|Home | About | Journals | Submit | Contact Us | Français|
In the CNS, serotonin, an important neurotransmitter and trophic factor, is synthesized by both mast cells and neurons. Mast cells, like other immune cells, are born in the bone marrow and migrate to many tissues. We show that they are resident in the mouse brain throughout development and adulthood. Measurements based on capillary electrophoresis with native fluorescence detection indicate that a significant contribution of serotonin to the hippocampal milieu is associated with mast cell activation. Compared to their littermates, mast cell deficient C57BL/6 KitW-sh/W-sh mice have profound deficits in hippocampus-dependent spatial learning and memory and in hippocampal neurogenesis. These deficits are associated with a reduction in cell proliferation and in immature neurons in the dentate gyrus, but not in the subventricular zone – a neurogenic niche lacking mast cells. Chronic treatment with fluoxetine, a selective serotonin reuptake inhibitor, reverses the deficit in hippocampal neurogenesis in mast cell deficient mice. In summary, the present studies demonstrate that mast cells are a source of serotonin, that mast cell deficient C57BL/6 KitW-sh/W-sh mice have disrupted hippocampus-dependent behavior and neurogenesis, and that elevating serotonin in these mice, by treatment with fluoxetine, reverses these deficits. We conclude that mast cells contribute to behavioral and physiological functions of the hippocampus and note that they play a physiological role in neuroimmune interactions, even in the absence of inflammatory responses.
Serotonin is implicated in hippocampal function during development and adulthood, both as a neurotransmitter and a trophic factor (Lauder and Krebs, 1978; Altman and Normile, 1988). The best known sources of hippocampal serotonin in the brain are the median raphe nuclei (Zhou and Azmitia, 1983) while resident mast cells are another, less studied source (Kushnir-Sukhov et al., 2007). Like other immune cells, mast cells are born in the bone marrow. They are found in the brain of rat, mouse, dove, voles, and human, with interspecies differences in their precise localization and numbers (Dropp, 1976, 1979; Persinger, 1979; Goldschmidt et al., 1985; Zhuang et al., 1993; Kriegsfeld et al., 2003). Whether or not serotonin of mast cell origin participates in hippocampal function in development or adulthood has not been examined.
It is known that mast cells can synthesize and store serotonin (Marathias et al., 1991; Kushnir-Sukhov et al., 2007; Ringvall et al., 2008). The specific mediator content of mast cell granules however, depends on the local microenvironment in which they reside (reviewed in Marshall, 2004). Serotonergic deafferentation of the hippocampus through ablation of serotonergic neurons results in a 60-80% decrease in serotonin in the hippocampus (Altman et al., 1990), suggesting that as much as 20-40% of serotonin may originate from mast cells. Determination of the role of this immune cell in hippocampal physiology and function is especially interesting given evidence for other immune system effects on an array of hippocampal functions (Yirmiya and Goshen, 2011).
The hippocampus is important in the regulation of anxiety and depressive behaviors as well as in spatial learning and memory (Scoville and Milner, 1957; Gray and McNaughton, 1983; Santarelli et al., 2003). Depletion of serotonin during development has profound effects on the formation of mature hippocampal synapses (Mazer et al., 1997), on neurogenesis, apoptosis and cell differentiation (Lauder, 1990; Yan et al., 1997; Gaspar et al., 2003). Disruption of hippocampal serotonergic signaling during development results in profound abnormalities in affective behavior and memory in adulthood (Mazer et al., 1997; Gaspar et al., 2003). In adulthood, increases in serotonin promote hippocampal neurogenesis (Gould, 1999), and also affects synapse formation and dendritic plasticity (Watanabe et al., 1992; Mazer et al., 1997). Finally, increased serotonergic signaling in adulthood has been implicated in the behavioral effects of antidepressants (Santarelli et al., 2003; Sahay and Hen, 2007).
