Abstract
Abnormal neuronal morphological features, such as dendrite branching, axonal branching, and spine density, are thought to contribute to the symptoms of depression and anxiety. However, the role and molecular mechanisms of aberrant neuronal morphology in the regulation of mood disorders remain poorly characterized. Here, we show that neuritin, an activity-dependent protein, regulates the axonal morphology of serotonin neurons. Male neuritin knock-out (KO) mice harbored impaired axonal branches of serotonin neurons in the medial prefrontal cortex and basolateral region of the amygdala (BLA), and male neuritin KO mice exhibited depressive and anxiety-like behaviors. We also observed that the expression of neuritin was decreased by unpredictable chronic stress in the male mouse brain and that decreased expression of neuritin was associated with reduced axonal branching of serotonin neurons in the brain and with depressive and anxiety behaviors in mice. Furthermore, the stress-mediated impairments in axonal branching and depressive behaviors were reversed by the overexpression of neuritin in the BLA. The ability of neuritin to increase axonal branching in serotonin neurons involves fibroblast growth factor (FGF) signaling, and neuritin contributes to FGF-2-mediated axonal branching regulation in vitro. Finally, the oral administration of an FGF inhibitor reduced the axonal branching of serotonin neurons in the brain and caused depressive and anxiety behaviors in male mice. Our results support the involvement of neuritin in models of stress-induced depression and suggest that neuronal morphological plasticity may play a role in controlling animal behavior.
- amygdala
- anxiety
- axonal morphology
- depression
- FGF
- serotonin
Significance Statement
Axonal atrophy of serotonin neurons is one of the representative neuroanatomical features of depression. We found that the secreted/membrane-anchored neurotrophic factor neuritin regulated axonal branch formation, which is involved in the development of depression and anxiety. In addition, neuritin and the secreted signaling protein fibroblast growth factor 2 (FGF-2) cooperate to promote axonal branching in serotonin neurons. Furthermore, the inhibition of FGF signaling promoted axonal branching impairments and depressive behavior in mice. Taken together, these findings suggest that neuritin regulates axonal branching in serotonin neurons and that the loss of neuritin is related to the development of depression. FGF signaling is involved in the neuritin-mediated axonal branching of serotonin neurons.
Introduction
Stress is linked to depression and anxiety disorders and has been implicated in various neuronal abnormalities in key brain regions associated with depression and anxiety. Chronic stress treatment is often used to induce depression- and anxiety-like behaviors in animal models (Stepanichev et al., 2014). Exposure to unpredictable chronic stress (UCS) has been shown to promote depressive and anxiety behavior in rats and mice (Willner, 1997; Pollak et al., 2010).
Stress-mediated morphological remodeling of neuronal cells underlies the development of depressive mood in model animals. Chronic stress treatment leads to decreases in axonal branching, dendritic branching, and the length and number of dendritic spines in the hippocampus, prefrontal cortex, and some amygdala nuclei (Sousa et al., 2000; Vyas et al., 2002; Radley et al., 2004; Kuramochi and Nakamura, 2009; Zahrai et al., 2020). However, the precise molecular mechanisms by which chronic stress alters dendritic arborization are still elusive. Serotonin has been thought to be a main player in the mechanism of depression for decades. A correlation between decreased serotonin and depression was first suggested in 1967 (Coppen, 1967). The effectiveness of selective serotonin reuptake inhibitors (SSRIs) as antidepressants supports the idea that depression is the result of abnormalities in serotonin signaling in the brain (Feighner, 1999; Kryst et al., 2022). Serotonin levels are correlated with depressive behavior, and SSRIs act as antidepressant drugs. However, how low-level serotonin induces depressive mood and the detailed mechanism by which SSRIs treat depressive disorders are still unknown (Blokhin et al., 2020).
Neuritin, also known as candidate plasticity gene 15, is an extracellular protein that is anchored to the cell surface by a glycosylphosphoinositide link (Naeve et al., 1997). Several studies have reported the role of neuritin in neuronal morphogenesis and function. Neuritin induces neurite sprouting, neurite arborization, neurite outgrowth, neural cell survival, and synaptic plasticity regulation (Yao et al., 2012, 2016; Shimada et al., 2013; Azuchi et al., 2018). The overexpression of neuritin increases axonal branch formation in granule cells and contributes to mossy fiber sprouting after epileptic seizures (Shimada et al., 2016). The overexpression of neuritin promotes increases in axonal branch formation in Xenopus motoneurons (Javaherian and Cline, 2005). Increasing evidence indicates that neuritin can regulate axonal and dendritic branching in various types of neurons. Interestingly, gene expression studies have shown that chronic stress treatment and social defeat stress decrease the expression level of neuritin in the mouse and rat hippocampus (Son et al., 2012; Sathyanesan et al., 2018) and that the expression of neuritin is correlated with the dendritic branching of granule cells in the dentate gyrus. In addition, shRNA-mediated decreases in neuritin expression in the hippocampus have been shown to induce depressive and anxiety behaviors in rats (Son et al., 2012). Chronic treatment with fluoxetine, an antidepressant drug, increased neuritin mRNA levels in the medial prefrontal cortex (mPFC) and dentate gyrus of the hippocampus in rats (Alme et al., 2007). Therefore, neuritin-mediated regulation of neural morphology may be involved in the development of depression and anxiety disorders.
Here, we elucidated the effects of stress-regulated neuritin expression and neuritin on the axonal branching of serotonin neurons both in dissociated cultures and in the mouse brain. Our results indicate that UCS decreases the expression of neuritin in the brain and that neuritin can promote axonal branch formation in serotonin neurons both in vitro and in vivo. We found that neuritin-dependent activation of fibroblast growth factor (FGF) signaling was also involved in the axonal branching of serotonin neurons in vitro. Finally, pharmacological inhibition of FGF signaling deceased the axonal branching of serotonin neurons in the brain and induced depressive behavior in mice. Our study suggested that a chronic stress-mediated reduction in neuritin expression decreases axonal branch formation in serotonin neurons in the brain and that this neuritin-dependent axonal branching may be involved in stress-mediated depressive and anxiety behaviors in mice.
Materials and Methods
Reagents
PD173074 and recombinant FGFs were obtained from Cayman and PeproTech, respectively. AZD4547 was obtained from Selleck Chemicals (for in vitro experiments) and ChemScene (for in vivo experiments).
Mice
Neuritin knock-out (KO) mice were obtained from the Jackson Laboratory (Nrn1tm1.2Ndiv, stock number 018402; https://www.jax.org/strain/018402). The mice were backcrossed into the BALB/c strain (CLEA Japan) for >10 generations. Only male animals were used in the behavioral and immunohistochemical experiments. All mice were group housed and maintained under a 12:12 h light/dark cycle. All animal experiments were approved by the Animal Care and Use Committee of the Tokyo Metropolitan Institute of Medical Science and were performed according to their recommendations.
Cell culture
Serotonin neurons were collected from Embryonic Day (E)14 Sprague Dawley rats (CLEA Japan) or E14 embryos of neuritin+/− × neuritin+/− mice. In brief, brains were collected from the embryos and cut along the median line. From the medial side, the caudal half of the dorsal mesencephalon was carefully collected and treated with 0.05% trypsin (Invitrogen). After gentle trituration, the isolated cells were plated onto ϕ18 mm coverslips coated with poly-L-lysine in 12-well plates at a density of 2–4 × 104 cells/well. The cells were cultured in 10% FCS/Neurobasal medium for 3 h, and the medium was then changed to Neurobasal medium/2% B-27 supplement (Invitrogen)/2 mM GlutaMAX (Invitrogen).
Preparation and application of Fc-tagged proteins
Soluble recombinant Neuritin-Fc and Fc were prepared as described previously (Uemura et al., 2010). In brief, the respective expression vectors were transfected into FreeStyle 293F cells (Invitrogen). Cell culture medium containing Fc-tagged protein was collected. The protein was purified on a protein A-Sepharose bead column and eluted with 3 M MgCl2. The eluted solution was concentrated with Amicon concentrators (10 kDa cutoff), and the buffer was changed to PBS. Purified proteins were administered to the cultured neurons at day in vitro (DIV) 1.
Immunocytochemistry
Cultured serotonin neurons were fixed with 4% paraformaldehyde for 15 min at room temperature, followed by three washes with PBS. After treatment with 0.1% Triton X-100/PBS (PBST) for 15 min at room temperature for permeabilization and a brief wash with PBS, the cells were treated with 2% goat serum/PBS for 1 h at room temperature. The cells were then immunostained using the following antibodies: anti-Tuj1 1:1,000 (BioLegend, 801201), anti-serotonin 1:4,000 (Sigma-Aldrich, S5545), and anti-GFP 1:1,000 (Merck Millipore, AB16901). Secondary antibodies were conjugated with Alexa Fluor 488 and Alexa Fluor 568 1:2,000 (Thermo Fisher Scientific, A11034, A11036, A11031, and A11039). We used Vectashield as the mounting medium.
