group).
Oligodendrocyte lineage cells and myelinated axons after 30 d of UNO. A, Comparison of GCL oligodendrocyte lineage cell density in the left and right bulbs of Control animals. We noted no significant differences between the left and right bulbs of Control animals in oligodendrocytes (paired t test, t(3) = −0.364) or OPCs (paired t test, t(3) = 1.53). Points represent within bulb means for each animal, p values represent paired t tests, n = 4 animals. A', Comparison of GCL oligodendrocyte lineage cell density in the Open and Occluded bulbs of Naris Occlusion animals. We found no significant difference in cell density between the Open or Occluded bulbs in oligodendrocytes (paired t test, t(3) = 2.43) or OPCs (paired t test, t(3) = 0.516). Points represent within bulb means for each animal, p values represent paired t tests, n = 4 animals. A, Comparison of LOT oligodendrocyte lineage cell density in Naris Occlusion versus Control animals. Taken as a group, Control and Naris Occlusion animals had similar numbers of oligodendrocytes (t test, t(5.41) = 0.192) and OPCs (t test, t(5.22) = −0.668). Points represent per-animal means, p values represent t tests, n = 8 animals. B, Comparison of LOT oligodendrocyte lineage cell density in the left and right bulbs of Control animals. We found no significant differences in oligodendrocyte (paired t test, t(3) = −0.995) or OPC density (paired t test, t(3) = 0.085) in LOT, quantification similar to A, n = 4 animals. B', Comparison of LOT oligodendrocyte lineage cell density in Open and Occluded bulbs of Naris Occlusion animals. We found no significant differences in oligodendrocyte (paired t test, t(3) = 0.606) or OPC density (paired t test, t(3) = −2.05) in LOT, quantification similar to A', n = 4 animals. B, Comparison of LOT oligodendrocyte lineage cell density in Naris Occlusion versus Control animals. Taken as a group, there was again no significant difference in oligodendrocyte lineage cell density between Naris Occlusion and Control animals (oligodendrocytes t test, t(4.45) = −1.86, OPCs t test, t(5.87) = 0.806). Quantification was similar to A, n = 8 animals. C, Comparison of g ratios of myelinated axons in the left and right LOTs of Control animals. There was no significant difference in mean g-ratio between the left and right sides of Control animals (t test, t(260.5) = 1.02, n = 265 axons from 5 animals). Points in scatter plot represent individual axons, error bars in bar plot inset represent per-group SD. C', Comparison of myelinated axons in the Open and Occluded LOTs of Naris Occlusion animals. There was no significant difference in mean g-ratio between the Open and Occluded sides of Naris Occlusion animals (t test, t(399.6) = 1.43). Data presented as in C, n = 402 axons from 8 animals. C, Comparison of myelinated axons in Naris Occlusion versus Control animals. Treated as a group, there was a significant difference between the mean g-ratio of Naris Occlusion and Control animals (t test, t(581.7) = 4.34).
Similarly, in the LOT, we found no significant differences between the left and right sides of Control animals in LOT oligodendrocytes (paired t test, t(3) = −0.995, p = 0.393, n = 4 animals) or OPCs (paired t test, t(3) = 0.085, p = 0.937, n = 4 animals). There were no significant differences between the Open and Occluded sides of Naris Occlusion animals in LOT oligodendrocytes (paired t test, t(3) = −0.606, p = 0.588, n = 4 animals) or OPCs (paired t test, t(3) = −2.05, p = 0.132, n = 4 animals). There was also no significant difference between the main groups (Control and Naris Occlusion) in LOT oligodendrocytes (Fig. 5, row B, t test, t(4.45) = −1.86, p = 0.13, n = 8 animals) or OPCs (t test, t(5.87) = −0.866, p = 0.421, n = 4 animals/group).