Our previous work with mast cell deficient. B6.Cg-KitW-sh/HNihrJaeBsmJ (Wsh/Wsh) mice suggests that brain mast cells are involved in behavioral and physiological components of anxiety-like behavior (Nautiyal et al., 2008). In the present studies we use these mice to explore the contribution of mast cells to the hippocampus and to its behavioral and physiological functions. Upon application of a mast cell degranulating agent, serotonin levels within hippocampi of control mice with mast cells were ~50% higher than in their mast cell deficient Wsh/Wsh littermates. Mast cell deficient Wsh/Wsh mice also have decreased hippocampal neurogenesis and marked deficits in spatial learning and memory. Chronic treatment with a selective serotonin reuptake inhibitor (SSRI) reverses the deficit in neurogenesis in Wsh/Wsh mice. The results indicate that mast cells and their chemical mediators may contribute to behavioral and physiological aspects of hippocampal function, even in the absence of inflammatory responses or disease states.
C57BL/6 wild-type (WT) and mast cell deficient Wsh/Wsh mice (B6.Cg-KitW-sh/HNihrJaeBsmJ, C57BL/6 background) were purchased from Jackson Laboratories (Bar Harbor, ME) and bred to establish colonies at Columbia University animal facilities. The Wsh/Wsh mice were crossed with WT C57BL/6 mice, to generate heterozygote (Wsh/+) mice which have brain mast cells. Wsh/Wsh mice have a naturally occurring mutation in the white spotting locus (W) which causes reduced c-kit receptor expression (Duttlinger et al., 1993) and results in abnormalities in pigmentation and an inability of mast cells to differentiate from their myeloid progenitors (Kitamura and Fujita, 1989; Chen et al., 2005). Wsh/Wsh mice appear normal in the numbers of other hematopoietic cells (Grimbaldeston et al., 2005; Galli et al., 2008), and in locomotor activity (Nautiyal et al., 2008). All animal care and testing was approved by the Columbia University Institutional Animal Care and Use Committee.
For studies of mast cell location in the mouse brain, WT mouse pups (n=5/age) were sacrificed at postnatal age (PN) 1, 5, 9, 12, 18, 21 or in adulthood (38 days). An additional group of WT mice were sacrificed at ~2 months of age (n=3) to localize mast cells in coronal sections with respect to the hippocampus. Pups aged PN1-9 were sacrificed by rapid decapitation. Mice PN 12 and older were deeply anesthetized with sodium pentobarbital (200mg/kg) and perfused transcardially with saline followed by 4% paraformaldehyde.
Homozygote Wsh/Wsh and heterozygote Wsh/+ male mice (at 12-18 wks) from Wsh/Wsh × Wsh/+ crosses were used in serotonin, behavior, neurogenesis and corticosterone experiments. Animals for neurogenesis experiments studies were anesthetized with sodium pentobarbital and sacrificed by transcardial perfusion as above. Animals used in brain serotonin and plasma corticosterone assays were sacrificed by rapid decapitation. Trunk blood was collected and brains were rapidly removed from crania and placed in ice cold Hank’s balanced salt solution (HBSS; Invitrogen, Carlsbad, CA).
Following weaning at 28 days, mice were housed 2-5/cage in transparent plastic bins (36×20×20 cm) in a 12:12 light-dark cycle at 22 ± 1°C. Cages had corn cob bedding (Bed-o’cobs, Maumee, OH) and food and water were provided ad libitum.
5-Bromo-2’-Deoxyuridine (BrdU; Sigma Aldrich, St. Lois, MO) was used to label dividing cells. For assessment of cell proliferation and survival in the dentate gyrus (DG), Wsh/+ and Wsh/Wsh mice were injected once daily for 3 days with BrdU at a dose of 150 mg/kg i.p. (concentration 7.5 mg/ml saline) and sacrificed 24 h following the 3rd injection. For assessment of cell survival in dentate gyrus and olfactory bulb, mice were sacrificed 28 days following the 3rd injection. For assessment of cell proliferation in the subventricular zone (SVZ) and following SSRI administration, mice were injected once with BrdU (150 mg/kg, i.p.) and sacrificed 24 h later.
Wsh/+ and Wsh/Wsh mice were given once daily i.p. injections of fluoxetine hydrochloride at a dose of 10 mg/kg/day (concentration of 1mg/ml saline; Biotrend AG, Zurich) or saline vehicle. Following 28 days of treatment, mice were tested in the open field and novelty suppressed feeding tests. One week later mice were injected with BrdU.
