The mouse olfactory peduncle 4: Development of synapses, perineuronal nets and capillaries

Abstract

Olfaction is critical for survival in neonatal mammals. However, little is known about the neural substrate for this ability as few studies of synaptic development in several olfactory processing regions have been reported. Odor information detected in the nasal cavity is first processed by the olfactory bulb and then sent via the lateral olfactory tract to a series of olfactory cortical areas. The first of these, the anterior olfactory nucleus pars principalis (AONpP), is a simple, two layered cortex with an outer plexiform and inner cell zone (layers 1 and 2, respectively). Five sets of studies examined age-related changes in the AONpP. First, immunocytochemistry for glutamatergic (VGlut1 and VGlut2) and GABAergic (VGAT) synapses demonstrated that overall synaptic patterns remained uniform with age. The second set quantified synaptic development with electron microscopy and found different developmental patterns between layers 1 and 2. As many of the interhemispheric connections in the olfactory system arise from AONpP, the third set examined the development of crossed projections using anterograde tracers and electron microscopy to explore the maturation of this pathway. A fourth study examined ontogenetic changes in immunostaining for the proteoglycans aggrecan and brevican, markers of mesh-like extracellular structures known as perineuronal nets whose maturation is associated with the end of early critical periods of synaptogenesis. A final study found no age-related changes in the density of vasculature in the peduncle from P5-P30. This work is among the first to examine early postnatal changes in this initial cortical region of the olfactory system.

Graphical Abstract

A series of studies characterized various aspects of the development of the anterior olfactory nucleus, pars principalis. Immunohistochemistry was used to examine development of glutamatergic and GABAergic synapses (shown in graphical abstract), perineuronal nets, and capillaries. Electron microscopy was used to further examine synaptic maturation, and anterograde tracers were used to visualize development of contralateral projections.

Odors binding to olfactory sensory neurons in the nasal cavity create patterns of neural activation within the olfactory bulb (OB). The lateral olfactory tract (LOT) carries information from the OB to the olfactory cortices (Mori, 2014). The olfactory peduncle, found just caudal to the OB, contains the first of these cortices: the anterior olfactory nucleus (AON; Fig. 1; Brunjes et al., 2005). The AON has two subdivisions: “pars externa”, a thin ring of cells found at the rostral end of structure, and the much larger “pars principalis” (AONpP), the primary region examined in the present work. It is a simple, two-layered cortex with an outer molecular layer (layer 1) and a deeper cellular region (layer 2) containing pyramidal projection neurons and interneurons similar to those found in all other forebrain cortices (Kay and Brunjes, 2014; Brunjes and Osterberg, 2015). Haberly (2001; Haberly and Price 1978b) subdivided the area into 4 zones based on cytoarchitecture and connections: pars lateralis (found just deep to the LOT; pPl), pars dorsalis (pPd), pars medialis (pPm) and pars ventroposterior (pPvp). AONpP has rich connections with most of the ipsilateral olfactory forebrain, including back projections to OB and feed-forward projections to the anterior portion of the piriform cortex (APC) and the olfactory tubercle, and substantial contralateral projections via the anterior commissure (AC) to the OB, AON and APC (Brunjes et al., 2005; Illig and Eudy 2009; Padmanabhan et al., 2019). The AON’s position in the sensory circuit allows it to interact with most of the areas that process odor information. The region has been demonstrated to compare binasal cues to aid in the localization of odors (Kikuta et al., 2010) and, through interactions with the hippocampus, process spatial and temporal odor memories (Aqrabawi and Kim, 2018a,b).

Fig. 1.

a. Coronal section (red box) through the olfactory peduncle revealing the region studied (grey). b. Photomicrograph of Nissl-stained section through peduncle. The deep border of the lateral olfactory tract is depicted by a green line; the two blue lines indicate the boundaries of the outer plexiform (layer 1) and inner cell dense (layer 2) regions of AONpP. The area outlined by the red box is depicted in greater detail in (c), where, once again, the green line indicates the deep border of the LOT and the blue lines the borders of layers 1 and 2. The deepest region of the peduncle contains the anterior limb of the anterior commissure (ALAC) and the subventricular zone (SVZ), which includes the rostral migratory stream. Depicted on the right side is a) a single axon (dark blue) leaving the LOT and descending into the superficial portion of layer 1 (layer 1a), and b) the sites of termination of axons from the ipsilateral anterior and posterior piriform cortices (yellow) and tenia tecta and dorsal peduncular cortex (red, Luskin and Price 1983). Scale bar = 100 μm.

Relatively little is known about the development of AONpP. It arises from the ventral pallium of the developing telencephalon (Puelles et al., 2013), and, in the rat, its constituent neurons are born between embryonic (E) days 15–21. Neurons are added from the outside to the inside, the reverse pattern as seen in other forebrain cortices (Bayer, 1986; Brunjes and Osterberg, 2015). The LOT matures very early in development in mice and rats (Lopez-Mascaraque et al., 1996; Sato et al., 1998) with mitral cell axons entering it as early as E11.5 and the tract beginning to emerge by E13 (Walz et al., 2006). Axons first enter the AC much later (E16; Silver et al., 1982). A similar pattern is seen in the onset and rapid phase of myelination in the two tracts: the LOT undergoes myelination between postnatal day (P) 9–13, but the process does not occur in the AC until P11–14 (Collins et al., 2018). Considerable volumetric growth occurs in AONpP from postnatal day P10 to P20, with reduced expansion from P20 to P30; similar changes are found in the growth in dendrites of its pyramidal cells (Brown and Brunjes, 1990; Brunjes et al., 2014; Brunjes and Kenerson, 2010).