RNA interference (RNAi)
The sequence of the targeted region of rat FGFR1 mRNA for Stealth RNAi (Invitrogen) was 5′-CAGCUGCCAAGACGGUGAAAUUCAA-3′, which corresponds to nucleotide residues 506–530. As a control, we used a sequence (5′-AAGGAGGAAAGACCGCUGAAUCCUG-3′), which does not target any known vertebrate gene. The efficiency of the siRNA and control RNA has been confirmed previously (Shimada et al., 2016). The neurons were transfected with RNA and GFP-expressing plasmids using a Nucleofector (Amaxa, Lonza Bioscience) before plating.
Immunohistochemistry
The mice were anesthetized and perfused transcardially with 4% paraformaldehyde in 0.1 M PB, pH 7.4. The brains were rapidly removed and postfixed for >24 h with 4% paraformaldehyde and then cryoprotected in 20% sucrose/PBS. The brains were frozen on the stage of a microtome (Microm). The sections were coronally cut at 30 μm and preserved at 4°C with 0.04% sodium azide/PBS. The brain slices were washed with 0.1% PBST for 10 min three times before being blocked with 4% goat serum in PBST for 1 h. The slices were then immunostained using an anti-serotonin transporter antibody 1:1,000 (Merck Millipore, PC177L), an anti-serotonin antibody 1:3,000 (Sigma-Aldrich, S5545), and an anti-GFP antibody 1:1,000 (Merck Millipore, AB16901). Secondary antibodies were conjugated with Alexa Fluor 488 or Alexa Fluor 568 1:1,000 (Thermo Fisher Scientific, A11034, A11036, and A11039). The slices were stained with DAPI (Thermo Fisher Scientific, D1306) with 10 ng/ml in PBST for 10 min at room temperature after secondary antibody treatment. Vectashield was used as the mounting medium.
Microscopy and image processing
Images were captured with an AxioImager Z.1 (Carl Zeiss; for the dissociated serotonin neurons and for the dorsal raphe nucleus in brain sections) and an LSM 710 confocal microscope (Carl Zeiss; for the mPFC, hippocampus, and amygdala in brain sections). For 3D images, stack images were acquired with the interval between images set to 0.300 μm. The region of interest template size was 212.55 × 212.55 × 11.70 µm (40 sections per one stack images).
Quantification of axons of serotonin neurons
Quantification of in vitro axon length was performed using ImageJ with the NeuronJ plug-in (Meijering et al., 2004). The longest neurite was defined as an axon, and protrusions shorter than 20 μm were excluded from axon branches. The number of primary branches divided by the axonal length was defined as the number of primary branches per unit length of axon (100 μm axon). Quantification of in vivo axon density in the mPFC, hippocampus, and amygdala was performed using the Imaris software as previously described (Zahrai et al., 2020), with slight modifications. Briefly, serotonin transporter-immunolabeled fibers were reconstructed using Imaris' filament tool, with a filament diameter >0.5 μm, and axonal volumes, axonal lengths, and the number of axonal branch points were determined by surface reconstruction of serotonin transporter-positive fibers. A minimal ratio of branch length to trunk radius of 10 was used to reduce the background signal, with further filter processing to remove short fibers (total length <5 μm) to eliminate artificial staining from quantification.
Immunoblot
Mouse mPFC, hippocampus, amygdala, and dorsal raphe nucleus were collected after decapitation. In brief, we collected mPFC regions from coronal sections from approximately bregma 1 to 3 mm and amygdala regions and hippocampal regions from approximately bregma −1 to −3 mm. We also collected the dorsal raphe region from coronal sections from approximately bregma −4 to −5 mm. The collected brain regions were homogenized with lysis buffer [150 mM NaCl, 5 mM EDTA, 20 mM HEPES, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, phosphatase inhibitors (10 mM NaF and 1 mM Na3O4V), and Roche protease inhibitor cocktail], pH 7.2, by mechanical homogenization on ice and then centrifuged. The resultant soluble fractions were immunoblotted. Primary antibodies against neuritin 1:250 (Merck Millipore, MABN1840), α/β tubulin 1:1,000 (Cell Signaling Technology, 2148), FGF-2 1:300 (Cell Signaling Technology, 46879), phospho-FGFR1 1:7,000 (Merck Millipore, 06-1433), and FGFR1 1:200 (Cell Signaling Technology, 9740), and HRP-conjugated secondary antibodies 1:4,000–10,000 (Jackson Laboratory, 111-035-033 and 115-035-003) were used.
Elevated plus maze test (EPMT)
The EPM is a cross-shaped maze consisting of four arms arranged in the shape of a plus sign, two opposing open arms (30 × 6 cm), enclosed arms (30 × 6 with 15 cm height walls), and a central zone (6 × 6 cm; Muromachi). All arms were set 40 cm above the ground and illuminated via dim light (including the enclosed arms). The mice were placed in the central zone facing one of the open arms and were allowed to move freely for 5 min. The time the mice spent in the open arms was measured with the Anymaze software (Stoelting).
Sucrose preference test (SPT)
During the adaptation period, the mice were individually housed and provided two bottles containing pure water or 1% sucrose solution for 24 h. The mice were allowed free access to both bottles. The 24 h test period followed immediately after the adaptation period. The intake of the sucrose solution and pure water was recorded after the test. The positions of the two bottles were exchanged every 12 h over the whole period. The sucrose preference index was calculated as follows: [intake of the sucrose solution (g)] / [intake of the sucrose solution (g) + intake of pure water (g)] × 100.
Novelty-suppressed feeding test (NSFT)
The floor of the box (37 × 31 × 19 cm) was covered with bedding. Before the test, the mice were deprived of food for 18 h. During the test, a single pellet of chow was placed in the center of the box. The animal was placed at one corner of the box, and the latency time to eat the pellet was recorded. Immediately after the bite, the mouse was moved back to its home cage with sufficient food. The amount of food consumed in the subsequent 10 min was measured (home cage feeding). For the NSFT test, there was a cutoff time of 20 min.
Tail suspension test (TST)
An automated TST device (NeuroScience) was used to measure the duration of behavioral immobility. Mice were suspended by the tail with a tape to a metal chain connected to a strain gauge. Mice were positioned such that the tip of their tail was attached to the bottom of the chain. A strain gauge detected the force generated by any movements of the mouse. The test duration was 10 min, and the data from the latter 8 min were used for the analysis. The total duration of immobility was calculated as the time the force of the mouse's movements was below a preset threshold. The following settings were used in all the experiments: sampling rate, 100 Hz; input range, 2.5 V; threshold, 5 (rest %); event gap, 0.5 s; maximum frequency, 10 Hz; minimum frequency, 1 Hz; and minimum duration, 0.1 s.
UCS
Six-week-old BALB/c male mice were divided into two groups: the control group and the UCS group. Mice in the control group were subjected to gentle handling (three times a week). Mice in the UCS group were subjected to daily stressors for 4 weeks. Two stressors were applied daily and were chosen pseudorandomly with no repetition of the same stressor for 2 consecutive days to avoid habituation. All UCS group mice were subjected to the same type, onset, and duration of daily stressors. The interval between the two daily stressors was at least 1 h. The stressors used are shown in Table 1.
UCS used in this manuscript
Virus injection
For the virus experiments, we used 8-week-old BALB/c male mice. Under anesthesia, the mice were placed in a stereotaxic device (Narishige). A total of 500 nl of purified virus, AAV9-CMV > mNrn1:T2A:EGFP:WPRE (1.85 × 1012 GC/ml; VectorBuilder) for neuritin expression or AAV9-CMV > EGFP:WPRE (1.81 × 1012 GC/ml;VectorBuilder) for the control, was delivered bilaterally into the basolateral region of the amygdala (BLA) using a 10 μl Hamilton syringe (50 nl/min). The location of the BLA was as follows: AP, −1.5 mm; ML, ±2.9 mm; and DV, 4.9 mm. After injection, the needle remained in place for 10 min before being removed, and the skin was sutured. Following surgery, the animals were kept in group housing for 4 weeks to allow sufficient expression, after which UCS was applied. After the behavioral test, we investigated GFP expression by immunohistochemistry. Mice in which GFP expression could not be confirmed in the amygdala were excluded from the results.
Experimental design and statistical analysis
A GraphPad Prism version 8.0 software was used for statistical analyses of the data. The detailed methods are described in the figure legends. Differences were considered significant when p < 0.05. All experiments requiring the use of animals, directly or as a source of cells, were subjected to randomization.