While we found no change in the density of oligodendrocyte lineage cells, myelin sheaths themselves can undergo remodeling by preexisting oligodendrocytes (Auer et al., 2018). To determine whether myelin sheaths changed in response to UNO, we analyzed myelinated axons in the LOT of Naris Occlusion and Control animals by measuring the g-ratio (see Materials and Methods; Fig. 2F). We noted no significant differences between the left and right LOTs of Control animals (t test, t(260.5) = 1.02, p = 0.307, n = 265 axons from 5 animals) or between the Open and Occluded sides of Naris Occlusion animals (t test, t(399.6) = 1.43, p = 0.154, n = 402 axons from 8 animals). However, when we compared the main groups (Control and Naris Occlusion), we noted that axons from Naris Occlusion animals had significantly lower g-ratios (thicker myelin relative to axon thickness) than Control (t test, t(581.7) = 4.34, p = 1.1e-5, n = 667 axons; Fig. 5, row C).
Together, we noted a significant change in g-ratios with no evidence of changes in oligodendrocyte lineage cells in response to 30 d of UNO in adult animals. The system-wide changes in Naris Occlusion animal myelin sheaths indicate that myelin remodeling is occurring independently of oligodendrocyte differentiation and maturation (see Discussion).
MC firing properties change following UNO
Sensory deprivation and enrichment can have dramatic effects on axon morphology and neuronal firing properties (Kuba et al., 2010; Evans et al., 2015; Jamann et al., 2021). We have described subtle length changes in the AIS and nodes of Ranvier following UNO, but does UNO also affect MC physiology?
To assess MC physiology, we performed whole-cell current clamp recordings on UNO and Control mice. Cells were held at –60 mV for all experiments. Membrane resting potential was not significantly different between the Control and Naris Occlusion groups (Control −52.9 ± 0.26 mV, n = 23 cells, Naris Occlusion −52.5 ± 0.423 mV, n = 19 cells, t test, t(32.3) = −0.155, p = 0.878, presented as mean ± SEM), or between the Control, Open, and Occluded groups (Control −52.9 ± 0.256 mV, n = 23 cells, Occluded = −52.8 ± 0.76 mV, n = 12 cells, Open = −52 ± 0.915 mV, n = 7 cells, ANOVA, F(2,39) = 0.0436, p = 0.957, presented as mean ± SEM). MCs are involved in complex circuits and receive both excitatory and inhibitory inputs which affects firing patterns and synchrony (Schoppa and Westbrook, 2001; Egger and Urban, 2006; Fukunaga et al., 2014). To isolate MCs from OB circuits and measure intrinsic firing patterns, we performed all recordings in the presence of the glutamatergic inhibitors DNQX (10 μm), APV (50 μm), and the GABAergic inhibitor gabazine (5 μm). MCs have large, complex dendrites and axons which can easily be damaged during the slicing procedure. To ensure damaged MCs were not influencing recordings, we performed a subset of recordings with biocytin in the patch pipette (∼2 mg/ml), then fixed and immunolabeled the slices for AISs after recording (Fig. 6A; see Materials and Methods). Our data indicate that the majority of recorded cells contained a discernible, AnkG+ AIS and axon (89%; Fig. 6A, white arrow), and a full primary dendrite innervating a glomerulus (72%; Fig. 6A, yellow arrow, n = 18 cells, 5 Control, 13 Naris Occlusion).