Many synaptic connections in the AONpP are segregated and orderly. Axon collaterals descend from the LOT into the superficial portion of layer 1 (layer 1a) where they contact the apical dendrites of pyramidal cells (Fig. 1). While many regions innervate AONpP (including hypothalamic, amygdaloid, thalamic and neocortical regions and neuromodulators including acetylcholine, histamine, norepinephrine, orexin, and serotonin, Brunjes, 2013; Brunjes et al., 2005, 2014), Luskin and Price (1983a,b) defined two major associational fibers systems serving the area. The first is a topographically organized input from the APC and posterior piriform cortex (PPC) into the deep portion of layer 1 (layer 1b; Fig. 1). Axons from the APC are found more superficially in this zone than those from the PPC. Furthermore, axons from APC cells located near the LOT preferentially project to pPl and pPvp, while cells located laterally to the LOT target pPd and pPl. Fibers from cells near the LOT in the PPC do not project to AONpP, but the remaining regions preferentially target layer1b in pPd and pPm. Luskin and Price’s (1983a,b) second projection targets layer 2 and arises from two small regions in the peduncle. The tenia tecta exhibits a topographic projection, with the inferior region projecting to pPvp and pPl and the superior to pPd and pPm while the dorsal peduncular cortex projects uniformly to the entire AONpP.

The present work contains five studies designed to expand our understanding of the development of the AONpP. The first three examined age-related changes in synapses. The density and distribution of both glutamatergic (VGlut1 and VGlut2) and GABAergic (VGAT) synapses was examined in postnatal day (P) 5–30 mice with immunofluorescence. A second set of quantitative electron microscopic studies examined synapses in P5, P11, P13 and mature function in the interhemispheric pathway (Collins et al., 2018). Because little is known about these crossed projections a third set of studies characterized both the development of axons from pPl to the contralateral AONpP (Illig and Eudy, 2009) and the synapses formed by the projection. A fourth study examined the development of perineuronal nets. These mesh-like extracellular structures envelop neuronal processes and regulate the number and types of synapses. Their maturation is associated with the end of early critical periods of synaptogenesis in several brain regions (Clarris et al., 2000; Valenzuela et al., 2014; Bikbaev et al., 2015; Sorg et al., 2016; van ‘t Spijker et al., 2017; Lensjø et al., 2017; Quraishe et al., 2018). Finally, the rapid changes associated with growth in the structure necessitate substantial energy: a small study addressed this issue by examining age-related changes in vasculature.

Methods

The studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All protocols were approved by the University of Virginia’s Institutional Animal Care and Use Committee (PHS service assurance number #A3245–01; USDA registration #: 52-R-0011). All surgeries were performed in aseptic conditions with isoflurane anesthesia, and all efforts were made to minimize suffering. C57Bl/6J mice were obtained from Jackson Labs, Bar Harbor, ME. Hemizygous knock-in Gad1 EGFP (Δ neo) mice on a C57BL6/J (RRID: ISMR_JAX: 000664) background, were kindly provided by Dr.Yuchio Yanagawa (Gunma University Graduate School of Medicine; Tamamaki et al., 2003, Kay and Brunjes 2014), referred to here as “GAD67-GFP”.

Mice were housed in standard polypropylene cages with food (8604, Harlan, Frederick, MD) and water ad libitum. The colony was maintained on a 12:12 light:dark cycle in a temperature-and humidity-controlled room. Preliminary results indicated no male/females differences at these early ages; therefore, each of the studies below employed mixed sex groups. Five sets of studies were performed:

1). Synaptic development in pars lateralis of the AONpP: Changes in glutamatergic (VGlut1 and VGlut2) and GABAergic (VGAT) synapses.

Tissue preparation: Mice were deeply anesthetized with sodium pentobarbital (Euthasol; 0.39 mg drug/g body weight; 150 mg/kg; APP Pharmaceuticals; Schaumburg, IL., RRID:SCR_004098) and perfused intracardially with 0.01 M phosphate buffered saline containing 0.001% heparin (PBS; pH 7.4) followed by 4% buffered formaldehyde. Brains were post-fixed for at least 2 hrs. Fluorescent immunohistochemistry was used to stain free-floating 60μm-thick coronal vibratome sections. Tissue sections from postnatal day (P) 5, 10, 15, 20 and 30 pups representing 3–4 different litters/age was used. Subjects were perfused with freshly prepared paraformaldehyde and allowed to postfix for 2 hrs at 4°C. Sections were rinsed 4 times in 0.01M phosphate buffered saline (PBS pH 7.4). Next, the tissue was incubated in 0.01 M citrate buffer pH 8.5 at 80°C (2 × 15 min, Jiao et al., 1999). After cooling at room temperature for 5 min, the sections were washed in PBS (2 × 2.5 min), permeabilized in 0.03% Triton in PBS (TW: 4 × 5 min), and placed into blocking solution (0.5% normal donkey serum in TW; Jackson ImmunoResearch,West Grove PA) for one hr. Sections were then placed in a solution containing three primary antibodies (VGlut1, VGlut2 and VGAT; Table 1) overnight. They were then washed (PBS 4X5 minutes) and incubated in secondary antibody (1/250 to 1/450 in TW: Jackson ImmunoResearch donkey anti-rabbit: catalog number 711–165-152 or 711–545-152; donkey anti-Guinea pig: 706–165-148; donkey anti-mouse: 715–485-150 or 715–545-151; donkey anti-rat: 712–165-153) for 1 hr, washed again (PBS 4 × 5 min) and mounted on slides with SlowFade mounting media (Invitrogen: S36937). Control experiments in which the primary antibodies were deleted resulted in no staining. All results were consistent with those previously reported.