As seen in Figure 1, mast cells lie within and near the hippocampus. Upon activation, mast cell granules release their contents over a substantial volume of tissue (Ruoss et al., 1991; Marszalek et al., 1997; Nautiyal et al., 2008). Thus, following degranulation, granular contents from mast cells lying both within and nearby the hippocampus can be measured. To measure the contribution of mast cells to the hippocampal milieu, coronal slices were cut on a vibratome (300 μm) in ice cold HBSS. The caudal hippocampal slice (bregma - 2.92 mm) was chosen as this is a mast cell rich region. The rostral hippocampal slice (bregma – 1.70 mm) served as control as this region bears few mast cells.
To measure the contribution of mediators of mast cell origin, degranulation was induced by applying 400 μl of Compound 48/80 (C48/80; Sigma Aldrich, St. Louis MI) at a concentration of 0.1 mg/ml in HBSS, while the mast cells contribution at baseline was examined after application of HBSS alone. The brain slice was incubated with C48/80 or the HBSS vehicle for 3 min. Next, the hippocampus was dissected out and placed into Eppendorf tubes, weighed, and manually homogenized in 20 μl of extraction matrix [49.5% LC-MS Chromasolve Methanol (Sigma Aldrich), 49.5% LC/MS grade water (Thermo Scientific, Waltham, MA), 1% acetic acid (Sigma Aldrich)] per mg of dissected tissue. Homogenized tissue was incubated with extraction matrix for 90 min at 4°C before centrifugation at 15,000 g for 15 min. The supernatants were then transferred to PCR tubes and frozen at -80 °C until analysis by capillary electrophoresis with laser induced native fluorescence (CE-LINF).
Each sample was analyzed in triplicate using a laboratory-built CE-LINF instrument equipped with a wavelength-resolved detector similar to that previously described (Fuller et al., 1998; Squires et al., 2007). Briefly, 10 nl of the supernatant was hydrodynamically injected into a fused silica capillary and 21 kV applied for separation. Eluting analytes were excited via 264 nm radiation from a frequency-doubled argon ion laser. The native fluorescence was collected and the spectrum imaged onto a CCD array where it was recorded at a rate of 2 Hz. The peaks resulting from the elution of serotonin were identified by comparison of both retention time and fluorescence characteristics to a standard of serotonin (Sigma Aldrich, St. Louis MO) spiked into extraction matrix. For quantification, the total integrated signal was calculated from the electropherograms by comparison to a linear working curve derived from 5 serotonin standards of known concentration (15-250 nM) and was then normalized to the weight of dissected hippocampal tissue. Outlying data points were removed based on the results of the Q-test. If either of the two sections of hippocampus were found to be a statistical outlier the corresponding data for the remaining section of hippocampus was removed. Differences in the populations were evaluated by the Kruskal-Wallis non-parametric analysis using Origin 8.0 (Kruskal and Wallis, 1952). We also measured Tyr, and tested for the presence of several serotonin-related catabolites [HIAA, serotonin sulfate, HITCA, NAS: see (Squires et al., 2007)]. These catabolites were below the instrument’s limit of detections and are therefore not reported here.
Mice were sacrificed by rapid decapitation within 2 h following lights on and trunk blood was collected into EDTA coated tubes (BD, Franklin Lakes, NJ) and spun down at 4°C at 1400 g for 15 min. Plasma was collected and stored at -80°C until processed. Plasma corticosterone was assayed using a commercially available EIA kit (Cayman Chemical, Ann Arbor, MI) according to the manufacturer’s directions. Samples were run in duplicate at two dilutions with resulting concentrations falling on the linear portion of the standard curve. The coefficient of variation between duplicates was less than 7%, otherwise the samples were rerun.
For mast cell localization during development, the crania were cleaned of skin, muscles and lower jaw structure and post-fixed in 4% paraformaldehyde overnight. Post-fixed crania were placed in a mild formic acid bone decalcifier (Immunocal, Tallman, NY) for 24 h, then washed in water and phosphate buffer (pH 7.3) and then gelatin embedded. Brains were cut into 50 μm-thick sagittal (for developmental study) or coronal (for adult hippocampal localization) sections on a cryostat (Microm HM 500M, Walldorf). Sections were thaw-mounted onto slides (Probe-on-plus slides, FisherBrand, Waltham, MA), and stored at -80 °C until processed.
