Genetic Increases in Olfactory Bulb BDNF Do Not Enhance Survival of Adult-Born Granule Cells

Abstract

Adult-born neurons produced in the dentate gyrus subgranular zone (SGZ) develop as excitatory hippocampal granule cells (GCs), while those from the subventricular zone (SVZ) migrate to the olfactory bulb (OB), where most develop as GABAergic olfactory GCs. Both types of neurons express TrkB as they mature. Normally ~50% of new olfactory GCs survive, but survival declines if sensory drive is reduced. Increases in endogenous brain-derived neurotrophic factor (BDNF) in hippocampus, particularly with wheel running, enhance dentate GC survival. Whether survival of new olfactory GCs is impacted by augmenting BDNF in the OB, where they mature and integrate, is not known. Here, we determined if increasing OB BDNF expression enhances survival of new GCs, and if it counters their loss under conditions of reduced sensory activity. Neurogenesis was assessed under normal conditions, and following unilateral naris occlusion, in mice overexpressing BDNF in the granule cell layer (GCL). OB BDNF levels were significantly higher in transgenic mice compared to controls, and this was maintained following sensory deprivation. Bromodeoxyuridine (BrdU) cell birth dating showed that at 12–14 days post-BrdU, numbers of new GCs did not differ between genotypes, indicating normal recruitment to the OB. At later intervals, transgenic and control mice showed levels of GC loss in deprived and nondeprived animals that were indistinguishable, as was the incidence of apoptotic cells in the GCL. These results demonstrate that, in contrast to new dentate GCs, elevations in endogenous BDNF do not enhance survival of adult-born olfactory GCs.

Introduction

A variety of molecular signals regulate adult neurogenesis in the 2 neurogenic niches of the adult mammalian forebrain, the subventricular zone (SVZ) lining the lateral ventricles, and the subgranular zone (SGZ) of the hippocampal dentate gyrus (Ming and Song 2005). Resident stem cells in these regions give rise to distinct types of new neurons in adulthood. The adult SGZ generates glutamatergic dentate granule cells, while neuroblasts born in the SVZ migrate in the rostral migratory stream to the olfactory bulb (OB), where they develop as inhibitory interneurons (Whitman and Greer 2009). The majority of the latter differentiate as gamma-aminobutyric acid (GABA)-synthesizing granule cells (GCs), anaxonic neurons that inhibit excitatory bulb output neurons, the mitral and tufted cells, through dendrodendritic synapses (Shepherd 1972; Carleton et al. 2003). The integration of new GCs is an adaptive turnover process that replaces pre-existing GCs that are continuously eliminated by programmed cell death (Imayoshi et al. 2008; Mouret et al. 2009). Functionally, adult-born GCs exhibit properties that impact odor processing (Breton-Provencher and Saghatelyan 2012). In comparison to mature GCs, they show unique physiological responses to sensory stimulation and long-term potentiation induction, and their continued integration endows OB circuitry with unique functional properties that mediate types of associative odor learning (Lazarini and Lledo 2011).

Factors provided by cells in the SVZ, as well as those from the vasculature, choroid plexus, and cerebrospinal fluid, control neurogenesis by regulating the proliferative activity of the neural stem cells and the intermediate progenitors they generate. Growth factors, including epidermal growth factor (EGF), fibroblast growth factor 2 (FGF2), ciliary neurotrophic factor (CNTF), and the neurotrophin NT3, play critical roles in this regulation (Kuhn et al. 1997; Lee et al. 2013; Delgado et al. 2014). Studies also have implicated the neurotrophin brain-derived neurotrophic factor (BDNF), though with conflicting results (Zigova et al. 1998; Benraiss et al. 2001; Bath et al. 2008; Galvao et al. 2008). BDNF binds the tropomyosin-related receptor tyrosine kinase type B (TrkB), the truncated TrkB receptor, and the low-affinity neurotrophin receptor p75 (Reichardt 2006). Each receptor type is expressed by distinct cell types in the rodent SVZ-derived lineage (Gascon et al. 2005; Chiaramello et al. 2007; Galvao et al. 2008). In vitro, BDNF increases neuron production when applied to SVZ explants or p75+ isolated progenitor cells (Kirschenbaum and Goldman 1995; Young et al. 2007). In vivo manipulations of BDNF in the ventricular zone/SVZ, or throughout the brain, have produced mixed effects. Germline BDNF knockout mice that survive to 2–4 weeks of age have smaller OBs that contain more apoptotic cells, while adult mice haploinsufficient for BDNF produce fewer new GCs (Linnarsson et al. 2000; Berghuis et al. 2006; Bath et al. 2008). In contrast, ventricular infusions of BDNF, or BDNF-encoding adeno-associated viruses (AAV), increase numbers of new GCs recruited to the bulb, though this effect has not been found consistently, potentially due to differences in the methods, location, and duration of BDNF administration (Zigova et al. 1998; Benraiss et al. 2001, 2012; Henry et al. 2007; Galvao et al. 2008).