Table 1.

Antibodies

Antibody	Immunogen	Supplier	Catalog /lot number	RRID	Species	Dilution
Aggrecan	Aa 1–298 of human aggrecan	Proteintech Rosemont, IL	13880–1-AP 00023864	AB_2722780	Rabbit polyclonal	1/200
Brevican	aa 219–655 of rat brevican, clone N294A1/10	NeuroMab, Davis CA	75–294 449–3AK-25	AB_2315824	Mouse monoclonal	1/200
CD-31 (PECAM-1)	aa 610–681 of mouse CD31	Dianova, Hamburg Germany	DIA-310 1662/08	AB_2631039	Rat monoclonal	1/250
VGAT (VIAAT (vesicular GABA transporter/vesicular inhibitory amino acid transporter)	aa 75–87 of rat coupled to key-hole limpet hemocyanin	SYSY, Göttingen, Germany	131 011 131011/52	AB_2619818	Mouse monoclonal	1/500
VGlut1	Strep-Taq fusion protein of rat VGlut1 aa 456–560	SYSY, Göttingen, Germany	135 302/ 135302/30	AB_887877l	Rabbit polyclonal	1/5000
VGlut2	guinea pig recomb inant GST-tagged VGlut2	Millipore (Temecul a, CA)	AB2251/2712426	AB_2665454	Guinea pig polyclonal	1/2500

Imaging and Analysis: Confocal images were obtained using a 40X planapo objective (NA 0.95) beginning at the pial surface over the LOT and extending perpendicularly to the subventricular zone (SVZ, Fig. 1c, Fig 2). Images gathered from a single section were combined in a montage and the brightness and contrast of the images was adjusted to be uniform across the tissue with Adobe Photoshop (San Jose, CA; RRID:SCR_014199) and the images combined into montages (Adobe Illustrator; RRID:SCR_010279). Default threshold settings were applied to all images with ImageJ (Rasband et al., 1999–2007). Regional differences in staining density were quantified by measuring the area fraction (the proportion of labeled pixels in a test field) for each antigen. Measurements were made in six regions: the outer, middle and deep thirds of both layer 1 and layer 2. In order to control for potential inter-animal variations in staining efficacy, for each section data from the six test zones were combined and this overall area fraction defined as “100%”. Values found in the 6 test areas were expressed as the deviation from this average.

Fig 2.

Developmental changes in immunostaining for markers of two varieties of glutamatergic synapses (VGlut1, green and VGlut2, red) and GABAergic synapses (VGAT, blue). 1,2 =layers 1 and 2, respectively. These photomicrographs represent the portion of AON pars lateralis depicted in Fig 1b, c. Scale bar = 100 μm.Fig. 3. a1–a4. Higher power details of VGLUT 1, 2 and VGAT staining from a P15 subject. VGAT synapses often encircle somata (e.g., arrows), while staining for the other two antigens was more diffuse. Scale bar = 100 μm. b. Age-related changes in the staining density in the deep, middle and superficial portions of layer 1 and layer 2. These measures were normalized for each antigen in each section by considering the average area fraction of staining across the entire test area as “100%” (dotted line). Therefore, areas with densities above or below 100% were more or less dense than the section mean. Patterns were quite similar across ages: for example, in layer 1 VGlut2 and VGAT was densest just below the LOT, while VGlut1 staining was more uniform.

2). Developmental changes in synapse density, synapse length, and terminal bouton area.

Tissue Preparation: Four ages were selected for study: P5 (early in olfactory system development), P13 and P17, (ages which coincide with the period of rapid myelination of the AC; Collins et al., 2018), and P30 (when OB growth plateaus; Hinds and Hinds, 1976). At least 3 animals were examined per age. Mice were anesthetized as above and perfused with PBS followed by a 2% glutaraldehyde, 2% paraformaldehyde solution. Brains were dissected, post-fixed at least 12 hrs, and sectioned at 60μm with a vibratome. Tissue was rinsed in 0.1M phosphate buffer (PB) three times followed by an hour incubation in a 1% osmium tetroxide solution in PB. It was then rinsed two more times with PB, once with 50% EtOH, once with 70% EtOH, and finally placed in 4% uranyl acetate in 70% EtOH overnight. Tissue was then serially dehydrated with ethanol, placed in a 1:1 solution of acetone and EPON resin and kept overnight. The acetone-EPON solution was exchanged for full EPON for at least 2 hrs. Sections were flat-embedded in aclar and placed in an oven at 60℃ overnight. The tissue was then transferred to EPON-filled capsules and again placed in an oven overnight. After polymerization, the tissue was sectioned with an ultramicrotome (Leica UC7) at 80nm.