Slide mounted sections were stained with the acidic (pH 2.0) toluidine blue (TB) stain (Sigma, St. Louis, MO) a commonly used marker for mast cells reported in both our work (e.g. Silverman et al., 1994; Zhuang et al., 1999; Wilhelm et al., 2000) and by many other labs (e.g. Florenzano et al., 2000; Esposito et al, 2001; Menétrey et al., 2003). Acidic toluidine has metachromatic properties and color changes to a deep blue/purple in tissues with acidic polyanions such as the proteogylcans and heparin found in mast cell granules (Presnell and Schreibman, 1997). Confirmation of the mast cell identity of TB stained cells has previously been confirmed in the brain by double labeling with alcian and/or safrinin and based on ultra-structural evidence using electron microscopy (Silverman et al., 1994; Zhuang et al., 1999; Yang et al., 1999). In the present study, for TB staining, slides were washed in 60% EtOH for 2 min, before stained with TB (4 mg/ml TB in 60% EtOH, pH 2.0), washed briefly in distilled water, then dehydrated through a series of ethanols (50% for 15s, 70% for 45s, 2 × 95% for 1min, 2 × 100% for 1 min) then cleared in Citrasolv (3 × 5 min) and coverslipped with Permount.
To confirm the identity of the mast cells, additional slide mounted sections were initially stained with avidin which labels mast cells by binding to the heparin found in mast cells (Bussolati and Gugliotta, 1983; Tharp et al., 1985; Presnell and Schreibman, 1997). Next, the sections were photographed, then the coverslip was removed, and the sections were re-stained with TB (as described above) to show double-labeling of brain mast cells with avidin and TB. For avidin staining, slides were washed in phosphate buffer (PB) with 0.1% triton X (PBT), then incubated in PBT with Cy3-conjugated egg-white avidin (1:5000; Jackson ImmunoResearch, West Grove, PA) for 2h. Slides were then washed in PB, dehydrated, cleared n Xylenes and coverslipped with Krystalon.
Every fourth section through the brain was analyzed by two observers for the number and location of mast cells using a light microscope (BH-2, Olympus America Inc. Center Valley, PA; inter-rater reliability: r2>0.95). Mast cells were identified by their deep purple granules against the light blue background Nissl stain. The location of each mast cell was mapped using a brain atlas (Paxinos and Franklin, 2004). Mast cells were localized to the following regions: olfactory bulb, cortex, hippocampus, amygdala, thalamus, hypothalamus, and cerebellum, and included mast cells located within the parenchyma as well as those in the perivascular spaces, choroid plexus and meningeal layers.
Perfused brains of animals injected with BrdU were post-fixed overnight, then submerged in sucrose for 48 h, then cut into 50 μm coronal sections on a cryostat (Microm HM 500M, Walldorf) into phosphate buffer. Every eighth section of hippocampus and olfactory bulb and every fourth section of the SVZ was processed by free-floating immunohistochemistry as previously described (Meshi et al., 2006). In brief, sections were washed in tris-buffered saline (TBS), denatured in 50% Formamide/2×SSC at 60 °C, washed in SSC, then in HCl at 37 °C and Boric Acid, then again in TBS. Sections were incubated for 36 h at 4 °C, in primary antibodies against one or two of the following antigens: BrdU [AbD Serotec Cat#OBT0030G, Raleigh, NC, 1:200], doublecortin (DCX) [Santa Cruz Biotechnology Cat#sc-8066, Santa Cruz, CA, 1:1000], neuronal-specific nucleus protein (NeuN) [Chemicon Cat#MAB377, Billerica, ME, 1:500]. Sections were then washed and incubated in the appropriate fluorescent secondary antibodies – donkey anti-rat secondary conjugated to CY3 for BrdU, donkey anti-goat conjugated to CY2 for DCX, and donkey anti-mouse conjugated to CY2 for NeuN (Jackson ImmunoResearch, 1:200) for 2 h, then washed, mounted, dehydrated and coverslipped.