New GCs express full-length TrkB as they reach the OB and begin to mature, therefore BDNF may act in this location to regulate later stages of their development (Bergami et al. 2013). The rodent OB normally expresses low levels of BDNF, while higher olfactory regions that project to the OB GCL, including the piriform cortex, express high levels (Guthrie and Gall 1991; Conner et al. 1997; Clevenger et al. 2008; Matsutani and Yamamoto 2008; McDole et al. 2015). Between 2 and 4 weeks after their birth, 25–30% of new GCs are eliminated by apoptosis (Petreanu and Alvarez-Buylla 2002; Yamaguchi and Mori 2005). Sensory activity regulates this process, and odor enrichment, as well as some types of olfactory learning, increase GC survival, suggesting that new cells are selectively recruited into bulbar circuits (Lazarini and Lledo 2011). Conversely, sensory deprivation increases GC loss (Cummings et al. 1997; Petreanu and Alvarez-Buylla 2002; Yamaguchi and Mori 2005). New cells are vulnerable to deprivation during a critical phase of maturation and synaptogenesis, and it has been proposed that reduced activity during this time impairs their integration, thereby promoting cell death (Petreanu and Alvarez-Buylla 2002; Yamaguchi and Mori 2005; Kelsch et al. 2009). Dendritic spine and synapse density are indeed reduced in those GCs that survive deprivation for 2–5 weeks, supporting this view (Kelsch et al. 2009; Dahlen et al. 2011).

Olfactory deprivation also reduces OB BDNF levels, while sensory stimulation increases this, consistent with the activity-dependent nature of normal BDNF gene regulation (Shieh and Ghosh 1999; McLean et al. 2001; Jones et al. 2007; Degano et al. 2014). BDNF is widely expressed by excitatory forebrain neurons, with normally very little detected in GABAergic populations (Hofer et al. 1990; Schmidt-Kastner et al. 1996). However, seizure activity reportedly leads to modest increases in BDNF mRNA in the OB GCL, while alcohol administration upregulates BDNF mRNA in dorsolateral striatum (Isackson et al. 1991; McGough et al. 2004). BDNF promotes growth and maturation of dendrites and spines, and we have shown that overexpression of BDNF by morphologically mature olfactory GCs increases their dendritic spine density (Gottmann et al. 2009; Vigers et al. 2012; McDole et al. 2015; Guo et al. 2018). Whether this affects survival of adult-born GCs by aiding their structural integration is not known, but such effects have been found in adult-born dentate GCs, which also show activity-dependent maturation (Bergami et al. 2008; Aasebø et al. 2011; Wang et al. 2015). Increases in hippocampal BDNF promote structural development and survival of new dentate GCs, while reductions decrease dendritic complexity, spine maintenance, and survival rate (Scharfman et al. 2005; Rossi et al. 2006; Chan et al. 2008; Choi et al. 2009; Gao et al. 2009).

To determine if increases in BDNF can similarly enhance survival of new olfactory GCs, we examined neurogenesis in transgenic mice that overexpress BDNF constitutively under the alpha calcium/calmodulin-dependent protein kinase II (CAMKIIα) promoter. Because olfactory deprivation can reduce OB BDNF levels in normal mice, we also tested if its increase could partially rescue new GCs from reduced sensory drive (McLean et al. 2001). BDNF overexpression in developing visual cortex compensates for deprivation-induced declines in native BDNF, and prevents defects in circuit maturation that occur with dark-rearing (Gianfranceschi et al. 2003). We report that, in contrast to its effects in the SVZ and hippocampus, augmenting BDNF in the OB has no effect on net olfactory neurogenesis. Bromodeoxyurdine (BrdU) cell labeling and assessments of apoptosis yielded no evidence for increased addition or survival of new cells with BDNF overexpression, either in the normal or sensory-deprived OB. Under the latter condition, GCL volume decreased to the same extent as that seen in control mice. Our results indicate that as new olfactory GCs mature, increases in endogenous BDNF do not significantly enhance their survival.

Methods

Animals and ethics

Animal procedures were carried out in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals under protocols approved by the Florida Atlantic University Animal Care and Use Committee. The mice, hereafter termed TgBDNF mice, were maintained on a C57Bl6/J background (Jackson Laboratories strain #006579). In addition to the native BDNF gene, the mice carry a transgene that encodes rat preproBDNF downstream of 8.5 kb of the CAMKIIα promoter, which replaces the native promoter region containing response elements that regulate activity-dependent BDNF expression (Huang et al. 1999; Shieh and Ghosh 1999). Brain expression of the transgene results in a significant increase in levels of a 1.6 kb BDNF transcript. Native CAMKIIα mRNA expression in developing mouse brain is detected in the first postnatal week, and increased CAMKIIα-driven BDNF transgene expression is detected as early as postnatal day 2 (Bayer et al. 1999; Huang et al. 1999). In adult brain, the transgene is widely expressed by forebrain excitatory neurons, but is also expressed by CAMKIIα-expressing GABAergic olfactory GCs and striatal neurons, resulting in increased BDNF levels in the OB and striatum (Burgin et al. 1990; Zou et al. 2002). We previously used in situ hybridization/quantitative densitometry to localize and quantify a more than 3-fold increase in BDNF mRNA expression in the olfactory GCL of TgBDNF mice, as well accompanying increases in BDNF protein (McDole et al. 2015). In the striatum, expression of the BDNF transgene improves motor behaviors and normalizes medium spiny neuron spine density in TgBDNF mice crossed with Huntington’s Disease mouse models (Gharami et al. 2008; Xie et al. 2010). Unlike most excitatory CNS neurons, CAMKIIα is not expressed by OB mitral/tufted cells, which also lack transgene expression (Zou et al. 2002; Néant-Fery et al. 2012). However, the transgene is highly expressed by excitatory neurons that project to the OB from the olfactory cortices (McDole et al. 2015). TgBDNF males were bred with WT females to obtain litters containing both TgBDNF and WT offspring. To verify the presence of the transgene, genomic DNA was isolated from tail samples for PCR amplification using the following primers: 5′-GTGAAGGAACCTTACTTCTGTGGTG-3′ AND 5′-GTCCTTGGGGAAGCTGAGGA-3′. Cycling parameters were: 94 °C-30 s, 54 °C-30 s, 72 °C-45 s (32 cycles). The total number of mice used was 98 and included both males and females.