Imaging and analysis: A JEOL 1010 electron microscope fitted with a 16MB camera was used to obtain images at 10,000X, yielding a pixel size of 0.73nm, sufficient for synaptic membranes to be discriminated. Between 8 and 15 images were collected from layer 1a, layer 1b and the superficial and deep halves of layer 2 of pPl per animal at each age. Measurements were made with Image ProPlus 5.1 (Media Cybernetics, Silver Spring, MD). Synapse density was estimated by dividing total numbers of synapses by the total area of neuropil, defined as the region remaining after the exclusion of any space occupied by cell somas, myelinated axons, or blood vessels. Synapses were counted and measured if a full postsynaptic density and synaptic cleft could be discerned. Perforated synapses were measured as single synapses. In order to correct for potential measurement biases due to large differences in synapse terminal size relative to section thickness, volumetric density was calculated using correction methods described by DeFelipe et al., 1999, Erisir and Harris, 2003 and Wang et al., 2012. Volumetric densities were calculated by dividing the mean area density (number of synapses divided by the total area of each image) by the mean synapse length. Komolgorov-Smirnov tests confirmed that synapse density values were normally distributed; therefore, parametric statistics were used to assess differences between ages and regions. Averages were calculated for each animal in each layer and one-way ANOVA with Bonferroni post-hoc tests used to determine group differences. In addition, both synapse length and terminal bouton area were measured (Fig. 4a–c). Terminal areas were measured if the full bouton was inside the image view. Non-parametric statistics (Kruskal-Wallis with Dunn’s multiple comparisons post-hoc tests) were used to compare changes in these measures across layers and ages as their distributions were not normally distributed.

Fig. 4.

a. Asymmetric synapses were identified by the presence of presynaptic vesicles (asterisk), a synaptic cleft, and a postsynaptic density (arrow). b, c. Methods used to define presynaptic terminal area (dotted line) and synaptic length (red lines). d. Synaptic density. Bars represent average synaptic density across three animals. Average density from each animal is shown as a red dot. In layers 1a and 1b, density did not significantly change across the period examined. In both layers 2a and b, density increased through P17 before decreasing at P30. * = p<0.05 e. Cumulative frequency histogram for synaptic length at P5 (red), P13 (green), P17 (blue), and P30 (black). Layer 1 exhibited no differences across ages. In layer 2a, synapse length measurements were significantly different at P30 compared to younger ages. f. Cumulative frequency histogram for terminal bouton area. Similar to synaptic length measures, no differences were found between any ages in layer 1a or 1b, but a significant increase in terminal bouton area occurred between P5 and P17 in layer 2a.

3). Development of AONpP’s interhemispheric projections.

a). Visualizing the contralateral pathway.

An anterograde tracer (approximately 1μl of adeno-associated virus AAV9.hSyn.TurboRFP.WPRE.rBG, 2x10⁴ particles/μl in PBS: Penn Vector Core) was pressure-injected into the right pPl of three P1, 5, 20, and 40 mice (Fig.5a,b). Animals were monitored during recovery and daily after surgery. Animals were then perfused as above at P5, 12, 30, and 50, respectively, allowing ample time for the tracer to spread to the contralateral hemisphere. The ages were chosen to correspond with those used in the previous study. Brains were post-fixed for at least 12 hrs. Sixty-µm vibratome sections were mounted on slides with SlowFade mounting media (Invitrogen: S36937). The left AON was imaged at 20x and 40x magnification with a confocal microscope. Montages were made with Photoshop to visualize the distribution of labeled fibers through the AON.

Fig. 5.

a. Injections of AAV were made at the junction between pPl and pPd, producing strong somatic labeling at the injection site (b). c. As early as P5, several labeled fibers can be seen coursing across the midline in the anterior commissure (AC; dotted lines represent borders), though very few labeled fibers have reached the contralateral AON. d. By P12, axons within the contralateral ALAC are strongly labeled and a band of labeled fibers can be seen primarily in pPl, with occasional axons extending into pPd (arrow). Some retrograde labeling occurred with AAV injections, evidenced by labeled cell bodies (arrowhead). e. By P30 labeled axons are distributed throughout layer 1b/2a of both pPl and pPd. These bands thicken between P30 and P50 (f, arrowhead). Labeling is mostly restricted to pPl and pPd, with little labeling in AONpPvp and sparse labeling in pPm (asterisk). Scale bars = 100μm. g. BDA injections into the right AONpP produced axonal labeling restricted to the border between layer 1b and 2a (arrowheads). Scale bar = 100 μm. Labeled terminals were visualized using electron microscopy. Axon terminals (asterisk) were found making contact with both dendritic shafts (h) and spines (i). Scale bars = 1 μm. j. Synaptic length did not differ between labeled terminals and population data. k. Labeled terminals were larger than the general population.

b). Characterizing the pathway.

As little detailed information is available about the projection, the location and morphology of axon terminals or the distribution of synapses in AONpP, these features were examined. Approximately 1μl of 10,000 molecular weight BDA (Invitrogen: D1956) in PBS was pressure-injected into the right AON in three mice at P30. Animals were perfused 7–10 days later with Tyrode’s solution followed by a 2% paraformaldehyde, 1% glutaraldehyde solution. Brains were dissected and postfixed for at least 2 hrs. Sixty-µm vibratome sections were collected and rinsed in 0.1M PBS 4 times for 5 minutes. Tissue was then placed in 1% sodium borohydride in PBS for 30 minutes and then incubated in ABC solution (Vector Elite) for 1–2 hrs while gently agitating on a shaker. Tissue was then placed in a 0.05% DAB, 3% hydrogen peroxide solution for 15–20 minutes, rinsed in PBS four more times and post-fixed with osmium tetroxide for one hr. Subsequently the tissue was sequentially dehydrated, counterstained with 4% uranyl acetate in 70% ethanol overnight, embedded in EPON resin, and then sectioned at the interface between the tissue and EPON with an ultramicrotome. These procedures allowed for the best visualization of DAB-labeled structures. Imaging and Analysis. EM images were collected as described above. Images were only taken in regions containing at least one labeled synapse. For all BDA-labeled pre-synaptic terminals, synapse length and terminal area were measured (Holtz et al., 2015). Across three mice, 20 synapses and 14 terminals were measured. Measures from labeled synapses were compared to population data using a Mann-Whitney nonparametric test.