Results
Neuritin promotes axonal branching in serotonin neurons in vitro and in vivo
To investigate the function of neuritin in the axonal morphogenesis of serotonin neurons, we applied purified Neuritin-Fc protein to the culture medium of serotonin neurons. Serotonin neurons were prepared from the dorsal raphe nucleus of E14 rats and stained with an anti-serotonin antibody on DIV 4. Serotonin-positive neurons were defined as serotonin neurons. The serotonin neurons were plated at low density on poly-L-lysine-coated coverslips and cultured, avoiding contact with other cells as much as possible. The longest neurite was evaluated as an axon. We measured the length of the axon and the number of branches per 100 μm of axon. The branching of serotonin neuron axons was promoted by the Neuritin-Fc administration (Fig. 1A,B) in a dose-dependent manner up to 1 µg/ml, whereas axonal length was not altered by neuritin treatment (Fig. 1A,B).
The axonal branching of cultured serotonin neurons is regulated by the neuritin protein level. A, Images of serotonin neurons from rat embryos. Immunodetection of serotonin-positive neurons treated with purified Fc or Neuritin-Fc protein (1 µg/ml) is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. B, Quantification of axon length (left) and the number of primary branches per 100 µm of axon (right) with various concentrations of Neuritin-Fc protein [n = 30/3 cells/independent experiments, one-way ANOVA followed by Dunnett's posttest; F(5,12) = 0.1109; p = 0.9963 (0.1 μg/ml Neuritin-Fc); p = 0.8566 (0.3 μg/ml); p = 0.9988 (1 μg/ml); p = 0.8480 (3 μg/ml); p = 0.6494 (10 μg/ml; axon length); F(5,12) = 0.3111; p = 0.9977 (0.1 μg/ml Neuritin-Fc); p = 0.0575 (0.3 μg/ml); ***p = 0.0006 (1 μg/ml); ***p = 0.0003 (3 μg/ml); and ***p < 0.0001 (10 μg/ml; number of branches) compared with neurons with Fc protein. The data were averaged and are presented by experimental conditions]. C, Images of serotonin neurons from neuritin+/+ (WT) or neuritin−/− (KO) mouse embryos with purified Fc or Neuritin-Fc protein (1 µg/ml). Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. D, Quantification of axon length (top) and the number of primary branches per 100 µm of axon [bottom; n = 30/3 cells/independent experiments, two-way ANOVA followed by Sidak's posttest, treatment × genotype interaction F(1,8) = 0.0236; p = 0.882; genotype, F(1,8) = 2.151; p = 0.181; treatment, F(1,8) = 0.495; p = 0.502; p = 0.915 (WT) and p = 0.808 (KO; Fc vs Neuritin-Fc, respectively); with Fc, p = 0.616, and with Neuritin-Fc, p = 0.489 (WT vs KO; axon length); treatment × genotype interaction, F(1,8) = 10.11; p = 0.013; genotype, F(1,8) = 1.745; p = 0.223; treatment, F(1,8) = 57.13; p < 0.001; *p = 0.029 (WT); and ***p = 0.001 (KO; Fc vs Neuritin-Fc, respectively); with Fc, #p = 0.026, and with Neuritin-Fc, p = 0.400 (WT vs KO; number of branches). The data were averaged and are presented by experimental conditions]. Error bars, SD.
An increase in extracellular neuritin through its addition promoted axonal branching on serotonin neurons. Then, to evaluate the effect of a decrease in neuritin protein on axonal branching, we observed cultured neurons from neuritin KO mice. We compared the axonal branching of serotonin neurons from neuritin+/+ mice (WT) and neuritin−/− mice (KO) at DIV 4. KO neurons exhibited poor axonal branch formation (Fig. 1C,D). Axonal length, however, was not altered by neuritin KO (Fig. 1C,D). In addition, the application of purified Neuritin-Fc protein rescued and promoted axonal branch formation (Fig. 1C,D). These results suggest that poor axonal branching in neurons from KO mice is a neuritin-dependent phenotype.
We next investigated the effect of neuritin on the branching of serotonin neurons in vivo. We prepared brain sections from 8- to 12-week-old adult WT and KO mice; observed the mPFC, hippocampal, and BLA regions; and evaluated the density of axons and the axonal branching of serotonin neurons. We calculated the serotonin transporter-staining–positive volume ratio and the number of branch points per axonal length unit. Consistent with cultured neurons, neuritin KO mice exhibited fewer dense axons and a smaller number of axonal branches in the mPFC and BLA. However, the density of serotonin neuron axons in the hippocampus was not altered in the KO mice, suggesting that factors other than neuritin may play a more dominate role in the regulation of serotonin axons in the hippocampus (Fig. 2A–C). We next examined whether the number of serotonin neurons was reduced in KO mice, resulting in a decrease in the density of serotonin axons in other brain regions. We counted serotonin-positive cell bodies in the dorsal raphe nucleus and found that the loss of neuritin did not affect the number of serotonin-positive neurons (Fig. 2D). We also investigated the expression levels of neuritin and found that the mPFC and amygdala exhibited relatively higher levels of neuritin than did the hippocampus, albeit not significantly. The dorsal raphe region showed significantly lower neuritin levels. A previous study showed that neuritin can function as a secreted protein (Putz et al., 2005). Therefore, this result suggests that the axonal morphology of serotonin neurons in vivo is more strongly regulated by neuritin in brain regions where serotonin neurons are projected than by neuritin expressed by serotonin neurons. Overall, the loss of neuritin impaired the axonal branch formation of serotonin neurons in the mPFC and amygdala. These experiments demonstrate the impact of neuritin on axonal branch formation both in vitro and in vivo.
The axonal density and branching of serotonin neurons in brain regions are regulated by the neuritin protein. A, Schematic illustration of regions of interest for immunofluorescence observation. Regions of interest are indicated as red squares and are located in the mPFC (prelimbic, PL), hippocampus (CA3), and BLA. Line drawings are adapted from the Allen Mouse Brain Atlas. B, Reconstructed flattened images of the PL, CA3, and BLA from WT and KO mice. Immunodetection of serotonin transporter-positive neurons is shown. Scale bar, 100 µm. C, Quantification of the number of branch points along the serotonin transporter-positive axons and axon volume in the images [n = 4/6 region images/mouse, two-tailed t test with Welch's correction, t(8.7992) = 5.7429 and ***p < 0.001 in the mPFC; t(9.7826) = 0.7969 and p = 0.4444 in the hippocampus; and t(7.1474) = 2.9325 and *p = 0.021 in the amygdala (number of axonal branches); t(9.8606) = 2.8624 and *p = 0.017 in the mPFC; t(9.9553) = 0.3769 and p = 0.7141 in the hippocampus; and t(10.0000) = 3.0379 and *p = 0.012 in the amygdala (axonal volume). The data are averaged and presented by mouse experimental group]. D, Schematic illustration of the region of interest for immunofluorescence observation. The region of interest, located in the dorsal raphe nucleus (DR), is indicated by a red rectangle. The line drawing is adapted from the Allen Mouse Brain Atlas. E, Left, Images of DR from WT and KO mice. Immunodetection of serotonin-positive neurons is shown. Scale bar, 200 µm. Right, Quantification of the number of serotonin-positive cells in the DR (n = 3/3 sections/mouse; two-tailed t test with Welch's correction; t(3.9938) = 0.2195; p = 0.837. The data are averaged and presented by mouse experimental group). F, Left, Immunoblot images of lysates from the indicated brain regions from WT mice. Right, The quantification of the neuritin levels in each homogenate shown in the left panel [n = 6 animals; one-way ANOVA followed by Dunnett's post hoc test; F(3,20) = 5.9603; p = 0.9534 (mPFC); p = 0.1285 (hippocampus); ***p < 0.0001 (dorsal raphe), compared with the amygdala]. Error bars, S.D.
Neuritin KO mice exhibit depressive and anxiety behaviors
Because a decrease in the density of serotonin neuron axons in the amygdala is one of the hallmarks of depression and anxiety (Kuramochi and Nakamura, 2009; Zahrai et al., 2020), it is plausible that neuritin-mediated poor axonal branching of serotonin neurons results in depressive and anxiety behaviors. To test whether poor axonal branching of serotonin neurons caused by the loss of neuritin leads to depressive and anxiety behaviors, we investigated the influence of neuritin KO on behavioral models of depression and anxiety in mice.