BrdU treatment, naris occlusion, and tissue collection

WT and TgBDNF mice at 7.5–8 weeks of age were treated once daily with 5-bromo-2′-deoxyuridine (BrdU: i.p., Roche Life Sciences, #10-280-879) at a dose of 50 mg/kg for 4 consecutive days. Mice survived for 2, 4, or 9 weeks after the first BrdU injection (n = 4, 2 males and 2 females per genotype and time point). Animals were euthanized with sodium pentobarbital (150 mg/kg in Euthasol, i.p.) and transcardially perfused with phosphate-buffered saline (pH 7.3), followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.35, 4 °C). Brains were dissected and postfixed overnight at 4 °C and cryoprotected in 30% sucrose. Forebrains were embedded in 10% gelatin, postfixed overnight, and returned to 30% sucrose (4 °C). Embedded brains were snap frozen in isopentane (−45 °C) and stored at −80 °C. Brains were sectioned in a cryostat (−21 °C; 30 µm) and serial, free-floating coronal sections were collected through the bulbs (1 in 5) and forebrain SVZ (1 in 3) for immunostaining and TUNEL labeling. A separate cohort of mice was treated with a single dose of BrdU (150 mg/kg; i.p.) at 7.5–8 weeks of age, and 24 h later underwent unilateral naris occlusion. Under ketamine/xylazine anesthesia (100 mg/kg ketamine; 10 mg/kg xylazine; i.p.), sterile eye lubricant (Puralube vet ointment, Dechra, Overland Park, KS) was applied to the right external naris, and a small plug made of PE50 tubing filled with hard-set dental molding wax (4–5 mm) was inserted into the naris. Control mice received anesthesia only. Mice were euthanized and brains collected as above at 12 days, 4 or 8 weeks after BrdU treatment (n = 4 deprived, 3 controls per genotype and time point). Animals, and slides of tissue sections from each, were assigned a number code for blinded analyses.

BDNF enzyme-linked immunosorbent assay (ELISA)

Mice were anesthetized and nares occluded as described above. At 4 weeks postocclusion, 3 control mice and 4 deprived mice of each genotype were overdosed with sodium pentobarbital (150 mg/kg; i.p.), decapitated, and individual bulbs were dissected and collected for ELISA measurement of total bulb BDNF. Replicate assays were performed with Promega’s BDNF EMax Immunoassay kit (Promega, cat# G7610) using duplicate samples of bulb supernatants (200 mg protein/well) and overnight incubation at 4 °C as described (McDole et al. 2015). The monoclonal and polyclonal BDNF antibodies in the kit were raised against recombinant human prepro-BDNF (119 amino acids) and each detect 50 ng of recombinant rat BDNF at a molecular mass of ~26kDa on Western blots, with no cross reactivity to 50 ng of NGF, NT3 or NT4 (manufacturer’s information). Following final reaction development, absorbance of samples and serially diluted BDNF standard controls were measured with a SpectraMax M5 plate reader and BDNF sample concentrations were calculated with SoftMax Pro software (Molecular Devices).

Antibodies

The rat monoclonal BrdU IgG antibody was made using the thymidine analog bromodeoxyuridine as antigen, and was obtained from Accurate Chemical (clone BU1/75, cat. #OBT0030, RRID antibody registry: AB_2313756). The antibody recognizes nuclear staining in olfactory neurons born from SVZ progenitors that incorporate BrdU in vivo (while in S phase) shortly after its administration (Magavi 2005). Mouse monoclonal NeuN antibody was obtained from Millipore-Sigma, and was made using purified mouse brain nuclei as antigen (cat#MAB377; RRID antibody registry: AB_2298772). It recognizes several bands corresponding to NeuN at ~47 kDa in immunoblots of mouse brain nuclear protein isolates, and stains nuclear NeuN in most brain neurons, but does not label olfactory mitral cells, which lack NeuN expression (Mullen et al. 1992). The rabbit antibody to Ki-67 was kindly provided by Drs. C. Isgor and S. Watson (University of Michigan, RRID antibody registry AB_2314695). The immunizing antigen was the synthetic peptide TPKEKAQALEDLAGFKELFQT, corresponding to amino acids 1213–1233 of human nuclear protein Ki-67 (Swiss-Prot ID: P46013.2). The antibody labels nuclei in proliferating cells of the adult mouse dentate gyrus SGZ (Isgor and Watson 2005). The rabbit polyclonal TH antibody (#AB152, EMD Sciences/Millipore; RRID antibody registry: AB_390294) detects a single band of ~59 kDa on Western blots of lysates prepared from the mouse nucleus of the solitary tract (NST), and stains noradrenergic NST neurons in fixed tissue sections. Immunostaining and western blots performed with this antibody show reduced levels of TH following viral-mediated knockdown of TH in NST neurons in vivo (Bathina et al. 2013).