4). Developmental changes in perineuronal nets.

The immunostaining procedures, confocal imaging techniques and Image J analysis methods used in the first group of studies were used to examine age-related changes in the expression of two perineuronal net markers (aggrecan and brevican, Table 1) in strips of pPl extending from the LOT to the SVZ. These antibodies are specific to extracellular matrix components in the mouse central nervous system (aggrecan: Yamada and Jinno, 2016; brevican: manufacturer’s product information) and labeling patterns in our samples were consistent with known distributions of aggrecan and brevican.

5). Vascular development.

Once again, methods similar to those described in the first group of studies were employed. Confocal images were taken at 10X from sections immunostained with an antibody recognizing CD-31 (Table 1). Montages were produced and the area fraction of stained pixels analyzed with ImageJ in two zones: the white matter/SVZ core and the remaining portion of the olfactory peduncle (i.e., the LOT and layers 1 and 2 of AONpP). Care was taken at avoid measurements at the pial surface: while it contains many blood vessels it is difficult to accurately measure due both to the ease of tissue damage during brain removal and the folding of the region when sections are mounted.

The specificity of each of the antibodies employed has been demonstrated previously by their respective manufacturers: Aggrecan (Proteintech, Cat. # 13880–1-AP) by western blots and staining patterns in the extracellular matrix of cartilaginous tissues. Brevican (NeuroMab, Cat. # 75–294) by western blots and immunohistochemistry in knockout mice and by observing expected staining patterns in overexpressing cells: CD-31 (Dianova, Cat. # DIA-310) by western blots from endothelioma lysates: VGAT (SYSY, Cat. # 131 011) and VGlut1 (SYSY, Cat. # 135 302) by the absence of staining in knock out mice, VGlut2 (Millipore, Cat. # AB2251) by western blots of rat brain membrane extract. In all of our experiments the staining patterns observed were highly consistent with those reported widely in the literature.

Results

1). Synaptic development

Changes in the relative density of glutamatergic (VGlut1 and VGlut2) and GABAergic (VGAT) synapses were quantified in the upper, middle and lower thirds of both layer 1 and 2 (Fig. 2, 3). The procedures outlined in the methods section were employed to normalize measures between different sections, ages and animals. The average area fraction (ratio of stained/unstained pixels) of the entire section was designated as 100%. Staining densities in the outer, middle, and inner thirds of both layer 1 and 2 were then expressed with respect these averages resulting in values above or below “100%” corresponding values greater or lower staining than the overall section mean. These laminar subdivisions were made in an attempt to focus measurements on regions with known projections. For example, the outermost third of layer 1 is densely populated by synaptic connections between LOT axon collaterals and the distal dendrites of layer 2 pyramidal cells (Scott, 1987; Scott et al., 1985; Ghosh et al, 2011). The innermost third of layer 1 contains association fibers from APC, PPC and other regions of AONpP (Luskin and Price, 1983a,b; Kay et al., 2008). The superficial portion of layer 2 is thought to have synaptic connections similar to that of deep layer 1, while the regions beneath it contains association axons from the dorsal peduncular cortex and tenia tecta (Luskin and Price, 1983a). It should be noted that no clear deep boundary exists for layer 2 (Fig. 1b,c), so measures in this area include portions of the deep white matter of the olfactory peduncle.

Fig. 3.

a1-a4. Higher power details of VGLUT 1, 2 and VGAT staining from a P15 subject. VGAT synapses often encircle somata (e.g., arrows), while staining for the other two antigens was more diffuse. Scale bar = 100 μm. b. Age-related changes in the staining density in the deep, middle and superficial portions of layer 1 and layer 2. These measures were normalized for each antigen in each section by considering the average area fraction of staining across the entire test area as “100%” (dotted line). Therefore, areas with densities above or below 100% were more or less dense than the section mean. Patterns were quite similar across ages: for example, in layer 1 VGlut2 and VGAT was densest just below the LOT, while VGlut1 staining was more uniform.

Composite measures taken across all deep-to-superficial regions remained relatively constant across age, with immunostaining for VGlut1 representing about half of total staining, VGAT about 37% and VGlut2 about 12%. In layer 2 the amount of staining was directly related to the amount of neuropil: at young ages little space was observed between somata, but by P30 they were widely separated, doubtlessly due to the elaboration of axons and dendrites (Fig. 2). Distinct patterns of staining for each antigen were observed. VGlut1 staining was highest in layer 1 at all test ages (Fig. 2, 3b). At the earliest ages (P5 and P10) differences between layers 1 and 2 were striking, but by P15 staining became much more uniform throughout the tissue. In contrast, patterns of VGlut2 staining were similar across the test ages. The most intense staining was found in a band just deep to the LOT, while layer 2 exhibited patchy and variable staining, with highest labeling found in superficial regions. Dense VGAT staining was also present just beneath of the LOT, perhaps due to the sparse populations of resident GABAergic neurons (Kay and Brunjes, 2014). Overall, increases in the density of VGAT terminals were observed from P5-P30. Fig. 3a1–a4 displays details of the staining patterns in superficial layer 1 and layer 2 from a P15 subject. In layer 1 VGlut1 staining is relatively uniform, while both VGlut2 and VGAT densities are highest just under the LOT (Fig. 3a3–a4). In layer 2 VGAT synapses are often found perisomatically, while staining for the other two antigens is more diffuse.