Perfused mouse brains were post-fixed overnight, then submerged in sucrose for 48 h, then cut into 50 μm coronal sections on a cryostat (Microm HM 500M, Walldorf) into phosphate buffer. Every eighth free-floating section of hippocampus and cerebellum was processed for double-labeled immunohistochemistry against c-kit and either GFAP or DCX. In brief, sections were washed in phosphate buffer (PB) with 0.1% triton X (PBT), blocked in normal donkey serum, then incubated overnight in rabbit anti-c-kit receptor primary antibody (Ab-1, Oncogene Research Products, Cat# PC-34, Boston, MA, 1:500) and either goat anti-GFAP (C-19, Santa Cruz Biotechnology, Cat# sc-6170, 1:250) or goat anti-DCX (as above). Sections were then washed then incubated in PBT with fluorescent secondary antibodies – donkey anti-rabbit secondary conjugated to CY3 for c-kit, donkey anti-goat conjugated to CY2 for GFAP and DCX for 2 h. Slides were then washed in PB and coverslipped with ProLong Gold (Invitrogen, Grand Island, NY).
Images of single labeled sections were captured using a Nikon Eclipse E800 light microscope (Nikon Co., Tokyo, Japan) equipped with the GFP (480±20 nm) or Texas Red (560±40 nm) filters to detect CY2 or CY3 fluorophores, respectively. The microscope is fitted with a Q-Imaging Retiga EXi, fast 1394 camera (Quantitative Imaging) with QCapture software (version 2.95.0, Quantitative Imaging, Surrey, BC, Canada). Images were loaded into Photoshop 7.0 (Adobe Systems, San Jose, CA) for cell counting.
BrdU+ cells were counted by two observers blind to genotype and condition (inter-rater reliability: r2>0.95). Sampling of BrdU+ cells was performed as follows: 6 sections/animal in DG along the subgranular layer; 4 sections/animal in SVZ adjacent to the lateral ventricle olfactory bulb; 4 sections per animal, scored in 3 representative areas (0.1mm × 0.1mm) randomly placed in the granule cell layer of the olfactory bulb. Area of staining, rather than cell counting was used to quantify the DCX staining in 6 sections of the hippocampus in every animal in ImageJ 1.31 with the Otsu method to determine the threshold of positive staining (Otsu, 1979).
Double labeling of cells with BrdU and NeuN in the granule cell layer was assessed by first counting the number of BrdU cells as described above in Photoshop (Adobe, San Jose, CA). Then the number of cells labeled with both BrdU and NeuN were counted amongst the population of BrdU+ cells. Confocal microscopy was used confirm double labeled BrdU and NeuN cells. Specifically, bilateral hippocampus from one representative section (bregma - 2.2) per animal was scanned in with the Zeiss Axiovert 200 MOT fluorescence microscope (Carl Zeiss, Thornwood, NY, USA) with Zeiss LSM 510 laser scanning confocal attachment and LSM 3.95 software (Carl Zeiss). Cells within the captured images were examined in 2 μm optical slices using the LSM 3.95 software for confirmation of double labeling. The percentage of double labeled cells achieved with both methods was not different (r2>0.95, p<0.01).
The volume of the dentate gyrus was estimated using the Cavalieri method (Gundersen and Jensen, 1987). Briefly, the area of the NeuN+ dentate gyrus was measured in 6 sections of the hippocampus in each animal in ImageJ 1.31 with the Otsu method to determine the threshold of positive staining. The volume of the hippocampus was calculated by multiplying the area of the hippocampal staining in each section obtained from ImageJ (in μm2) by the thickness of the section plus the distance between each section (50 μm+150 μm). The total volume was calculated by summing this volume across the six sections.
For assessment of spatial learning and memory, Wsh/+ and Wsh/Wsh mice were tested in the radial arm maze. Animals were weighed once per day and kept on a restricted feeding paradigm to maintain their body weight at 80-85% of their free-feeding weight. The radial arm maze was made of black plexiglass with 8 arms extending 50 cm from the center, with 16cm high walls. Racks and other behavioral testing equipment in the room served as spatial cues. Mice were tested twice daily in the maze (2 h and 8 h after lights on). Mice were placed into the center of the radial arm maze restricted to a plexiglass cylinder for 15 s. The cylinder was then lifted and the session began, lasting 10 min or until all rewards were retrieved and eaten. Each session was videotaped and the maze was wiped down with 70% ethanol following each animal. Mice were given three habituation sessions during which pieces of Fruit Loops ® (Kellog’s) were scattered throughout the maze and then placed at the end of each arm. Mice were given 20 training trials in which 3 arms were baited with fruit loop pieces, followed by a reversal trial in which 3 different arms were baited. Each arm entered, latency to enter the first arm and latency to retrieve bait were measured. An entry into an arm was scored as an error if the mouse had already entered that arm during the trial (working memory error), or if the arm did not contain a reward (reference memory error). Other behaviors recorded include grooming, rearing, investigation of potential food location (weigh-boat), and scratching/biting weigh-boat. None of these behaviors were significantly different between genotypes and are therefore not reported here.