Ki-67+ immunostaining and SVZ cell counts

Proliferating SVZ cells were identified by immunoperoxidase staining for the nuclear antigen protein Ki-67, using brains from WT and TgBDNF mice at 8–10 weeks of age (n = 5 per genotype). Sections from these mice also were used for olfactory bulb TUNEL staining (below). Sections rinsed for 30 min in 0.05% Triton X-100 in Tris buffered saline (TBS; 0.1 M Tris-HCl, 0.9% NaCl, pH 7.4). For epitope retrieval, sections were placed in 10% Na-Citrate buffer (pH 6.0) in a 91 °C water bath for 1 h, then cooled to RT for 20 min. Tissue was treated with TBS containing 0.6% H₂0₂ and 0.05% Triton X-100 for 30 min. Sections blocked in 3% normal goat serum, followed by incubation in primary antibody (1:7000) overnight at 4 °C. Sections then incubated in biotinylated goat anti-rabbit IgG (Vector Laboratories: 1:200) for 1 h, followed by avidin-biotin-HRP complex (Elite ABC kit, Vector Laboratories). Chromagen was developed with Vector Labs Impact DAB kit (#sk-4105), using 12 µL of chromagen per mL of reaction buffer. Omission of the primary antibody was used as a negative control. Counts of proliferating SVZ cells were made along the lateral ventricular wall in both hemispheres from 6 serial sections (90 µm intervals) beginning at ~1.25 mm anterior to bregma (Paxinos and Franklin 2004). Cell counts were made at 40–60× objective magnification using an Olympus AX70 microscope. Calibrated digital images of sampled SVZ regions were collected at 20× objective magnification, and the contours of the SVZ were outlined in NIH ImageJ (Rasband 2010) to measure the sampled area. Mean density of Ki-67+ cells in each subject was used to calculate group mean density values, ± standard error of the mean (SEM).

Immunofluorescence

Combined labeling for BrdU and NeuN was used to identify adult-born GCs. Serial sections incubated in 0.6% H₂O₂ for 8 min, followed by 50% formamide in 1× saline-sodium citrate buffer for 30 min at 65 °C. Sections were transferred to 2N HCl for 30 min at 37 °C, then 0.1 M sodium borate (pH 8.5) for 10 min. After blocking in 5% normal goat serum, sections incubated for 3 nights at 4 °C in rat antibody to BrdU (1:400) combined with mouse antibody to NeuN (1:500). Secondary antibody cocktail contained mouse-adsorbed AlexaFluor (AF)-488-conjugated goat anti-rat IgG (#A-11006) and AF594-labeled goat anti-mouse IgG (#A-11032, Life Technologies), each diluted 1:1000. Additional bulb sections from deprived mice and control littermates were processed for tyrosine hydroxylase (TH) immunofluorescence to confirm effective sensory deprivation (Baker et al. 1993). Sections blocked in 5% goat serum in PB with 0.3% triton X-100, followed by overnight incubation in rabbit anti-TH diluted in the same blocker (1:700). Secondary incubation used AF594-conjugated goat anti-rabbit IgG (1:1000). Immunostaining was examined in the glomerular layer, and images were collected using a Zeiss 710 Laser Scanning Confocal Microscope (LSM, Zeiss AG).

BrdU+/NeuN+ cell density in the granule cell layer

Confocal image collection from coded samples was performed with a Zeiss 710 LSM at 20× objective magnification (zoom factor 1.3; 1 airy unit = 4 µm). Collection started in the anterior bulb where the GCL measured ~1 mm from dorsal to ventral. A sample box (150 µm × 250 µm; 0.16 µm/pixel) was placed in each of 4 regions of the GCL (ventral, medial, dorsal, lateral) with one outer edge positioned superficially along the internal plexiform layer, and the remaining box extending down over the GCL as previously described in (Smail et al. 2016). Images of immunolabeling for each marker were collected for each sample region (BrdU only, NeuN only, and combined labeling), and were used for cell counts using NIH ImageJ software (Schneider et al. 2012) and the Cell Counter plugin (Schneider et al. 2012). A minimum of 6 serial sections (maximum 7, both bulbs) were imaged per subject, with a minimum of 48 sample regions per animal. A total of 0.15 mm² of the GCL was sampled per section, for a minimum total sample area of 1.8 mm² per subject. Dense labeling for NeuN occurred in cells throughout the GCL, but mitral cells lacked NeuN staining in all sections. Numbers of BrdU+/NeuN+ and BrdU+-only cells were counted in the GCL sample regions, with these values used to calculate mean density of new GCs, and the proportion of presumptive glia (BrdU+ only). Orthogonal projections of z-stacks (0.7 µm steps) were used to verify nuclear colocalization of BrdU and NeuN in random labeled cells in the GCL of each subject. Group mean values were calculated from mean cell counts and densities measured in each subject, ± the standard error of the mean (SEM).

BrdU-immunoperoxidase staining and cell counts

Localization of BrdU was performed on every 10th section through the olfactory bulbs, with additional control sections processed without the primary antibody. Tissue treatment included incubation in H₂O₂, formamide/SSC, 2N HCl and sodium borate as above. Sections blocked in 5% normal rabbit serum, followed by overnight incubation in rat anti-BrdU. Sections were treated with biotinylated rabbit anti-rat IgG (1:200; Vector Laboratories), followed by avidin-biotin-HRP complex for 1.5 h (Vector’s Elite ABC kit). Chromagen was developed with Vector’s Impact-DAB kit at a dilution of 1:85 (chromagen:buffer) for 2 min. Beginning with the first section in which the GCL was visible, all BrdU+ cells in the GCL were counted microscopically (40× objective magnification) in every tenth section of both bulbs, up to the first appearance of the GCL of the accessory olfactory bulb (AOB). This distance was smaller in deprived mice due to bulb shrinkage (quantified below). The percentage of presumptive BrdU+ glia in the GCL, based on counts of BrdU+-only cells (described above for double-labeling) was subtracted from this total to estimate the total number of adult-born GCs in the section series. Group means in deprived mice were calculated from cell counts made ipsilateral and contralateral to occlusion (n = 4 per genotype and time point).