2). Developmental changes in synaptic density, synaptic length, and terminal bouton area.

Electron microscopy was used to compare synaptic profiles in the superficial and deep halves of layers 1 and 2 in P5, 13, 17, and 30 mice. At least eight 10,000x images were collected from each of three subjects/age yielding a minimum analyzed area of 177μm² per layer/animal.

Synaptic density was calculated in order to determine whether the expansion of neuropil and the addition of new synapses progress at similar rates and to compare regional synaptic development in AONpP. The density of synapses remained constant across ages in the both layer 1a and 1b, indicating that synapses are added proportionally as neuropil space expands (Fig. 4d: F=1.15, n.s.). Conversely, synaptic density in layers 2a and b increased through P17 (2a: F=4.57, p<0.05; 2b: F=6.42, p<0.05), indicating the rapid addition of new synapses during this period. Density slightly declined from P17 to P30. Bonferroni’s post-hoc tests confirmed that the developmental change observed can be accounted for by the large increases between P5 and P17 (2a: t=3.62, p<0.05; 2b: t=4.09, p<0.05).

A similar pattern was observed when calculating synapse length (a measure of the diameter of the synapse active zone, Fig. 4b–c). Little change in mean synaptic length occurred in either layer 1a or b with age (1a: H=0.53, n.s.; 1b: H=3.323, n.s.; Fig. 4e). Conversely, in both layer 2a and b age-related changes were observed with large increases between P13 and P30 (2a: H=25.2, p<0.0001; 2b: H=16.94, p<0.001). Once again, by P30 no differences were observed between layers (P30; H=5.04, n.s.).

Terminal bouton size affords both a measure of synaptic development and a way to catalog subpopulations of axons within a given region. For example, as the visual cortex matures developmental terminal area increases (Erisir and Dreusicke, 2005) and within the visual thalamus, inputs from different areas (i.e. cortex and brainstem) can be assessed by their caliber (Erisir et al., 1997). At P30, all layers of the AON exhibited statistically similar terminal area distributions (H=2.58, n.s.). Once again, in layer 1a and b no changes in terminal area were observed from P5-P30 (1a: H=2.20, n.s.; 1b: H=2.03, n.s.). However, developmental differences were observed in layer 2a (H=20.67, p<0.0001) with the largest increases in terminal area observed between P17 and 30 (difference in rank sum = -46.02, p<0.0001), but not in layer 2b (H-0.87, n.s., Fig. 4f). Interestingly, distribution of terminal areas in both layer 1 and 2 at P5 resembled that of P30 (difference in rank sum = 17.27, n.s.).

3). Visualizing contralateral projections

Confocal microscopy was used to localize interhemispheric axon terminals originating from pPl (Fig. 5a–b). AAV injections into the right pPl led to robust labeling in the contralateral pPl and pPd. Lighter staining was encountered in pPvp, and little to no label was observed in pPm (Fig. 5c–f), patterns similar to those reported previously in adult rats and mice (Haberly and Price, 1978a; Illig and Eudy, 2009). Fibers coursed from the deep ALAC and distributed through layer 2 of the lateral and dorsal AONpP with very few fibers continuing all the way into superficial layer 1a.

Developmental studies indicated that by P5 many labeled fibers extend across the midline in the AC (Fig. 5c), although very few labeled axons were present in the contralateral AON (Schwob and Price, 1984; Sturrock, 1975). The density of labeled axons terminating in the AON sharply increased by P12 and continued to increase until P30 (Fig. 5d–e). During this time the band of labeled fibers present in layer 1b/2a became substantially thicker. Labeling at P50 was stronger than that observed at P30, suggesting that contralateral connections continue to be added well after development of the rest of the olfactory system has plateaued (Fig. 5f).

In order to more fully characterize the projection, animals were examined after BDA administration at P30. BDA was detected in a very small number of contralateral terminals (n=14) and contralateral presynaptic synapses (n=20). Anterogradely-filled axon terminals were found deep in layer 1b and labeled axons could be seen coursing through layer 2 (Fig. 5g). Terminals were seen contacting both dendritic shafts and spines (Fig. 5h–i). Terminal and synapse sizes were normally distributed. Synapse length did not significantly differ between labeled and unlabeled synapses (U=1362, n.s.; Fig. 5j) but labeled terminal boutons were significantly larger than unlabeled terminals in the same region (unlabeled M = 347.85 μm², SD = 224.08; labeled M = 531.43 μm², SD = 338.31; U 3= 342, p<0.01; Fig. 5k). Despite the small sample size, such a large difference in terminal size could suggest that synapses originating from contralateral sources form a distinct subpopulation.

4). Developmental changes in perineuronal nets.