In the elevated plus maze test (Fernández Espejo, 1997; Bourin, 2015), neuritin KO mice spent less time in the open arms of the apparatus than did WT mice (Fig. 3A), and in the NSFT (Ibarguen-Vargas et al., 2008; Francois et al., 2022), neuritin KO mice also showed a significantly longer latency to feed than did WT mice (Fig. 3B), suggesting anxiety behavior. There was no effect on home cage feeding, indicating that neuritin KO had no general effect on metabolic status (Fig. 3B). In the sucrose preference test (Salazar et al., 2012; Fu et al., 2023), the KO mice showed less preference for the sucrose solution (Fig. 3C), and in the tail suspension test (Mul et al., 2016; Tsuchimine et al., 2020), the KO of neuritin markedly increased immobility time (Fig. 3D), which are typical depressive responses. Therefore, the loss of neuritin may be involved in the low axonal branching of serotonin neurons in the mPFC and BLA, which is related to depressive and anxiety behaviors in mice. However, we cannot exclude the possibility that the loss of neuritin affected other types of neurons and regulated other neuronal functions to promote depression and anxiety behaviors.
Neuritin KO mice exhibit depressive and anxiety behaviors. A, Time spent in the open arms of the EMPT by neuritin+/+ (WT) and neuritin−/− (KO) mice (n = 15 animals; two-tailed t test with Welch's correction; t(25.6710) = 3.3654; ***p = 0.0024). B, Latency time to feed in the NSFT (left; n = 15 animals; two-tailed t test with Welch's correction; t(20.1369) = 2.5142; *p = 0.0205) and food consumption of home cage feeding, following the NSFT (right; n = 15 animals; two-tailed t test with Welch's correction; t(27.9999) = 1.0963; p = 0.2823). C, Sucrose intake in the SPT of WT and KO mice (n = 15 animals; two-tailed t test with Welch's correction; t(27.9434) = 2.8203; **p = 0.0087). D, Immobility time in the TST of WT and KO mice (n = 15 animals; two-tailed t test with Welch's correction; t(25.5015) = 2.2941; *p = 0.0302). Error bars, SD.
UCS causes a decrease in neuritin expression and axonal branching of serotonin neurons
To examine the relationship between neuritin and depressive and anxiety behaviors caused by stress, we used a UCS model (Jung et al., 2014; Nollet, 2021) to assess the effects of neuritin on depression in mice. We exposed mice to UCS and investigated the neuritin expression levels and axon branching of serotonin neurons in different brain regions after the development of depression and anxiety (Fig. 4A). We collected mPFCs, hippocampi, amygdalae, and dorsal raphe nuclei from mice that experienced UCS for 4 weeks and compared their neuritin protein levels with those of control mice. Lower protein expression of Neuritin was detected in the mPFCs, hippocampi, and amygdalae of UCS group mice than in those of control mice as observed by immunoblotting (Fig. 4B), but the neuritin level did not change in the dorsal raphe. Using our UCS protocol, we succeeded in inducing depressive and anxiety behaviors in mice. In the TST, the UCS mice spent less time in the open arms in the EPMT (Fig. 4C) and exhibited a longer latency to feed in the NSFT (Fig. 4D), indicating that anxiety behavior has been induced. In addition, the UCS group of mice consumed less sucrose solution in the SPT (Fig. 4E), and the UCS group of mice exhibited longer immobility times (Fig. 4F), indicating that depressive behavior was facilitated. There was no difference in home cage feeding (Fig. 4E), indicating that food consumption was not affected by the UCS.
UCS induces a reduction in neuritin expression, causing depressive and anxiety behaviors and a decrease in axon density and axonal branches of serotonin neurons. A, Schematic illustration of the experimental design for the UCS, followed by the EPMT, SPT, NSFT, and TST. B, Top, Immunoblot images of lysates from the indicated brain regions from control and UCS mice. Bottom, Quantification of neuritin levels in each homogenate shown in the top panel (n = 6 animals; two-tailed t test with Welch's correction; t(8.9269) = 2.5036 and *p = 0.034 in the mPFC; t(9.8671) = 2.7715 and *p = 0.020 in the hippocampus; t(7.6485) = 3.0701 and *p = 0.016 in the amygdala; and t(8.2625) = 0.4106 and p = 0.692 in the dorsal raphe). C, Time spent in the open arms of the EMPT in control and UCS mice (n = 14 animals; two-tailed t test with Welch's correction; t(23.8605) = 3.3409; ***p = 0.0027). D, Latency time to feed in the NSFT (left; n = 14 animals; two-tailed t test with Welch's correction; t(25.5676) = 3.0824; ***p = 0.0049) and food consumption of home cage feeding, following the NSFT (right; n = 14 animals; two-tailed t test with Welch's correction; t(25.5443) = 1.1029; p = 0.2804). E, Sucrose intake index of the SPT in control and UCS mice (n = 14 animals; two-tailed t test with Welch's correction; t(25.7981) = 2.6447; *p = 0.0137). F, Immobility time in the TST of control and UCS mice (n = 14 animals; two-tailed t test with Welch's correction; t(21.1291) = 2.0905; *p = 0.0488). G, Reconstructed flattened images of the PL, CA3, and BLA from control and UCS mice. Immunodetection of serotonin transporter-positive neurons is shown. Scale bar, 100 µm. H, Quantification of the number of branch points along the serotonin transporter-positive axons and the volume of axons in the images [n = 4/6 region images/mice; two-tailed t test with Welch's correction; t(7.3741) = 3.8808 and ***p = 0.005 in the mPFC; t(9.0694) = 2.3795 and *p = 0.0411 in the hippocampus; and t(8.9506) = 2.3972 and *p = 0.040 in the amygdala (number of axonal branches); t(8.9912) = 2.3857 and *p = 0.041 in the mPFC; t(9.9988) = 0.7659 and p = 0.4615 in the hippocampus; and t(9.5170) = 5.2341 and ***p < 0.001 in the amygdala (axonal volume). The data are averaged and presented by mouse experimental group]. I, Top, Images of dorsal raphe nuclei from control and UCS mice. Immunodetection of serotonin-positive neurons is shown. Scale bar, 200 µm. Bottom, Quantification of the number of serotonin-positive cells in the dorsal raphe nucleus (n = 3/3 sections/mouse; two-tailed t test with Welch's correction; t(3.2184) = 0.3816; p = 0.727). The data are averaged and presented by mouse experimental group. Error bars, SD.
To investigate whether axonal degeneration of serotonin neurons can also be observed in UCS mice, we also compared the axonal density and branching level of serotonin neurons in the mPFC, hippocampus, and BLA. Immunohistochemical analysis revealed that UCS-exposed mice exhibited poor axonal branching in the mPFC and the BLA (Fig. 4G,H), indicating that axonal branching was impaired after UCS condition. Furthermore, UCS mice also exhibited fewer dense serotonin transporter-positive axons in the mPFC and the BLA, supporting the hypothesis that UCS induces axonal branch loss of serotonin neurons (Fig. 4G,H). Because neuritin functions as a soluble secreted protein (Putz et al., 2005), it is thought that neuritin expressed in the mPFC and BLA mainly regulates the axonal branching of serotonin neurons rather than neuritin expressed by serotonin neurons themselves. However, in the hippocampus, the density of serotonin transporter-positive axons was not altered after UCS, and axonal branching was slightly but significantly increased (Fig. 4G,H). The expression level of neuritin was decreased in the hippocampus, suggesting that the regulation of serotonin neurons axons in the hippocampus may occur in a neuritin-independent manner. The number of serotonin neurons was not decreased by UCS conditioning (Fig. 4I). A decrease in neuritin protein expression possibly contributed to this decrease in serotonin neurons axonal density and branching caused by UCS conditioning. Taken together, these results support the hypothesis that UCS decreases neuritin protein expression to impair the axonal branching of serotonin neurons in the mPFC and BLA and that decreased neuritin-mediated axonal branch impairment is involved in the development of depressive and anxiety behaviors in mice.
Neuritin overexpression in the amygdala suppresses stress-dependent behavioral changes and decreases axonal branching of serotonin neurons
To test directly whether neuritin is involved in stress-mediated axonal morphology regulation and the development of depression and anxiety, we investigated the influence of adeno-associated virus–mediated overexpression of neuritin in the amygdala on mouse behavioral models of depression and anxiety. Neuritin-overexpressing mice and control (GFP-expressing) virus-injected mice were randomly assigned to the nonstressed control or UCS conditioning group (Fig. 5A,B). Stress shortened the time spent in the open arms in the EPMT and increased the immobility time in the TST in GFP-expressing animals, and neuritin overexpression mildly reversed these phenotypes (Fig. 5C,F). Furthermore, UCS caused an increase in the latency to feed in the NSFT and a decrease in sucrose preference in the SPT in GFP-expressing animals (Fig. 5D,E), and these effects were completely blocked by overexpression of neuritin. Moreover, neuritin overexpression had no effect on nonstressed mice. In addition to these behavioral effects, the overexpression of neuritin before UCS exposure mildly reversed axonal branching impairment and completely prevented the stress-induced decrease in axonal volume in the amygdala (Fig. 5G,H). Moreover, neuritin overexpression did not affect axonal morphology in nonstressed mice. In summary, these results suggest that the expression of neuritin is involved in the regulation of the axonal branching of serotonin neurons in the amygdala and suggest that the stress-mediated reduction in the expression of neuritin could be one of the factors controlling the axonal morphology of serotonin neurons in the development of depression and anxiety in mice.