Terminal deoxynucelotidyl transferase dUTP nick end labeling (TUNEL)

Slide-mounted serial sections were processed for TUNEL labeling of apoptotic cells as described (Smail et al. 2016). Treatment included incubation with 40 mg/mL (~1.5 units/mL) proteinase K for 20 min at 37 °C, followed by 2.5 h of incubation in TUNEL reaction solution at 37 °C (25 mM Tris-HCl, 200 mM Na cacodylate, 0.25 mg/mL BSA, 1 mM cobalt chloride, 13 mM biotin-14-dATP [Life Technologies, #19524-016], 200 U/mL terminal transferase (New England Biolabs, #M0315]). Following overnight incubation in avidin-biotin-HRP complex (1:100; Vector Laboratories), chromagen was developed with Vector’s Impact-DAB kit as above. Controls included pretreatment of some sections with DNAse (5 units/µL, Sigma-Aldrich), as well omission of terminal transferase. Numbers of TUNEL+ cells per bulb (left and right) were counted microscopically in the GCL (40× objective magnification) from 6 serial sections (150 µm spacing), starting with the most caudal section anterior to the appearance of the AOB GCL, and moving rostrally. Counts included labeled cells in the subpendymal layer. Cell counts were made in 5 nondeprived WT and 5 TgBDNF mice. For the deprivation experiment, counts were made from 4 deprived and 4 control mice per genotype at 4 weeks postocclusion. Group means were calculated from total cell counts obtained for left and right bulbs in individual mice, for each genotype and sensory condition (±SEM).

Estimations of granule cell layer volume

Alternate sections (1 in 5) from deprived and control mice at 8 weeks postocclusion were stained with neutral red, and calibrated images were collected at 4× objective magnification using an Olympus AX70 microscope with a Microfire digital camera. Mean GCL area in each section (left and right) was calculated by measuring the GCL area twice using NIH ImageJ software (Rasband 2010), with the mitral cell layer/internal plexiform layer used to define the outer GCL boundary. GCL volume for each bulb was estimated as in Mirich et al. (2002). The areas of the most anterior section containing visible GCL and the next section in the series were measured, and the volume of each calculated using section thickness. The mean volume of these 2 sections was calculated and then multiplied to estimate the volume of intervening sections. This was repeated for subsequent serial sections, ending just anterior to the appearance of the AOB GCL. Total GCL volume (including the subependymal layer) was estimated by summing these section volumes for left and right bulbs in each subject. Group mean values ± SEM were calculated from these (n = 4 deprived and 3 nondeprived mice per genotype).

Statistical analysis

Unless otherwise stated, all data are presented as group mean values ± SEM, with individual values indicated on graphs. Analysis was performed using Graphpad InStat and Prism 7 software, with level of significance set at P < 0.05. All within-subject comparisons were made by paired, 2-tailed t-tests (left vs. right). All other pair-wise group comparisons at a given time point were tested with unpaired, 2-tailed t-tests with Welch’s correction.

Results

Subventricular zone cell proliferation is unaltered

TgBDNF mice overexpress BDNF in the striatum, and a subpopulation of SVZ progenitor cells express the low-affinity neurotrophin receptor p75 (Gascon et al. 2005; Galvao et al. 2008). As the SVZ is in close proximity to the striatum, we tested for potential BDNF effects on cell proliferation by measuring the density of cells expressing the protein Ki-67. As shown in Figure 1, the density of Ki-67+ cells in the SVZ of young adult TgBDNF mice did not differ significantly from that of WT controls (P ≥ 0.05, t-test with Welch’s correction, t(₆) = 0.6352).

Figure 1.

Ki-67 immunoreactivity in the SVZ. (A–B) Cell labeling along the lateral wall of the ventricle in a WT (A) and TgBDNF mouse (B). The inset in A shows nuclear labeling at higher magnification. (C) Cell counts indicate no significant differences in the density of Ki-67+ cells between genotypes (P ≥ 0.05, unpaired t-test with Welch’s correction, n = 5 per genotype). Cc, corpus callosum; lv, lateral ventricle; st, striatum. Bar in A= 6μm, B = 40 μm.

BDNF augmentation does not enhance GC survival under normal conditions

To test for changes in GC survival, we used 2 methods to quantify and compare the numbers and density of new cells in the GCL of sex- and age-matched TgBDNF and WT mice following BrdU treatment. Combined BrdU/NeuN immunolocalization was used to measure density of new GCs in the GCL, and BrdU-immunoperoxidase staining was used for full coronal serial section cell counts (described below). As shown in Figure 2, high density of BrdU+/NeuN+ GCs was observed at 2 weeks post-treatment in transgenic (Tg) mice, followed by declines in density over time that were indistinguishable from that seen in WT mice. At 2 weeks, mean density of Brdu+/NeuN+ cells in the WT GCL was 522.5/mm² ± 27 SEM, while that measured in transgenic mice was 560.4 cells/mm² ± 39 SEM (P ≥ 0.05, unpaired t-test with Welch’s correction, t(₅) = 0.441). By 9 weeks, this was reduced to 359.4 cells/mm² ± 55 SEM in WT mice, and 352.0 cells/mm² ± 34 in TgBDNF mice (P ≥ 0.05, unpaired t-test with Welch’s correction, t(₅) = 0.114). The overall decline in the density of new GCs between 2 and 9 weeks averaged ~31% for WT mice, and ~36% for TgBDNF mice. Levels of apoptotic cell death in the GCL were also unchanged by BDNF overexpression (Figure 3). Numbers of TUNEL+ cells counted in the GCL of 12-week-old TgBDNF mice did not differ significantly from that of WT controls (P ≥ 0.05, t-test with Welch’s correction, t(₆) = 0.9608).