MC spiking and APs adapt to UNO. A, Example of a MC filled with biocytin (magenta) and immunolabeled for the AIS (AnkG, green). White arrowhead indicates the cell's AIS, and the yellow arrowhead indicates the primary dendrite innervating a glomerulus. Scale bar: 25 µm. B, Diagram of an AP train with annotations of relevant features. The annotation letters correspond to the features measured in the plots below. C, Example current clamp trace from a Control, Open, and Occluded MC in response to a 150-pA 1-s current injection. D, MC spiking frequency across 25-pA current steps from 0 to 500 pA. There was a significant interaction between current and Occluded (GLM, Current × Occluded t = −2.91, p = 0.00375), and current and Open cells versus Control (GLM, Current × Open, t = −3.39, p = 0.000721). P values for the significant interactions are shown on the plot. N cells = 23 Control, 13 Occluded, and 7 Open for all experiments unless otherwise noted. E, Interspike interval time for current steps, similar to D. The main model was not significantly different from the current-only model, indicating Group had no significant effect on the overall interspike interval. F, Plot of CV for interspike interval. The Open group had significantly lower CV than control (t = −2.79, p = 0.00541). P value for the Open term is shown on the plot. G, Example traces of single APs from a Control bulb MC (black), Open bulb MC (blue), and an Occluded bulb MC (magenta) in response to a 150-pA current step. H, Plot of AP width in response to current steps. APs from Occluded bulb MCs were significantly wider than Control (GLM, t = 2.61, p = 0.00933). P value for the Occluded term shown on the plot. I, Plot of AP threshold in response to current steps. Open bulb MCs had only marginally lower thresholds than Control (GLM, t = −1.92, p = 0.0557), but the effect was not significant at α = 0.05. J, Plot of AP amplitude across current steps. APs from Occluded MCs had significantly larger amplitudes compared with Control (t = 4.27, p = 2.23e-5). The Current × Occluded interaction also had a significant effect on AP amplitude (t = −1.99, p = 0.0469). P values for the significant terms are shown on the plot. K, Plot of AP amplitude and resting membrane potential at a 150-pA step for a subset of cells. There was no significant correlation between amplitude and resting membrane potential for any of the groups in the membrane potential range observed (Pearson's product-moment correlation, Control, t(17) = −1.22, p = 0.238, n = 18 cells, Open t(4) = 1.39, p = 0.236, n = 5 cells, Occluded t(7) = 0.626, p = 0.551, n = 8 cells). All plots are mean ± SEM.
MCs are diverse, both in terms of morphology and physiology, displaying both bursting and regular spiking patterns in response to current injections (Chen and Shepherd, 1997; Padmanabhan and Urban, 2010, 2014; Fadool et al., 2011). We first investigated the spiking behavior of MCs in response to a series of current injections from 0 to 500 pA with a step size of 25 pA (Fig. 6). To test whether spiking patterns were different, we fit generalized linear models (GLMs) to the spike frequencies. The main model was of the form:
For MC spiking frequency, the main model was significant compared with the current-only model (Fig. 6D, F = 8.57, p = 8.7e-7). In the main model, we found that firing rates increased significantly as a function of current (GLM, t = 20.5, p < 2e-16). Additionally, the interactions between current and Occluded, and current and Open had a significant effect on firing rates versus Control (GLM, Current × Occluded interaction t = −2.91, p = 0.00375, Current × Open interaction t = −3.39, p = 0.000721, n cells = 23 Control, 13 Occluded, and 7 Open; for all GLM results, the reported t and p values are rounded to three significant figures in the text). See Table 4 for the model summary.
Spike count GLM summary
Previous computational modeling and experimental studies have indicated that spike timing (measured via interspike interval or onset threshold) is sensitive to minor changes in AIS length (Baalman et al., 2013; Evans et al., 2015; Jamann et al., 2021). Furthermore, voltage-gated potassium channels (Kv), known to be present at the AIS, strongly influence spiking diversity, AP shape, and AP reliability (Debanne, 2004; Kole et al., 2007; Padmanabhan and Urban, 2014). To test whether the MC morphologic adaptations were reflected in spiking patterns, we next measured the mean interspike interval at each current step. We found that the main model was not significantly different from the current-only model (Fig. 6E, F = 0.848, p = 0.495). Only current was significant in the main model, with interspike interval decreasing as a function of increasing current (GLM, t = −10.5, p < 2e-16). There were no significant interactions between any group and current compared with Control (see Table 5 for the model summary; n cells = 23 Control, 13 Occluded, and 7 Open).
Interspike interval GLM summary
4-Aminopyridine (4AP)-sensitive voltage-gated potassium channels are implicated in bursting/spiking variability in MCs (Balu et al., 2004; Padmanabhan and Urban, 2014). To test whether MCs had altered spiking variability, we calculated the coefficient of variation (CV), a unitless ratio of the SD/mean for interspike interval (Padmanabhan and Urban, 2010, 2014). The main model of interspike interval CV was significantly different from the current-only model (Fig. 6F, F = 3.27, p = 0.0114). Interspike interval CV decreased significantly as a function of increasing current (GLM, t = −3.28, p = 0.0011), and the Open group had significantly lower CV than Control (GLM, t = −2.79, p = 0.00541; see Table 6 for model summary; n cells = 23 Control, 13 Occluded, and 7 Open).