For further assessment of hippocampus-dependent spatial learning, mice were tested in the Morris water maze (Morris, 1984). A 150 cm diameter round pool was filled with water (25°C, 10 cm deep) and made opaque by the addition of non-toxic white tempera paint. Four black and white patterns (small checked, large vertical stripes, small horizontal stripes, large circle) on 8½×11” paper were placed on the inside wall of the Morris water maze to serve as spatial cues. Mice were subject to 4 trials daily, for 9 days. Mice were videotaped during each trial (Sony Handycam) and CleverSys software was used for analysis of swim path and time spent in each quadrant.
Mice were placed in the pool at each of 4 drop points (N, E, S, W) randomized between days, but kept consistent for all animals within a given day. A basic black and white patterned cue was placed at each of the 4 drop points above the water line to serve as spatial cues. A plexiglass platform (10×10cm) was weighted down and placed in the center of the SE quadrant. Time to reach the platform was recorded for each trial (escape latency) in seconds. Mice were allowed 90 s to reach the platform, on which they had to remain for 5s before being removed by the experimenter. If a mouse did not reach the platform within the 90 s, a latency of 90 s was assigned and the mouse was gently guided by the experimenter to the platform before being removed from the water. A visible platform was used on the first day for shaping trials. It was elevated 1cm from the water surface and clearly marked with a flag. For hidden platform trials, the platform was 1 cm below the water. On the 6th day, a single probe trial was run; the platform was removed from the pool and the swim path and time spent in the target quadrant was analyzed. On days 7-9, visible platform trials were run to assess non-spatial deficits such as swimming performance or sensory perception. Mice were always placed in the pool at the E drop point, and the location of the platform placement was varied among the center of each of the quadrants.
Two-tailed Student’s t-tests were used to compare hippocampal volume measurements, BrdU+ cell counts, DCX staining area and plasma corticosterone levels between genotypes. If the test for normality failed, a Mann-Whitney Rank Sum Test was run. A chi-square test was used to test differences in cell survival between genotypes. Two-way ANOVAs were used to assess effects of genotype and trial in behavioral experiments, and also effects of genotype and treatment in the SSRI experiment. Tukey’s HSD was used for post hoc testing when appropriate.
The number and localization of intracranial mast cells was studied throughout development and in adulthood (Fig 1). In mice, mast cells are located in the meninges and perivascular spaces on the brain side of the blood-brain barrier predominantly in thalamic and hippocampal brain regions (Hendrix et al, 2006; Kovács & Larson, 2006; Nautiyal et al., 2011). Brain maps, seen in sagittal (Fig 1A) and coronal (Fig 1B) views reveal mast cells around the hippocampus, where they lie predominantly in more caudal areas (bregma - 3.08). The total number of intracranial mast cells increases steadily after birth, reaching adult levels around PN21 (F6,30=7.4, p<0.01; Fig 1C). The total number of mast cells located within the parenchyma of the hippocampus and surrounding leptomeninges remains stable over this interval, with a gradual shift from the former to the latter compartment (Fig 1D). Thus, a larger proportion of total brain mast cells are located in and around the hippocampus during early postnatal development, peaking at around 25% at PN5 (Fig 1E).
Mast cells, stained with both toluidine blue and avidin, are seen within the brain (Fig 2A). Mast cells are located in the parenchyma of the hippocampus as well as the leptomeninges and choroid plexus adjacent to the hippocampus (Fig 2B). These results, together with previous work showing altered anxiety-like behavior in mast cell deficient mice (Nautiyal et al., 2008), led us to focus on mast cell functions in the hippocampus.
To investigate mast cell contribution of serotonin to the hippocampus, C48/80, a mast cell degranulating agent or control HBSS solution was applied to brain slices containing the caudal or the rostral hippocampus. The total serotonin content was subsequently measured in the dissected hippocampal parenchyma at two rostral-caudal levels using CE-LINF detection. Our laboratory built instrument allows for the trace analysis of natively fluorescence analytes from volume-limited samples, including indoles in the hippocampus (Fuller et al., 1998; Hatcher et al., 2008). Analysis of both rostral and caudal regions of the hippocampus in mast cell deficient Wsh/Wsh and mast cell competent Wsh/+ mice revealed that mast cells contribute significantly to the serotonin content in the caudal, but not rostral hippocampus in Wsh/+ mice (Fig 3).