Neuritin prevents UCS-induced abnormal behavior and decreased serotonin axon branching. A, Schematic illustration of the experimental design for virus injection followed by UCS and behavioral tests. B, Schematic showing the viral constructs used in this study. The CMV promoter sequence was followed by neuritin, the T2A cleavage peptide from Thosea asigna virus, and EGFP (top). Example coronal section prepared after the behavioral tests showing the expression of GFP in the BLA region indicated by the white dotted line (left side). Scale bar, 1 mm (bottom). C, Time spent in the open arms of EMPT in nonstressed control and UCS mice infected with control or neuritin virus [n = 15 animals; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,56) = 2.702; p = 0.1058; virus, F(1,56) = 0.7215; p = 0.3993; stress, F(1,56) = 6.822; p = 0.0115; **p = 0.0078 (control) and p = 0.7464 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.8207; and stress, p = 0.1598 (control vs neuritin)]. D, Latency time to feed in the NSFT [left; n = 15 animals; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,56) = 8.076; p = 0.0062; virus, F(1,56) = 9.603; p = 0.0030; stress, F(1,56) = 13.64; p = 0.0005; ***p < 0.0001 (control) and p = 0.7969 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.9794; and stress, ***p = 0.0002 (control vs neuritin)] and food consumption of home cage feeding, following the NSFT [right; n = 15 animals; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,56) = 0.05134; p = 0.8216; virus, F(1,56) = 0.0008; p = 0.9782; stress, F(1,56) = 0.1056; p = 0.7466; p = 0.9967 (control) and p = 0.9162 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.9803; and stress, p = 0.9873 (control vs neuritin)]. E, Sucrose intake index of the SPT in nonstressed control and UCS mice infected with control or neuritin virus [n = 15 animals; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,56) = 3.881; p = 0.0538; virus, F(1,56) = 0.4476; p = 0.0388; stress, F(1,56) = 8.248; p = 0.0057; **p = 0.0023 (control) and p = 0.7755 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.9933; and stress, *p = 0.0109 (control vs neuritin)]. F, Immobility time of the TST in nonstressed control and UCS mice infected with control or neuritin virus [n = 15 animals; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,56) = 0.2348; p = 0.6299; virus, F(1,56) = 0.9395; p = 0.3366; stress, F(1,56) = 11.17; p = 0.0015; *p = 0.0180 (control) and p = 0.0940 (neuritin) (nonstress vs UCS, respectively); nonstress, p = 0.9288; and stress, p = 0.5216 (control vs neuritin)]. G, Reconstructed flattened images of the BLA from both virus-injected control and UCS mice. Immunodetection of serotonin transporter-positive neurons is shown. Scale bar, 100 µm. H, Quantification of the number of branch points along serotonin transporter-positive axon and the volume of axons in the images [n = 4/6 region images/mouse; two-way ANOVA followed by Sidak's posttest; virus × stress interaction, F(1,20) = 1.820; p = 0.1924; virus, F(1,20) = 3.118; p = 0.0927; stress, F(1,20) = 0.6279; p = 0.0210; p = 0.0259 (control) and p = 0.6671 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.9477; and stress, p = 0.0775 (control vs neuritin; number of axonal branches); virus × stress interaction, F(1,20) = 2.333; p = 0.1423; virus, F(1,20) = 3.891; p = 0.0625; stress, F(1,20) = 7.367; p = 0.0134; *p = 0.0141 (control) and p = 0.6534 (neuritin; nonstress vs UCS, respectively); nonstress, p = 0.9406; and stress, *p = 0.0443 (control vs neuritin; axonal volume). The data are averaged and presented by mouse experimental group]. Error bars, SD.
FGF signaling is required for neuritin-mediated axonal branching
Next, we focused on the molecular mechanisms by which neuritin regulates the axonal branching of serotonin neurons. We previously reported that neuritin interacts with FGF receptor protein 1 (FGFR1) and regulate FGF signaling in granule cells (Shimada et al., 2016). To investigate the involvement of FGF signaling in axonal branching in serotonin neurons, we applied the FGFR1 inhibitors PD173074 and AZD4547 to the neuronal culture media. Interestingly, inhibition of FGFR1 resulted in a decrease in the basal axonal branching ability of serotonin neurons (Fig. 6A,B). This result may correspond to the fact that neuritin KO serotonin neurons had less axonal branching than do WT neurons. The basal neuritin expression level may have maintained the essential FGF signaling intensity and contribute to axon branch formation in WT neurons. Inhibiting FGF signaling also reduced axonal branching under Neuritin-Fc treatment conditions (Fig. 6A,B). In addition, there were no significant differences in axonal lengths between Fc- and Neuritin-Fc-treated neurons under FGFR1-inhibited conditions.
FGF signaling is necessary for neuritin-mediated axonal branch formation in serotonin neurons. A, Images of serotonin neurons from rat embryos. PD173074 (100 nM) or AZD4547 (200 nM), FGFR inhibitors, and purified Fc or Neuritin-Fc protein (1 µg/ml) were applied at DIV 1. Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. B, Quantification of the number of primary branches per 100 µm of axon (n = 30/3 neurons/independent experiments; two-way ANOVA followed by Sidak's post hoc test; neuritin treatment × inhibitor interaction, F(2,12) = 2.278; p = 0.1450; neuritin treatment, F(1,12) = 8.340; *p = 0.0136; inhibitor, F(2,12) = 24.37; ***p < 0.0001. *p = 0.0428 (Fc + DMSO vs Fc + PD173074); *p = 0.0142 (Fc + DMSO vs Fc + AZD4547); ***p = 0.0004 (neuritin-Fc + DMSO vs neuritin-Fc + PD173074); ***p = 0.0002 (neuritin-Fc + DMSO vs neuritin-Fc + AZD4547); *p = 0.0154 (Fc + DMSO vs neuritin-Fc + DMSO); p = 0.8321 (Fc + PD173074 vs neuritin-Fc + PD173074); p = 0.8194 (Fc + AZD4547 vs neuritin-Fc + AZD4547)]. The data were averaged and are presented as independent experiments. C, Images of serotonin neurons from rat embryos. siRNA against FGFR1 or control RNA was transfected before plating, and purified Fc or neuritin-Fc protein (1 µg/ml) was applied at DIV 1. Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. D, Quantification of the number of primary branches per 100 µm of axon [n = 30/3 neurons/independent experiments; two-way ANOVA followed by Sidak's posttest; neuritin treatment × RNA interaction, F(1,8) = 7.527; p = 0.0253; neuritin treatment, F(1,8) = 1.378; p = 0.2741; RNA, F(1,8) = 27.79; ***p = 0.0008. p = 0.2107 (Fc + scRNA vs Fc + siRNA); ***p = 0.0009 (neuritin-Fc + scRNA vs neuritin-Fc + siRNA); *p = 0.0480 (Fc + scRNA vs Neuritin-Fc + scRNA); p = 0.5091 (Fc + siRNA vs neuritin-Fc + siRNA). The data are averaged and presented as independent experiments]. Error bars, SD.
We further investigated the role of FGFR1 in neuritin-mediated axonal branch formation by knocking down FGFR1 via siRNA. Transfection with FGFR1 siRNA abolished the ability of neuritin to promote axonal branching, whereas basal axonal branch formation was not altered by FGFR1 siRNA treatment (Fig. 6C,D). These results suggest that pharmacological inhibition of FGFR has a stronger effect on attenuating FGF signaling than does siRNA. Taken together, these results indicate that neuritin-mediated axon branch formation in serotonin neurons is an FGF signal-dependent event and that FGF signaling is required for the basal branch formation ability of serotonin neurons. In addition, a certain level of FGF signaling is effective and necessary for axonal branching even under control conditions.