Figure 2.

Survival of adult-born granule cells is not altered by increasing BDNF. (A–B) Confocal images BrdU/NeuN colabeling in the GCL of a TgBDNF mouse. Arrows indicate double-labeled cells. (C) Orthogonal projection of a Brdu+/NeuN+ granule cell. (D) No differences are found in BrdU+/NeuN+ cell density between WT and TgBDNF mice at any survival interval (P ≥ 0.05, unpaired, 2-tailed t-tests, with Welch’s correction). Bars in A–B= 20 µm.

Figure 3.

Levels of programmed cell death in olfactory bulb are not altered by BDNF overexpression. (A–B) Examples of TUNEL-labeled nuclei (arrows) in the deep granule cell layer in a WT (A), and TgBDNF mouse (B). (C) No differences in numbers of TUNEL+ cells were observed between genotypes (P ≥ 0.05, unpaired t-test with Welch’s correction, n = 5 per genotype). sel, subependymal layer. Bar in B = 50 µm.

Olfactory bulb BDNF levels remain elevated in sensory-deprived TgBDNF mice

Sensory stimulation increases OB BDNF expression, while deprivation reduces levels of native BDNF (Degano et al. 2014; McLean et al. 2001). CAMKIIα gene expression is not upregulated by activity, nor reduced by sensory deprivation (Neve and Bear 1989; Murray et al. 1995). Under the CAMKIIα promoter, BDNF transgene expression is maintained and allows for constitutive BDNF expression during periods of reduced activity, as shown in the visual cortex of dark-reared TgBDNF mice (Gianfranceschi et al. 2003). ELISA comparisons of BDNF levels between left and right bulbs within nondeprived WT and TgBDNF controls showed no differences, and values were pooled within each genotype (paired, 2-tailed t-tests, P ≥ 0.05, t₍₂₎ = 0.596 for WT and 0.774 for transgenic mice). As expected, OBs from TgBDNF mice contained higher levels of BDNF protein compared to those from WT controls (Figure 4, P ≤ 0.05, 2-tailed, unpaired t-test with Welch’s correction, t(₉) = 7.64). There was no significant difference in BDNF content in deprived and nondeprived bulbs from TgBDNF mice at 4 weeks postocclusion (mean occluded = 147.5 pg/mg protein ± 44.0 standard deviation [SD], open = 158.6 pg/mg protein ± 39.8 SD, P ≥ 0.05, 2-tailed, paired t-test, t(₃) = 0.20). However, consistent with prior reports of BDNF changes with olfactory sensory activity, measures in deprived WT mice showed differences between left and right OBs (Jones et al. 2007; Degano et al. 2014). Nondeprived, stimulated OBs from WT mice contained more BDNF than deprived bulbs, which did not differ significantly from WT control levels (mean occluded = 54.9 pg/mg protein ± 18.5 SD, open = 86.7 pg/mg protein ± 26.0 SD, P = 0.047, paired t-test, t(₃) = 7.77).

Figure 4.

Box and whisker plots comparing ELISA measures of olfactory bulb BDNF in TgBDNF and WT mice. Individual values are shown. BDNF levels in transgenic mice are higher than those of WT controls (**P ≤ 0.01, unpaired t-test with Welch’s correction), and remain elevated above WT control levels after olfactory deprivation. In deprived WT mice, BDNF content in nondeprived bulbs is higher than that measured in deprived bulbs (*P ≤ 0.05, paired t-test). Cont, control bulbs.

Increased BDNF does not counter deprivation-induced loss of new GCs

New GCs are most susceptible to deprivation at 2–4 weeks after their birth, while undergoing morphological maturation (Yamaguchi and Mori 2005). Their dendritic spine density is reduced by odor deprivation and by depletion of TrkB, indicating a need for BDNF signaling, as well as activity, in the normal maturation of GC connectivity (Dahlen et al. 2011; Bergami et al. 2013). To test if transgene-sustained elevations in BDNF partially counter the effects of reduced activity, numbers of Brdu+ cells in the GCL were compared in odor-deprived WT and TgBDNF mice. To confirm effective unilateral sensory deprivation, OB sections were processed for TH-immunostaining (Baker et al. 1993). All deprived mice showed reduced glomerular layer TH expression in bulbs ipsilateral to occlusion by 4 weeks (Figure 5). Shrinkage of the deprived OB GCL (described below) precluded measurements of BrdU+ cell density.

Figure 5.

Deprived TgBDNF mice exhibit reductions in glomerular layer (GL) tyrosine-hydroxylase immunoreactivity (TH-IR) at 4 weeks postocclusion. (A) TH-IR in a nondeprived bulb. (B) TH-IR in the bulb ipsilateral to occlusion. The arrow in B indicates a TH+ cell body. Bar in B = 50 µm.