Tissue from P5-P30 pups was stained for aggrecan and brevican, both of which are major components of the matrix covering neurons (Fig 6 a–c; Zimmerman and Dour Zimmerman, 2008; Bikbaev et al, 2015; Sorg et al., 2016). Tissue from P5 subjects exhibited very diffuse aggrecan staining, primarily in layer 1 (Fig 6 a1). Image J analysis indicated that the area fraction of labeled pixels was 7 times denser than in layer 2 (means = 35 vs 5; U=0, P<0.05). This pattern remained (though not as exaggerated) through all ages examined: staining in Layer 1 was at least 3 times denser than Layer 2 (Fig 6 a1–4). By P10 staining began to localize over somata. The pattern became very apparent by P15 and P20 (Fig. 6b1–b3). Both interneurons and pyramidal cells exhibited somatic labeling (Fig. 6c). Brevican staining was minimal at P5 with area fractions of less than 5. By P10 labeling began to accumulate on apical dendrites (Fig. 6a1–a4) and by P20 staining density had increased to 10X that seen at P5. In layer 2 labeled dendrites could be seen projecting radially towards the pial surface. The apical dendrites of cells scattered throughout layer 2 accumulate as they ascend, resulting in denser staining in the superficial regions of the layer. In layer 1 staining was much less organized, presumably because dendritic processes assume varying orientations in this neuropil region (Fig. 6b1–3).

Fig. 6.

a-c The development of perineuronal nets was observed with immunostaining for aggrecan (green) and brevican (red). The images in a1-a4 depict age-related changes in these markers in sections oriented similarly to those see in Fig. 1c and Fig 2. Higher magnification images of layer 2 from a P20 subjects can be seen (b1- b3). Scale bar = 100 μm. Aggrecan labeling was diffuse in P5 subjects and seen primarily in layer 1 (a1). By P15 the densest staining was observed over somata (a3, b2), which was intensified in P20–30 day old subjects (b3). At P5, brevican staining was also diffuse and found deep within the peduncle (a1). By P10 brevican staining was found to be increasing in layer 2 and by P15 outlined the apical dendrites of pyramidal cells (a2, b1). At the oldest ages examined the ascending primary dendrites were obvious in layer 2; as branches entered layer 1 and assumed many orientations the stained figures became more complex (a4, b3). Panel c depicts tissue from a GAD67-GFP mouse line in which all GABAergic neurons express green fluorescent protein. Scale bar = 50 μm.Note that aggrecan labels these cells bodies as well as most of the other somas in the image, suggesting that all neurons in the region are involved. D. Developmental changes in capillary density. Immunostaining for CD-31, a marker of mature endothelial cells revealed no differences in total blood vessel density from P5-P30. The ALAC, found in the deep core of the peduncle, exhibited less density than the overlying layers 1 and 2. By P30, vessels in layer 1 were about twice as dense as of those in layer 2. Scale bar = 220μm.

5). Vascular development.

As neural maturation necessitates an abundant supply of resources, a small study examined if there were developmental changes in the vascularity of the olfactory peduncle (Fig 6d). Image J analysis of immunostaining for CD-31, a marker of mature endothelial cells (Vanzulli et al., 1997; Milner, 2014), revealed no difference in total blood vessel density from P5-P30 (H=1.067; p=0.785). The core of the peduncle, containing the axons running with the anterior limb of the anterior commissure and SVZ, exhibited less density than the overlying layers 1 and 2 (t=19.754, p<0.00001). By P30, vessels in layer 1 were about twice the density of those in layer 2.

Discussion

The present work examines several features of the development of AONpP, a large but relatively understudied component of the olfactory circuit even though it is the first recipient of odor information processed by the olfactory bulb. The AONpP shares many characteristics of the other forebrain cortices, including pyramidal-shaped projection neurons (Brunjes and Kenerson, 2010), a wealth of interneuron types (Kay and Brunjes, 2014) and similar sequences of cell production during gestation (Brunjes and Osterberg, 2015). It is an extremely simple cortex, with an outer molecular layer and a single cellular region, with inputs to the structure segregating in specific regions (Fig. 1).

The first set of studies provided an overview of synaptogenesis in pPl by examining age-related changes in markers of molecules involved in loading neurotransmitters into presynaptic glutamatergic (VGlut1 and VGlut2) and GABAergic (VGAT, also known VIAAT; Wojcik et al., 2006) vesicles. The two glutamatergic markers have been used to identify separate axon systems in other tissues: VGlut1 is associated with intracortical and corticostriatal pathways while VGlut2 markers label thalamocortical and thalamostriatal inputs (Fremeau et al., 2004; Kaneko et al., 2002; Raju et al., 2006). Differential distributions were found in pPl. VGlut1 was broadly distributed throughout the tissue but was densest in layer 1 (Fig. 2–3). A band of VGlut2 immunostaining was found just deep to the LOT (layer 1a), with only sparsely distributed figures throughout layer 2. GABAergic synapses (VGAT) were also prevalent throughout the tissue. As observed for VGlut2, a dense band of VGAT staining was observed just under the LOT, in a region known to contain GAD67 positive neurons (Brunjes et al., 2011; Kay and Brunjes, 2014). Furthermore, like VGlut1, both layer 1 and 2 exhibited dense staining, with layer 1 was most heavily stained. The relative levels of the three synaptic markers were constant across age, with VGlut1 representing about half of total staining, VGAT about 37% and VGlut2 about 12%. However, age-related changes were observed. In P5 pups most staining was observed in layer 1; a finding consistent with observations that the LOT forms very early during embryonic development (Collins et al., 2018) and that olfactory function is present at birth (Blass and Teicher, 1980; Logan et al., 2012). The superficial portion of layer 1 continued to be densely stained with all three markers through P30, doubtlessly because the region contains the large number of synapses between LOT fibers and processes of the resident neurons in the peduncle. At P5, layer 2 contains a narrow band of the densely packed cell bodies of relatively undifferentiated neurons. With age, dendrites and axons expand, and the neuropil regions between the somata increase. This growth results in an increased depth of layer 2 but no change in the relative density of the three markers. The observation is consistent with the idea that synapses fill all available space, and also with data indicating that synaptic density is generally constant across species (Cragg, 1967; Beaulieu and Colonnier, 1985; Schüz and Palm, 1989; Schüz and Demianenko, 1995). While VGAT staining was likely due to interneurons in the region, there are multiple sources for glutamatergic synapses including a) axon collaterals from LOT fibers that distribute olfactory input to to layer 1a (Scott, 1987; Scott et al., 1985; Ghosh et al., 2011), b) the resident neurons in the AONpP (Illig and Eudy, 2009), c) association fibers from the APC, DPC and tenia tecta on the ipsilateral side, d) axons from the contralateral AONpP arriving via the anterior commissure, and e) connections from a number of other neural regions innervating AONpP including the hippocampal formation, amygdala, and hypothalamus (Brunjes et al., 2005; Luskin and Price, 1983a,b; Friedman and Price, 1984). The exact contributions of these many of these inputs has yet to be studied.