FGF-2 and neuritin cooperate to promote axonal branching in serotonin neurons
There are 22 FGF subtypes in mammals, and they are divided into seven subfamilies (Itoh and Ornitz, 2011). As in our previous study (Shimada et al., 2016), we chose FGFs that represent these subfamilies, except the FGF-11/12/13/14 and FGF-19/21/23 subfamilies, because FGF-11/12/13/14 are not secreted, and FGF-19/21/23 shows weak FGFR signaling activation (Zhang et al., 2006). We applied FGF or vehicle (5 mM Tris and 0.1% bovine serum albumin), pH 7.6, to cultured serotonin neurons and quantified the number of axonal branches. Surprisingly, only FGF-2 promoted axonal branch formation in dissociated serotonin neurons (Fig. 7A,B). This result is different from the results obtained for granule cells, in which FGF-4, FGF-5, FGF-7, and FGF-8 promoted axonal branching and FGF-2 inhibited axon branch formation (Shimada et al., 2016). However, this result is consistent with that for FGF-2, which has been reported to promote branching in various types of neurons (Szebenyi et al., 2001; Klimaschewski et al., 2004; Rak et al., 2014). FGFs other than FGF-2 had no effect on the axonal branch formation of cultured serotonin neurons.
FGF-2 is involved in neuritin-mediated axonal branching. A, Images of serotonin neurons from rat embryos. Each FGF (40 ng/ml) was applied at DIV 1. Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. B, Quantification of the number of primary branches per 100 µm of axon [n = 30/3 cells/independent experiments; nested one-way ANOVA followed by Dunnett's posttest; F(7,711) = 1.720; comparing control, p = 0.9973 (to FGF-1); *p = 0.0489 (to FGF-2); p = 0.9789 (to FGF-4); p = 0.9999 (to FGF-5); p = 0.9995 (to FGF-7); p = 0.8013 (to FGF-8); and p = 0.9848 (to FGF-9). The data were averaged and are presented by experimental condition]. C, Images of serotonin neurons from WT and KO mouse embryos. FGF-2 (40 ng/ml) was applied at DIV 1. Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. D, Quantification of the number of primary branches per 100 µm of axon [n = 22-26/4 neurons/independent experiments; two-way ANOVA followed by Sidak's posttest; genotype × FGF-2 interaction, F(1,12) = 8.802; p = 0.0118; genotype, F(1,12) = 244.1; ***p < 0.0001; FGF-2, F(1,12) = 14.11; ***p = 0.0027. ***p = 0.0009 (WT + control vs WT + FGF-2); p = 0.8296 (KO + control vs KO + FGF-2). The data were averaged and are presented by experimental condition]. E, Images of serotonin neurons from rat embryos. FGF-2 (40 ng/ml) and purified Fc or neuritin-Fc protein (1 µg/ml) were applied at DIV 1. Immunodetection of serotonin-positive neurons is shown. Arrows indicate axonal branch points. Scale bar, 50 µm. F, Quantification of the number of primary branches per 100 µm of axon (n = 30/3 neurons/independent experiments; two-way ANOVA followed by Sidak's posttest; neuritin × FGF-2 interaction, F(1,8) = 2.101; p = 0.1852; neuritin, F(1,8) = 6.341; p = 0.0359; FGF-2, F(1,8) = 9.574; ***p = 0.0148. p = 0.0246 (Fc + control vs Fc + FGF-2); ***p = 0.4792 (neuritin-Fc + control vs neuritin-Fc + FGF-2); *p = 0.0455 (Fc + control vs Neuritin-Fc + control); p = 0.7207 (Fc + FGF-2 vs neuritin-Fc + FGF = 2). The data were averaged and are presented as independent experiments. Error bars, SD.
Next, we determined whether FGF-2 signaling was involved in neuritin-mediated axonal branching in serotonin neurons. Serotonin neurons from neuritin KO mice were cultured with FGF-2. FGF-2 did not promote axonal branching in neuritin-deficient serotonin neurons, whereas neuritin WT neurons responded to FGF-2 in terms of axonal branch formation (Fig. 7C,D). These findings indicate that neuritin is necessary to promote FGF-2-dependent axonal branching in serotonin neurons. We then performed additional experiments to investigate whether FGF-2 and neuritin have additive effects on promoting axonal branch formation. As shown in Figure 7, E and F, the administration of both FGF-2 and Neuritin-Fc did not further increase the axonal branching of the cultured serotonin neurons, suggesting that both factors work in the same pathway to regulate axonal morphogenesis. Neuritin can activate FGFR1, and neuritin is required for FGF-2-dependent FGFR1 activation, which is involved in the promotion of axonal branching in serotonin neurons. An increase in neuritin may increase the basal activation level of FGFR1 via FGF-2 but may not further increase the activation level in response to additional administration of FGF-2.
Inhibition of FGF signaling causes poor axonal branching of serotonin neurons in vivo and leads to the development of depressive and anxiety behaviors
To evaluate the physiological role of the FGF signaling- and neuritin-mediated axonal branching of serotonin neurons, we evaluated the importance of FGF signaling in the development of depressive and anxiety behaviors. First, we investigated the interaction between Neuritin and FGF signaling in vivo. The levels of phosphorylated FGFR1 in the brains of neuritin KO mice were lower than those in the brains of control mice (Fig. 8A). Therefore, pharmacological inhibition of FGF signaling activity may phenocopy neuritin KO mice. A previous study suggested that the FGF signaling inhibitor AZD4547 functions as an FGFR inhibitor in the brain by passing through the blood–brain barrier (Singh et al., 2012). We orally administered AZD4547 (50 mg/kg) or vehicle [1% carboxymethyl cellulose (CMC, Sigma-Aldrich), 0.9% NaCl, and 5% DMSO, 10 ml/kg], and AZD4547 treatment resulted in a reduction in phosphorylated FGFR1 levels (Fig. 8B). Then, we orally administered AZD4547 or vehicle (CMC) to BALB/c male mice for 4 weeks, followed by behavioral tests and preparation of brain sections from these mice (Fig. 8C). AZD4547 treatment shortened the time spent in the open arms in the EPMT (Fig. 8D), and the NSFT results revealed that AZD4547 treatment increased the latency to feed (Fig. 8E), suggesting the development of anxiety-like behavior. In addition, AZD4547-treated mice showed less preference for sucrose solution (Fig. 8F) in the SPT and longer immobility times in the TST (Fig. 8G), indicating that inhibiting FGF signaling promotes depression in these mice. Therefore, these results indicate that chronic inhibition of FGF signaling could promote depressive and anxiety behaviors in mice. Previous studies have shown that FGF-2 administration during and after UCS rescued the depressive phenotype in both mice and rats (Elsayed et al., 2012; Wang et al., 2018) and that depressive behaviors induced by olfactory bulbectomy were also reversed by FGF-2 administration in mice (Jarosik et al., 2011), indicating that an increase in the FGF-2 protein level prevents the development of depressive behaviors. Taken together with these reports and our results of AZD4547 administration, these findings suggest that the regulation of FGF-2 signaling is crucial for the induction of depressive and anxiety behaviors.