Interspike interval CV GLM summary
We next investigated the kinetics of individual APs in the Control, Open, and Occluded groups. We extracted the first AP from spike trains in current steps from 50 to 500 pA (leaving out the lower current steps where few spikes were evoked) to measure AP kinetics. We again fit GLMs to compare relationships.
AP width (measured as full width at half maximum) is sensitive to changes in Kv channel composition at the AIS, and influences downstream synaptic efficiency (Debanne, 2004; Kole et al., 2007). The main model for AP width was significantly different from the current only model (Fig. 6H, F = 11.6, p = 4.4e-9). Current was significant in the main model (GLM, t = −3.17, p = 0.00159), and APs from Occluded MCs were significantly wider than Control (GLM, t = 2.61, p = 0.00933). There were no significant interactions between group and current (see Table 7 for the model summary; n cells = 23 Control, 13 Occluded, and 7 Open).
AP Width GLM summary
AP threshold marks the point where an AP becomes an all-or-none response. We defined AP threshold as the voltage when the derivative of the rising phase of the AP reached 25 V/s. AP threshold is typically inversely related to AIS length, with longer AISs often displaying lower thresholds (Kuba et al., 2010; Jamann et al., 2021). AP threshold is also affected by channel composition (Katz et al., 2018). The main model for AP threshold was significantly different from the current only model (Fig. 6I, F = 6.98, p = 1.67e-5). Current had a significant effect on threshold in the main model (GLM, t = 2.72, p = 0.0067). Surprisingly, APs from Open MCs were only marginally lower than Control (GLM, t = −1.92, p = 0.0557). There were no significant interactions between group and current (see Table 8 for the model summary; n cells = 23 Control, 13 Occluded, and 7 Open).
AP Threshold GLM summary
The main model for AP amplitude (measured from threshold voltage to peak voltage) was significantly different from the current-only model (Fig. 6J, F = 21.9, p < 2.2e-16). AP amplitude increased significantly as a function of increasing current in the main model (GLM, t = 4.27, p = 2.23e-5). APs from Occluded bulb MCs had significantly larger amplitudes than Controls (GLM, t = 5.21, p = 2.64e-07). There was also a significant interaction between Current and Occluded versus Control (GLM, Current × Occluded, t = −1.99, p = 0.0469). These results are consistent with previous sensory deprivation studies in the chick auditory brainstem (Kuba et al., 2010). See Table 9 for the model summary; n cells = 23 Control, 13 Occluded, and 7 Open. To control for different resting potentials possibly affecting AP amplitude, we plotted cell resting potential versus amplitude for a subset of cells in each group. There was no significant correlation between amplitude and resting membrane potential for any of the groups in the membrane potential range observed (Fig. 6K, Pearson's product-moment correlation, Control t(17) = −1.22, p = 0.238, n = 18 cells, Open t(4) = 1.39, p = 0.236, n = 5 cells, Occluded t(7) = 0.626, p = 0.551 n = 8 cells), which is consistent with previous results (Balu et al., 2004).
AP Amplitude GLM summary
Discussion
In the OB, trains of APs at the γ frequency are generated in response to odors (Eeckman and Freeman, 1990; Kashiwadani et al., 1999; Bathellier et al., 2006; Li and Cleland, 2017). These oscillations must reliably travel large distances for further processing in regions such as the piriform cortex, where precise, synchronized arrival determines whether a cell fires (Franks and Isaacson, 2006; Luna and Schoppa, 2008; Nagayama et al., 2010; Bolding and Franks, 2018). Despite the importance of reliable AP transmission over large distances, little is known about the myelinated axons that generate and propagate APs in the OB, and whether they adapt in response to changes in sensory input. Here, we characterized the myelinated axons of MCs in the OB and LOT, and tested whether they adapt to altered sensory input using UNO. We found that 30 d of adult UNO led to an increase in AIS length of 8% (∼2 µm) in MCs from the Open bulbs of Naris Occluded animals relative to both Occluded bulbs and bulbs from Control animals (Fig. 4). Because of nonlinear AP initiation, a change of similar magnitude was predicted by computational modeling to result in significant physiological changes in AP characteristics (Baalman et al., 2013). Indeed, we found Naris Occlusion had a significant effect on spiking patterns, reducing spiking frequency and spike variability (Fig. 6D,E; Tables 4, 6).