Results shown in Figure 6 summarize the progression of BrdU+ cell loss out to 8 weeks post-BrdU in normal and deprived mice of both genotypes. Counts in left and right bulbs within nondeprived WT and TgBDNF controls were not significantly different at any time point (P ≥ 0.05, paired, 2-tailed t-tests), and were combined to generate group mean control values (n = 6 samples per genotype). By 4 weeks, the GCL ipsilateral to naris occlusion contained fewer BrdU+ cells in WT animals, compared to contralateral counts (mean occluded = 1403 ± 197 SEM, open = 1978 ± 261, P ≤ 0.05, paired, 2-tailed t-test, t₍₃₎ = 3.62). TgBDNF mice showed similar differences in BrdU+ cell numbers (mean occluded = 1427 ± 128 SEM, open = 1982 ± 162, P ≤ 0.05, paired, t-test, t₍₃₎ = 12.94). Cell loss continued for both genotypes out to 8 weeks (mean WT occluded = 899 ± 52 SEM, open = 1440 ± 152, P ≤ 0.05, paired t-test, t₍₃₎ = 3.88; Tg occluded = 966 ± 46, open = 1301 ± 66, P ≤ 0.05, paired, t-test, t₍₃₎ = 3.99). At this time, numbers of remaining BrdU+ cells in deprived bulbs did not differ significantly between genotypes (P = 0.376, unpaired t-test with Welch’s correction). Lack of a survival effect was further indicated by measures of apoptotic cell death in the GCL (Figure 7A). At 4 weeks after naris occlusion, TUNEL+ cells were more abundant in deprived bulbs than in contralateral bulbs in both genotypes (P ≤ 0.05, paired t-tests, t(₃) = 3.66 for WT and 11.78 for Tg).

Figure 6.

Increased BDNF does not counter deprivation-induced loss of adult-born GCs. Counts of BrdU+ cells in the GCL show that cell loss in bulbs ipsilateral to occlusion is significantly higher compared to contralateral counts in both WT and TgBDNF mice (*P ≤ 0.05, paired 2-tailed t-tests).

Figure 7.

(A) Deprivation increases apoptotic cell death in the GCL of both WT and TgBDNF mice. In all deprived mice, bulbs ipsilateral (ipsi) to occlusion contain more TUNEL+ cells compared to contralateral bulbs (contra) and bulbs from nondeprived control mice (*P ≤ 0.05, paired, 2-tailed t-tests. (B) Estimations of granule cell layer (GCL) volume 8 weeks after unilateral deprivation indicate that in mice of both genotypes, deprivation significantly reduced GCL volume (*P < 0.05, paired, 2-tailed t-tests).

Prolonged olfactory deprivation in rodents reduces OB volume due to substantial GC loss and impaired GC replacement (Cummings et al. 1997). We noted all deprived bulbs appeared smaller at 8 weeks postocclusion. Estimations showed significant loss of GCL volume (~15–23%) in deprived bulbs relative to control bulbs in both WT and TgBDNF mice (Figure 7B; P ≤ 0.05, paired, 2-tailed t-tests, t(₃) = 5.35 and 6.59, respectively). Paired t-test comparisons in nonoccluded controls showed no differences between left and right GCL volumes (P ≥ 0.05, t(₂) = 1.66 and 0.585, WT and Tg, respectively), and these values were combined for each genotype (n = 6 bulb measures). Comparison of these volume estimates in WT and TgBDNF controls showed no significant differences (P ≥ 0.05, unpaired t-test with Welch’s correction, t(₈) = 1.276).

Discussion

BDNF and adult neurogenesis

Survival of adult-born olfactory GCs is regulated by multiple factors, with neural activity playing a pivotal role in controlling functional incorporation and cell viability. Odorant-driven activity activates mitogen-activated protein kinase signaling (MAPK; aka Erk1/2) in new GCs, leading to downstream increases in bulbar levels of the anti-apoptotic protein Bcl-2, as well as in numbers of surviving GCs (Miwa and Storm 2005). The activity-dependent nature of BDNF secretion and the resulting TrkB signaling, which also converges on MAPK, makes it ideally suited to promote development of young CNS neurons (Reichardt 2006; Gottmann et al. 2009). Indeed, BDNF infusions, and conditions that elevate hippocampal BDNF, particularly wheel running-induced activity, promote dendrite growth, spine maturation and enhanced survival in new dentate GCs (Scharfman et al. 2005; Rossi et al. 2006; Kobilo et al. 2011). In contrast, eliminating BDNF signaling through selective TrkB knockout in these new GCs reduces dendritic arborization, spine density and ultimately cell survival (Bergami et al. 2008). We previously found with Golgi staining that morphologically mature olfactory GCs in TgBDNF mice exhibit increased spine density and more mature spines than their WT counterparts (McDole et al. 2015). Their responsiveness to increased BDNF raised the possibility of survival effects in adult-born GCs similar to those seen in hippocampus, however we found no evidence of either increased GC production or survival in this model. TrkB deletion in new olfactory GCs reduces apical dendritic spine development by ~50%, but remarkably, even this manipulation appears to have no effect on GC survival, although dopaminergic PG cell survival is clearly reduced (Galvao et al. 2008; Bergami et al. 2013). The morphological effects of altered BDNF-TrkB signaling in olfactory GCs are independent of survival outcomes.

Consistent with prior reports, we found BDNF showed evidence of activity-dependent expression in the OBs of deprived WT mice. Levels of native BDNF protein in bulbs receiving stimulation were higher than those in deprived bulbs which were comparable to control levels. As expected, deprivation had no effect on transgene-derived bulb BDNF levels in TgBDNF mice, yet reduced GC survival to the same extent as in WT animals, indicating that unlike the beneficial effects seen in deprived visual cortex circuitry, increased BDNF does not compensate for reduced olfactory activity in terms of outright neuron viability (Gianfranceschi et al. 2003). A severe reduction in activity-dependent BDNF release with naris occlusion would be expected, although GC survival in nondeprived TgBDNF mice also was not improved. Activity per se plays a dominant role in regulating GC survival during the critical period of their maturation. Young GCs require a minimal level of activation to survive, and genetically increasing GC intrinsic excitability prevents deprivation-induced deficits in their dendritic and synaptic development, while also rescuing their survival (Kelsch et al. 2009; Lin et al. 2010; Dahlen et al. 2011).