Inhibition of FGF signaling causes depressive and anxiety behaviors and decreases the axon density and the axonal branch number of serotonin neurons in different brain regions. A, Top, Immunoblot images of lysates from the indicated brain regions of WT and neuritin KO mice. Bottom, Quantification of the phospho-FGFR1 levels in each homogenate shown in the top panel (n = 6 animals; two-tailed t test with Welch's correction; t(9.8348) = 2.4149 and *p = 0.036 in the mPFC; t(8.3236) = 3.6104 and **p = 0.006 in the hippocampus; and t(9.8689) = 3.2916 and **p = 0.008 in the amygdala). B, Top, Immunoblot images of brain lysates from CMC- and AZD4547-treated mice. Bottom, Quantification of the phospho-FGFR1 levels shown in the top panel (n = 4 animals; two-tailed t test with Welch's correction; t(5.9983) = 3.6516; *p = 0.011). C, Schematic illustration of the experimental design for the inhibition of FGF signaling. AZD4547 (50 mg/kg) or vehicle (1% CMC, 0.9% NaCl, 5% DMSO; 1 ml/kg) was orally administered for four sets of 5 consecutive days at 2 d intervals for a total of 20 d, followed by EPMT, SPT, NSFT, and TST. D, Time spent in the open arms by vehicle (CMC)-treated and AZD4547-treated EMPT mice (n = 14 animals; two-tailed t test with Welch's correction; t(23.2076) = 2.5752; *p = 0.0169). E, Latency time to feed in the NSFT (left; n = 14 animals; two-tailed t test with Welch's correction; t(22.2276) = 3.7419; ***p = 0.0011) and food consumption of home cage feeding, following the NSFT (right; n = 14 animals; two-tailed t test with Welch's correction; t(25.6728) = 1.3447; p = 0.1905). F, Sucrose intake in the SPT of CMC- and AZD4547-treated mice (n = 14 animals; two-tailed t test with Welch's correction; t(25.9522) = 2.6447; *p = 0.0116). G, Immobility time of the TST in CMC- and AZD4547-treated mice (n = 14 animals; two-tailed t test with Welch's correction; t(24.9604) = 2.5753; *p = 0.0163). H, Reconstructed flattened images of the PL, CA3, and BLA from CMC- and AZD4547-treated mice. Immunodetection of serotonin transporter-positive neurons is shown. Scale bar, 100 µm. I, Quantification of the number of branch points along the serotonin transporter-positive axons and axon volume in the images [n = 4/6 region images/mouse; two-tailed t test with Welch's correction; t(7.8441) = 2.8100 and *p = 0.023 in the mPFC; t(5.9624) = 0.5689 and p = 0.5902 in the hippocampus; and t(8.5825) = 3.4783 and **p = 0.007 in the amygdala (number of axonal branches); t(9.0980) = 2.3038 and *p = 0.046 in the mPFC; t(8.4435) = 2.7328 and *p = 0.025 in the hippocampus; and t(9.7989) = 5.1331 and ***p < 0.001 in the amygdala (axonal volume). The data are averaged and presented by mice]. J, Top, Images of dorsal raphe nuclei from CMC-treated and AZD4547-treated mice. Immunodetection of serotonin-positive neurons is shown. Scale bar, 200 µm. Bottom, Quantification of the number of serotonin-positive cells in the dorsal raphe nucleus [n = 3/3 sections/mouse; two-tailed t test with Welch's correction; t(3.2184) = 0.3816; p = 0.727. The data are averaged and presented by mouse experimental group]. K, Top, Immunoblot images of lysates from the indicated brain regions of CMC-treated and AZD-treated mice. Bottom, Quantification of the neuritin levels in each homogenate shown in the top panel (n = 6 animals; two-tailed t test with Welch's correction; t(5.7366) = 2.6420 and *p = 0.040 in the mPFC; t(9.1375) = 2.3643 and *p = 0.042 in the hippocampus; t(9.4551) = 0.3952 and p = 0.701 in the amygdala; and t(8.8334) = 0.7257 and p = 0.487 in the dorsal raphe). L, Schematic illustration of the role of neuritin in FGF signaling. Top, With neuritin, FGFR1 is recruited and remains on the cell surface, which enables FGFR1 to interact with FGF-2 more easily, activate FGF signaling, and induce axonal branching. Bottom, Without neuritin, FGFR1 levels on the cell surface is insufficient. Therefore, the interaction between FGF-2 and FGFR1 is decreased, resulting in a decrease in axonal branch formation. Error bars, SD.
In addition, we compared the axonal branching and density of serotonin neurons in the brain regions. AZD4547-treated mice presented poor axonal branching and axonal density in the mPFC and amygdala (Fig. 8H,I), suggesting that the inhibition of FGF signaling decreased the axonal density of serotonin neurons in the mPFC and amygdala. The hippocampal axons of serotonin neurons were not altered by AZD4547 treatment (Fig. 8H,I), indicating that the FGF signal is not necessary to regulate axonal morphology in the hippocampus. Inhibition of FGF signaling did not change the number of serotonin neurons in the dorsal raphe region (Fig. 8J). Therefore, these results indicate that FGF signaling is required for the axonal branching of serotonin neurons in vivo. In addition, we examined neuritin expression after AZD4547 treatment. Interestingly, the expression levels of the neuritin protein varied by brain region (Fig. 8K). The expression of the neuritin protein decreased in the mPFC and increased in the hippocampus, whereas the neuritin level did not change in the amygdala or dorsal raphe, suggesting that neuritin expression is regulated by FGF signaling in a brain region-dependent manner. For example, neuritin expression and FGF signaling may form a positive feedback loop in the mPFC. Therefore, we cannot exclude the possibility that the reduction in the level of the neuritin protein by AZD4547 is the main mechanism that induces changes in axonal morphology in the mPFC and promotes the development of depression. However, the decrease in neuritin levels attenuated FGF signaling in the mPFC (Fig. 8A). Overall, AZD4547 directly inactivates FGFR1 or indirectly inactivates FGFR1 through a decrease in the level of neuritin, which could be involved in the development of depressive and anxiety behaviors. In addition, because oral administration of an FGF signal inhibitor could affect the whole brain by decreasing the activity of the FGF signaling pathway and global neuritin KO could affect neurons in the whole brain, it is possible that the main targets of FGF and neuritin are brain regions other than the mPFC and amygdala and neurons other than serotonin neurons. However, our results suggest that the inhibition of FGF signaling decreases the axonal branching of serotonin neurons in the mPFC and amygdala, which may be one of the factors involved in the development of depression and anxiety in mice.
Neuritin KO does not affect the expression of FGF-2, whereas stress conditioning reduces FGF-2 production
To examine whether neuritin is involved in the regulatory effect of FGF-2 on the development of depression and anxiety behaviors, we quantified the expression levels of FGF-2 in neuritin KO mice and UCS-conditioned mice. There are three types of FGF-2 with different molecular weights, p18, p22, and p24, and their expression levels may differ slightly during the development of depression and anxiety (Y. Cheng et al., 2015; Wang et al., 2018). In the neuritin KO mice, the expression of FGF-2 in the mPFC, hippocampus, and amygdala was comparable with that in the WT mice, indicating that neuritin may not be involved in the regulation of FGF-2 expression (Fig. 9A). In contrast, UCS-conditioned mice showed lower levels of FGF-2 in brain regions than control mice. Although all types of FGF-2 did not show decreased expression in all regions, one or two types of FGF-2 were clearly reduced by the stress condition in each region (Fig. 9B), probably in a neuritin-independent manner. These results indicate that stress conditioning independently downregulated the expression levels of neuritin and FGF-2 and that a reduction in neuritin weakened the FGF signal. In addition, it is possible that both a decrease in FGF-2 and a decrease in neuritin-mediated FGF signaling induce stress-mediated axonal branch impairment, promoting depression and anxiety in mice.
The expression of FGF-2 is not altered by neuritin KO, whereas UCS conditioning decreases FGF-2 levels in the brain. A, Immunoblot images of lysates from the indicated brain regions of WT and KO mice (left). Quantification of FGF-2 levels in each homogenate shown in the left panel [n = 6 animals; two-tailed t test with Welch's correction; t(9.5067) = 0.8029 and p = 0.442 (p18); t(9.9793) = 0.3943 and p = 0.702 (p22); and t(9.9935) = 0.3881 and p = 0.706 (p24) in the mPFC; t(9.0873) = 0.1160 and p = 0.910 (p18); t(8.1054) = 1.3213 and p = 0.222 (p22); and t(8.9200) = 1.0118 and p = 0.338 (p24) in the hippocampus; and t(7.3899) = 1.9547 and p = 0.089 (p18); t(9.8938) = 0.9786 and p = 0.351 (p22); and t(9.7849) = 0.2906 and p = 0.777 (p24) in the amygdala; right]. B, Immunoblot images of lysates from the indicated brain regions of control and UCS mice (left). The quantification of FGF-2 levels in each homogenate shown in the left panel [n = 6 animals; two-tailed t test with Welch's correction; t(8.4195) = 1.8086 and p = 0.106 (p18); t(5.7998) = 1.2097 and p = 0.273 (p22); and t(7.5382) = 2.7151 and *p = 0.028 (p24) in the mPFC; t(9.5575) = 0.7081 and p = 0.496 (p18); t(9.6633) = 2.9531 and *p = 0.015 (p22); and t(8.5514) = 3.2652 and *p = 0.010 (p24) in the hippocampus; and t(9.7837) = 3.5130 and **p = 0.006 (p18); t(9.2929) = 0.3411 and p = 0.741 (p22); and t(9.9027) = 2.4668 and *p = 0.034 (p24) in the amygdala; bottom]. Error bars, SD.
Discussion
The present study demonstrates that neuritin promotes axonal branching in serotonin neurons, similar to what we have previously shown in granule cells. The addition of purified neuritin enhanced axonal branching, and the loss of neuritin decreased branch formation in cultured serotonin neurons. Immunohistochemical analysis revealed that neuritin-deficient mice exhibited lower axonal density and branching of serotonin neurons in the mPFC and BLA than did WT mice. Pharmacological inhibition of FGFR1 and knockdown of FGFR1 expression abolished neuritin-mediated axonal branching in vitro, indicating that neuritin-mediated axonal branching is an FGF signal-dependent event. In addition, neuritin is necessary for FGF-2-mediated axonal branch formation, suggesting that neuritin and FGF-2 corporately promote axonal arborization. On the other hand, the loss of neuritin expression induced depressive and anxiety behaviors, likely because of the impaired axonal branching of serotonin neurons. Chronic stress on mice decreased the neuritin expression and axonal branching of serotonin neurons, and overexpression of neuritin inhibited the stress-mediated reduction in axonal branching of serotonin neurons. Neuritin overexpression also blocked depressive and anxiety behaviors caused by stress conditioning. Neuritin KO mice showed less phosphorylated FGFR1, and chronic oral administration of FGFR1 inhibitor led to the development of depression and anxiety behaviors in mice. Although we cannot completely exclude the possibility that neuritin regulates neurons other than serotonin neurons to promote behavioral effects, these results suggest that both neuritin and FGF-2 are involved in the regulation of axonal arborization in serotonin neurons and that impaired axonal branching of serotonin neurons can lead to depressive and anxiety behaviors in mice.