A recent study examined how axons in the OB adapt to short-term deprivation (Galliano et al., 2021). Following 24-h UNO, the group found a significant change in AIS length in a population of dopaminergic periglomerular cells without significant differences in MC AIS length. We detected changes in MC AISs after 30 d of occlusion. Together, these data indicate that a longer period of deprivation is required for MCs to adapt than periglomerular cells.
Of note were the significant differences in AP width in MCs from the Occluded side, despite the Occluded cell AIS length being similar to Controls (Figs. 4, 6). Why would AP width increase in Occluded MCs? One possibility is adaptation in Kv channel composition or number. More Kv channels at the AIS are known to broaden AP width (Debanne, 2004; Kole et al., 2007), and 4AP-sensitive Kv channels increase MC firing pattern diversity (Padmanabhan and Urban, 2010, 2014). Ion channel adaptations, including voltage-gated sodium channel phosphorylation are known to produce functional changes in AP generation at the AIS (Evans et al., 2015). Broadening of APs would be expected to lead to an increase in synaptic strength at proximal axonal terminals (Kole et al., 2007). This may translate to a counterintuitive increase in odor sensitivity on the Occluded side because of a higher probability of depolarizing a downstream cell in the piriform cortex. This hypothesis is supported by a previous study showing that Naris Occlusion mice outperformed Control mice in a habituation-dishabituation olfactory discrimination task (Angely and Coppola, 2010). Our results could represent a novel mechanism to explain that finding, although future work will determine whether AP shape changes measured in Occluded bulbs translate to physiological adaptations in synapses at the piriform cortex.
We found that the variance of AIS lengths is larger in Naris Occlusion animals compared with Control animals. This increased diversity was counter-intuitively reflected in an overall lower interspike interval CV in the Open group compared with Control (Fig. 6F; Table 6). What explains this increased length variance and decreased spiking variance? MCs are biophysically diverse cells, with cell-level firing differences largely attributed to local circuits and OB oscillations, as well as ion channel diversity, particularly in Kv channels (Heyward et al., 2001; Padmanabhan and Urban, 2010, 2014; Angelo and Margrie, 2011). Spiking diversity is thought to increase the information carrying capacity of neurons by de-correlating the firing patterns of groups of neurons (Padmanabhan and Urban, 2010). One hypothesis is that the increased range of AIS lengths in Naris Occlusion animals reflects an attempt to encode more information with half the resources. The less variable spiking patterns we measured in vitro are likely different from MC spiking patterns in vivo. In vivo, respiratory rhythm, local circuits/oscillations, and intrinsic membrane oscillations lead to different MC spiking patterns than in vitro (Cang and Isaacson, 2003; Cury and Uchida, 2010; Angelo and Margrie, 2011; Li et al., 2017). Additional experiments will determine how neuronal spike rates change over the course of UNO in vivo.
Nodes of Ranvier allow for fast saltatory conduction along myelinated axons by regenerating propagating APs at successive gaps in the myelin sheath (Susuki et al., 2013; Castelfranco and Hartline, 2016). Recent work has implicated nodes of Ranvier as potential sites for plasticity (Ford et al., 2015; Arancibia-Cárcamo et al., 2017; Dutta et al., 2018; Cullen et al., 2021). We measured nodes of Ranvier in the LOT and ALAC following UNO. We found that node lengths in LOT were significantly shorter when comparing Naris Occlusion animals (both Open and Occluded sides) to Controls (both left and right sides; Fig. 4). Surprisingly, we also found that nodes of Ranvier in the ALAC of Naris Occlusion animals were significantly longer than nodes of Ranvier in the ALAC of Control animals (Fig. 4). The ALAC carries bilateral olfactory axons projecting to both bulbs and from both brain hemispheres (Brunjes, 2012; Rabell et al., 2017; Collins et al., 2018). Changes in the ALAC could represent a possible mechanism for bilateral control over LOT axonal activity. While it is unclear why Naris Occlusion animals exhibit different node lengths compared with Controls, bilateral adaptations following UNO are not unprecedented. In particular, a study investigating in vivo synaptic responses of olfactory receptor neurons following UNO found that odor-induced synaptic release was equally reduced in the Open and Occluded sides compared with Controls (Kass et al., 2013). Similarly, gene expression in the olfactory mucosa changes on both the Open and Occluded sides compared with Controls (Coppola and Waggener, 2012). Bilateral changes highlight the importance of using separate Control animals in UNO studies in addition to within animal comparisons (Coppola, 2012).