Under the CAMKIIα promoter, GCs themselves express the transgene, as do projection neurons in the olfactory cortices. In CNS neurons, both native BDNF and overexpressed BDNF tagged with green fluorescent protein, are chiefly targeted to the regulated secretory pathway for activity-dependent release, with less directed to constitutive secretion (Brigadski 2005; Kuczewski et al. 2009). While secretory granules containing BDNF localize to axons and cell somata, they are preferentially trafficked to the dendritic compartment, and are released from this location (Kuczewski et al. 2009; Dieni et al. 2012; Leschik et al. 2019). In nondeprived TgBDNF mice, transgene-derived BDNF is therefore potentially available in an activity-dependent manner to young, TrkB- and BDNF-expressing GCs through anterograde and autocrine routes. Within the first week of their arrival in the OB, new, anaxonic GCs receive inputs to their basal and proximal apical dendrites from projection neurons in the AON and piriform cortex, areas with increased BDNF expression in TgBDNF mice (Deshpande et al. 2013). Interestingly, we noted that mature GCs in these animals show increased spine density in the proximal apical dendritic domain (McDole et al. 2015).

Autocrine mechanisms regulating adult neurogenesis have been elegantly demonstrated by Mu-Ming Poo’s laboratory, in work that selectively deleted or overexpressed BDNF in adult-born dentate GCs in vivo (Wang et al. 2015). Deletion only in new GCs impairs their morphological development to the same extent as that produced by genetic BDNF knockout in forebrain. Overexpression of BDNF only in new GCs normalizes their dendritic development in these BDNF knockout mice (Wang et al. 2015). Moreover, selective BDNF deletion only in young GCs blocks increases in their dendritic growth in response to wheel running-induced neural activity, demonstrating a need for activity-dependent secretion by the cells themselves. Technical advances in imaging active autocrine BDNF-TrkB signaling dynamics in cultured neurons have revealed that stimulation of an individual dendritic spine triggers BDNF release from that spine and subsequent TrkB activation in the same spine (Harward et al. 2016). Other trophic factors also show evidence of eliciting dendritic autocrine signaling, including insulin-like growth factor-1, which is released by mitral cells at dendrodendritic mitral:GC synapses (Liu et al. 2017). It is tempting to speculate that BDNF overexpressed by olfactory GCs may exert autocrine/paracrine effects at mitral:GC contacts, contributing to the increased spine density measured in the distal apical dendritic compartment of TgBDNF mice (McDole et al. 2015).

Comparison to BDNF augmentation in the adult SVZ

Our finding that increases in OB BDNF do not enhance GC survival rate contrasts with effects reported following BDNF administration in the SVZ or ventricle. By acting on SVZ progenitor cells, these treatments increase the numbers of new OB GCs 2–4 weeks after BrdU treatment (but see Galvao et al. 2008). In these studies, BDNF augmentation was combined with repeated, daily BrdU treatment for 14–30 days, with animals euthanized 1 or 2 days after termination of treatment (Zigova et al. 1998; Benraiss et al. 2001, 2012; Bath et al. 2008).The resulting mixed population of BrdU+ cells therefore ranged from those born in the SVZ 2–4 weeks earlier to those born within the previous 5–8 days (the minimum time needed for neuroblast migration to the OB). Survival rates for GCs at defined ages or maturational state were not assessed. For rats given intraventricular administration of AAV-BDNF, this ongoing addition of new GCs did not increase either total neuron numbers or bulb volume. It did however increase the incidence of apoptotic cells, suggesting that neuron lifespan was limited by enhanced elimination to maintain normal cell numbers (Benraiss et al. 2012). This is supported by a study by Reumers et al. (2008), in which mice received SVZ injections of LacZ- or BDNF-encoding lentivirus (LV-BDNF), combined with a green fluorescent protein-lentivirus (LV-GFP). Increased addition of GFP+ bulb cells was seen 3 weeks later in LV-BDNF treated mice. However, this was followed by a reduction in the density of GFP+ and BrdU+ OB cells by 8 weeks that exceeded levels of cell loss measured in controls, suggesting that long-term GC survival after SVZ BDNF administration is limited by cell elimination to maintain stable numbers of GCs over time (Reumers et al. 2008, in Supplementary Information).

Summary

Our findings indicate that endogenous increases in olfactory GCL BDNF do not improve the survival of adult-born OB GCs. Increasing BDNF expression in GCs and olfactory cortex enhances GC dendritic spine density, while elimination of TrkB expression in new GCs impairs their ability to elaborate normal spine density, however neither of these morphological effects significantly impacts survival (Bergami et al. 2013; McDole et al 2015). Under normal conditions, the role of BDNF-TrkB signaling may be to regulate adaptive structural plasticity and spine maintenance as new GCs are recruited into those local olfactory networks that provide sufficient levels of activity to foster their stable incorporation.

Funding

This work was supported by the National Institute on Deafness and Other Communication Disorders grant DC012425 to K.G.

Conflicts of Interest

The authors declare there are no conflicts of interest in the publication of this material.