Neuritin and downstream signaling regulate neuronal morphology and function
Previous studies have shown that neuritin promotes the arborization of neurites, axons, and dendrites (Nedivi et al., 1998; Javaherian and Cline, 2005; Fujino et al., 2011). The molecular mechanism by which neuritin regulates neural morphology remains elusive, but several studies have shown that neuritin is involved in the activation of insulin receptor, insulin-like growth factor receptors, and FGFR1 (Yao et al., 2012; Shimada et al., 2016; Lee et al., 2021). We previously reported that neuritin interacts with FGFR1 via its Ig2 and Ig3 domains and that neuritin recruits FGFR1 to the axonal cell surface, resulting in the promotion of axon branching. FGF also recruits FGFR1 to the cell surface, but loss of neuritin abolishes the FGF-dependent recruitment of FGFR1 (Shimada et al., 2016). Thus, neuritin can recruit FGFR1 to the cell surface and retain it there, resulting in increased FGFR1 availability to bind FGF-2 and activate FGF signaling. In the absence of neuritin, FGFR1 activation is insufficient even in the presence of FGF. Therefore, we assume that the protein level of neuritin can regulate axonal branch formation in serotonin neurons (Fig. 8L).
Axonal branching of serotonin neurons in the brain and the development of depression
Animal studies have shown that exposure to chronic stress decreases the neuritin mRNA translation level in the hippocampus (Son et al., 2012; Sathyanesan et al., 2018), while we have shown that stress exposure reduced the protein level of neuritin in the mPFC, hippocampus, and amygdala. The brain regions analyzed in this study, mPFC, hippocampus, and amygdala, are regions involved in stress-mediated behavioral alterations. The mPFC, hippocampus, and amygdala play various roles in regulating the stress response via the hypothalamic‒pituitary‒adrenal axis (Ulrich-Lai and Herman, 2009). Chronic restraint stress shortens dendritic length and decreases dendritic branches in the mPFC (Radley et al., 2004), and prolonged and severe stress induces the shortening and debranching of dendrites in hippocampal neurons (Conrad et al., 1999). Neuritin has been shown to be involved in the dendritic atrophy of granule cells in the hippocampus mediated by chronic stress (Son et al., 2012). We observed that neuritin regulated axonal branch formation of serotonin neurons in the mPFC and amygdala. Because the amygdala contributes brain functions involving emotions, neural alterations in the amygdala are likely to be involved in symptoms of stress disorders. In rodents, severe stress facilitates fear and anxiety-like behavior, and the BLA has been shown to be essential for stress-induced facilitation of aversive learning (Shors and Mathew, 1998). Repetitive stress controls dendritic morphology and spine density in the amygdala; however, there are differences in the effect of stress on different regions of the amygdala. Recent studies have shown that the innervation of serotonin axons in the amygdala is altered under conditions inducing depression. Ischemia-induced mice showed poststroke depression (PSD; Vahid-Ansari et al., 2016). Stroke targeting the left mPFC reduced the density of serotonin neuron axons in the left mPFC and the ipsilateral BLA, whereas axonal density of serotonin neuron was not altered in the hippocampus (Zahrai et al., 2020). In addition, postweaning isolation rearing conditions promote anxiety in rats (Walker et al., 2019), and the density of serotonin-immunopositive axon was significantly lower in the amygdala in isolated rats (Kuramochi and Nakamura, 2009). Antidepressant drug treatment rescues the axonal loss of serotonin neurons in the amygdala (Zahrai et al., 2020), suggesting that poor axonal branching of serotonin neurons in the amygdala is involved in the development of depression and anxiety. A stress-mediated decrease in neuritin is likely involved in these phenotypes in the amygdala.
Updated serotonin hypothesis
Originally, low levels of serotonin in the brain were thought to contribute to the development of depressive disorders, and treatment to increase serotonin levels was thought to improve symptoms of depressive disorders; however, a recent report suggested that simply increasing serotonin levels using SSRI treatment does not directly lead to improvements in depressive symptoms (Lacasse and Leo, 2005). The major question is why behavioral improvement takes weeks to observe, despite brain levels of serotonin increasing within hours after SSRI administration. This delay might reflect adaptive changes in pre- and postsynaptic neurons, including long-term changes in gene expression, protein translation, and synaptic plasticity (Krishnan and Nestler, 2008; Vahid-Ansari et al., 2019). Therefore, the serotonin level in the brain may not directly regulate symptoms of depressive disorders, and researchers have focused on determining the role of increased serotonin levels in regulating neuronal morphology and synaptic transmission.
The degree of the serotonin neuron axon innervation is one of the explanations for the delayed effect of SSRIs on behavioral recovery. In the postmortem human brain of depression patients, axonal staining of serotonin neurons in the deep layer of the prefrontal cortex was shown to be reduced, and the density of serotonin neurons in the orbitofrontal cortex has been shown to be reduced in an age-dependent manner in major depressive disorder patients (Austin et al., 2002; Rajkowska et al., 2017). Positron emission tomography analysis investigating the association between pretreatment SERT-binding ratios and treatment responses against depression provided data indicating that antidepressant (escitalopram and citalopram) treatment led to an increase in serotonin axon innervation in the living human brain (Lanzenberger et al., 2012). These studies indicate that a deficiency in serotonin neuron innervation occurs in major depression and that this deficiency can be modified by chronic treatment with SSRIs. Neuritin-mediated axonal innervation of serotonin neurons may participate in this process of regulating depressive behavior. However, it is still unknown whether SSRI treatment controls the expression level of neuritin in the amygdala.
FGF signaling and depression
Many studies have shown that FGF-2 levels are correlated with depression, and activation of FGF-2 signaling is another axis for ameliorating depression. Social defeat-conditioned rats and PSD rats presented decreased mRNA levels of FGF-2 (Turner et al., 2008; Ji et al., 2014). Antidepressant drug treatment increased the mRNA and protein levels of FGF-2 in the mouse and rat brain (Maragnoli et al., 2004; Bachis et al., 2008; J. Cheng et al., 2019, 2021), and antidepressant treatment did not rescue stress-mediated depression in FGF-2 KO mice (Simard et al., 2018), indicating that FGF-2 functions as a mediator of antidepressant effects and that FGF signaling is required for proper antidepressant function. Furthermore, postmortem studies have shown that the mRNA levels of FGF-2 in the brains of depressive patients are relatively lower than those in the brains of healthy controls (Evans et al., 2004; Gaughran et al., 2006). In this study, we have shown that stress conditioning leads to a decrease in FGF-2 expression in various brain regions. However, the molecular mechanism by which the loss of FGF-2 induces depression is still unclear. Our data suggest that UCS conditioning decreases FGF-2 expression in the brain and that neuritin and FGF-2 cooperate to regulate the axon branching of serotonin neurons in vitro. Therefore, neuritin could be involved in the antidepressant function of FGF-2, and one of the functions of FGF-2 may be the regulation of the axonal branch formation of serotonin neurons. Recent studies revealed that FGFR1 and the serotonin receptor 5-HT1A form a heteroreceptor complex and lead to a marked synergistic increase in FGF signaling in the midbrain raphe (Borroto-Escuela et al., 2015; Borroto-Escuela et al., 2016), indicating the involvement of FGF-2 in the development of depression. Further analysis will reveal whether neuritin may participate in the formation of heterocomplexes by interacting with FGFR1, which can regulate FGF signaling.
Footnotes
- Received January 23, 2023.
- Revision received August 19, 2024.
- Accepted August 21, 2024.
This work was supported in part by Japan Society for the Promotion of Science KAKENHI Grants-in-Aid for Scientific Research 18H02536 and 22K07949 (K.Y.) and 17K07086 and 22K06493 (T.S.). We thank Drs. Akiyo Natsubori and Ran Inoue for their critical instructions concerning virus injection, Dr. Hiroshi Sakuma for the helpful discussion, and Ms. Fumie Masuda and Ms. Nobuko Ogawa for their support in the quantification and genotyping methods.
The authors declare no competing financial interests.
- Correspondence should be addressed to Tadayuki Shimada at shimada-td{at}igakuken.or.jp.
- Copyright © 2024 the authors