Despite finding no changes in oligodendrocyte lineage cells, we found that, when counted as a group, the g-ratios on both sides of Naris Occlusion animals were significantly lower than the g-ratios in Control animals (Fig. 5). This bilateral change in myelination mirrors the bilateral adaptations in node length seen in Naris occlusion animals (Fig. 4). What mediates nodal length changes? Several potential mechanisms have been proposed, including mediation by perinodal astrocytes (Dutta et al., 2018), breakdown of the nodal complex (Huff et al., 2011), or microglial involvement (Hughes and Appel, 2020). Additionally, previous work in the visual system reported monocular deprivation caused proliferation and maturation of oligodendrocytes, as well as shorter sheaths on the deprived side (Etxeberria et al., 2016). Motor learning, somatosensory stimulation, and optogenetic stimulation of neuronal activity in motor cortex have also been reported to increase oligodendrocyte lineage cell proliferation, differentiation, and cause myelin sheath remodeling (Gibson et al., 2014; McKenzie et al., 2014; Xiao et al., 2016; Hill et al., 2018; Hughes et al., 2018). However, lower g-ratios in the same regions as we saw nodal length changes could also indicate myelin remodeling by existing oligodendrocytes, a phenomenon well documented in animals undergoing learning active learning paradigms (McKenzie et al., 2014; Xiao et al., 2016; Hill et al., 2018; Hughes et al., 2018).
How could thicker myelin affect olfaction? Conduction velocity is increased in axons with thicker myelin sheaths (Fields, 2015). Based on our finding of lower g-ratio, one would assume that conduction velocity is faster in UNO animals compared with Controls. Faster conduction velocity would result in more reliable simultaneous arrival of APs to piriform cortex. While speed is not thought of as essential to olfactory processing, coordinated signals and the reliable transmission of oscillations are associated with olfactory processing and learning (Laurent et al., 1996; Franks and Isaacson, 2006; Kay et al., 2009; Losacco et al., 2020). However, it is hard to predict how adaptations in myelinated axons will affect the system-level information transfer in oscillations. Computational modeling studies emphasize the importance of myelin and conduction speed tuning for oscillation synchrony (Pajevic et al., 2014), and our lab has previously shown that mild myelin disruption in the Plp1-null mouse leads to increased oscillatory power in the theta and β frequencies (Gould et al., 2018). Future work will elucidate the effects of lower g-ratios in Naris Occlusion animals on conduction velocity and downstream signal integration. Our work provides evidence for a novel form of cellular plasticity in the olfactory system where myelinated axons adapt to changing experience input in adult animals. It is unclear what the systems level consequences of these novel adaptations are, but they are likely to have important consequences for downstream olfactory system information processing.
Footnotes
This work was supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health Awards F31 DC018459 (to N.M.G.) and R01 DC000566 (to D.R.). This work was also supported by the University of Colorado Anschutz Neuroscience Training Program Grant T32 HD041697 and the Colorado Clinical and Translation Sciences Institute training Fellowship TL1 TR001082 (to N.M.G.). We thank Dr. Jennifer Bourne, manager of the University of Colorado Anschutz Medical Campus Electron Microscopy Center, for advising on the design of, and performing, electron microscopy experiments for this publication; Nicole Arevalo, MA, for her expertise and extensive assistance with experiments, animal care, and animal breeding; Katie Given, MS, for her invaluable support and technical expertise when performing experiments for this manuscript; Dr. John Caldwell and Dr. Nathan Schoppa for experimental design and data analysis advice and for reviewing and providing helpful input to this manuscript; and Dr. Alexandria Hughes for helpful conversations, support, and review of this manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Diego Restrepo at diego.restrepo{at}cuanschutz.edu