Electron microscopic studies confirmed these findings. For instance, consistent with the immunostaining results detailed above, synaptic density was higher in layer 1 in early life but, due to developmental changes in layer 2, by P30 both regions were similar. Additionally, synaptic maturation in the plexiform layer of the AON occurred earlier than the cellular layer. Changes in synapse density or morphology (length and terminal bouton size) are indicative of synapse formation or circuit refinement. Interestingly, within the plexiform layer of the AON, these parameters did not change over the developmental period examined, indicating that synapse maturity is achieved earlier than P5. Such early maturation within layer 1 would allow the necessary neural activity for other developmental processes, such as early myelination (Collins et al., 2018), and is consistent with similar findings from the rat piriform cortex demonstrating early synaptic development (Westrum, 1975). Finally, EM analysis also confirmed extensive neuropil expansion throughout the period examined.

Differences in the timing of maturation between the LOT and AC are mirrored in functional differences within the olfactory pathway (Collins et al., 2018). As mentioned above, the LOT arises early in order to allow olfactory function at birth. It undergoes a period of rapid myelination between P7 and 11. The AC forms 3 days later than the LOT, is quite immature at birth and does not undergo its period of rapid myelination until P13 (Collins et al., 2018). Interestingly, behavioral studies also suggest that the interhemispheric transfer of information does not begin until about P12 (Kucharski and Hall, 1990). Since very little work has examined the commissural pathway, it was of considerable interest to characterizing its topography, synaptic morphology and development. Anterograde tracing revealed that contralateral projections terminate in a thin band deep within layer 1. Furthermore, EM studies revealed that these terminals synapse onto dendrites, presumably apical dendrites extending from the pyramidal cells in layer 2, and form large terminal boutons. The tight regulation of terminal location at the proximal end of these apical dendrites and large terminal size suggests an important role for contralateral projections in driving activity of the AON.

Perineuronal nets around neurons regulate neuritic growth and synaptogenesis during early life. Maturation of these extracellular matrices restricts synaptic plasticity and stabilizes functional synapses (Clarris et al., 2000; Valenzuela et al., 2014; Bikbaev et al., 2015; Sorg et al., 2016; van ‘t Spijker et al., 2017; Lensjø et al., 2017; Quraishe et al., 2018) and disrupting them can extend early sensitive periods (e.g. monocular deprivation in the cerebral cortex or thalamus; Lensjø et al., 2017) and allow synaptic potentiation of normally plasticity-resistant neurons (Carstens et al., 2016). Substantial regional differences in the expression of extracellular matrix proteins during development have been reported (Dauth et al., 2016, Sontag et al., 2015). Aggrecan and brevican, both chondroitin sulfate proteoglycans, are prominent among these molecules. Data included in Fig.6 reveal that the first indication of their mature patterns is evident by P15 in AONpP. By P20 aggrecan staining is found primarily around the cell bodies of both excitatory and inhibitory neurons, while brevican lines neuronal processes.

Finally, as development necessitates considerable energy, we examined age-related changes in the vascularity of the olfactory peduncle using an antibody recognizing mature endothelial cells (Vanzulli et al., 1997). Developing blood vessels sprout from the pial vasculature and grow gradually deeper into the brain towards the SVZ (Wittko-Schneider et al., 2014). No developmental differences were observed in the density of blood vessels across age. However, it was found that vessel area was lowest in the central zone containing the axons in the ALAC and the SVZ and highest in layer 1.

The results outlined above are the first systematic examinations of synaptic development in any structure located in the olfactory peduncle. While the function of the areas comprising the peduncle are poorly understood there are indications that it is involved in localizing stimuli in space by comparing information from the left and right olfactory bulbs (Kikuta, 2010) and for temporal and spatial aspects of odor learning (Aqrabawi and Kim, 2018 a,b; Kucharski et al., 1990). The AONpP is positioned in the olfactory circuit to have broad control over most of the olfactory forebrain: it has strong connections with both the ipsilateral and contralateral OB and APC, and the contralateral AON (Brunjes et al., 2005). Further studies are needed to refine our knowledge of the structure and function of the olfactory peduncle.