In the rostral hippocampus, there were no significant differences between Wsh/+ (n=4) and Wsh/Wsh (n=5) mice in baseline levels of serotonin, measured following application of HBSS vehicle (H1=0.00, p>0.05; Fig 3A). Additionally, following application of C48/80, there were no differences in serotonin content in the rostral hippocampus between Wsh/+ (n=7) and Wsh/Wsh (n=7) mice (H1=0.10, p>0.05, Fig 3B).
In the caudal hippocampus, there were no significant differences between Wsh/+ and Wsh/Wsh mice in baseline levels of serotonin measured in HBSS vehicle condition (H1=0.02, p>0.05; Fig 3A). In contrast, following application of C48/80 to slices, serotonin content in Wsh/+ mice (n=7) was increased compared to their HBSS baseline and to that of Wsh/Wsh (n=8) mice (H1=6.48, p<0.05). Approximately 50% more serotonin was measured in Wsh/+ mice compared to Wsh/Wsh mice in C48/80 stimulated caudal hippocampus. This represents a dynamic mast cell-mediated increase in serotonin in the hippocampus. Interestingly this increase corresponds to the larger population of mast cells resident near the caudal vs. rostral hippocampus in adulthood and indicates that degranulation of mast cells in brain tissue contributes measurable levels of this neurotransmitter to the hippocampal milieu.
In measuring serotonin with our CE-LINF instrument, tyrosine (not found in mast cells) is also measured. As expected, levels in Wsh/+ and Wsh/Wsh animals did not change with mast cell stimulation by C48/80 (data not shown).
Wsh/Wsh mice showed deficits in hippocampus dependent spatial learning and memory in the radial arm maze (Fig 4A). Both Wsh/Wsh (n=4) and Wsh/+ (n=5) littermates reduced the number of errors made over the 5 training blocks suggesting that they learned to locate the rewards (F5,35=12.42, p<0.05). There were no significant differences between Wsh/+ and Wsh/Wsh mice in the number of errors made during training trials (F1,35=0.06, p>0.05), nor in the total number of errors made in the last training block (4.2±0.5 and 3.9±0.5, respectively; t7=0.71, p>0.05). However, on the reversal trial, the total number of errors increased from the last training block in both genotypes (t8=2.21, p=0.057 for Wsh/+ and t6=0.96, p>0.05 for Wsh/Wsh mice), with the Wsh/Wsh making fewer total errors than their littermates during the reversal trial (4.3±0.3 vs. 7.2±1.3: F1,14=4.31, p=0.057). Interestingly, in Wsh/Wsh mice, there was no increase in the number of reference memory errors (entries into unbaited arms) in the reversal trial, indicating that they did not learn the location of the rewards based on spatial cues. Rather, their seemingly successful performance is due to a learned non-spatial search strategy, in which working memory errors (re-entries into arms) are low (Hodges, 1996). In keeping with this idea, analysis of the pattern of arm entries in Wsh/Wsh mice reveals a serial search pattern (Fig 4B). Serial search is defined as entering successive arms in a clockwise or counter clockwise fashion during a single trial. Over 21 trials (training+reversal), Wsh/Wsh mice performed an exact match serial search 30.9% of the time while control mice did so only 1.9% of the time (t7=2.43, p<0.05) supporting the suggestion that Wsh/Wsh mice were not learning the location of the rewards.
Wsh/Wsh mice also showed impaired learning and memory in the Morris water maze test of hippocampus dependent spatial learning (Fig 4C). Compared to Wsh/+ mice (n=9), Wsh/Wsh mice (n=8) took longer to reach the hidden platform (Fig 4C; F1,59=5.61, p<0.05). By the 5th day, Wsh/+ mice reached the platform in 13±5s while Wsh/Wsh mice took 31±6s (t15=2.04, p=0.059). Collapsed across both genotypes, the escape latency decreased over days (main effect of time; F4,59=17.05, p<0.01). There was no significant difference between genotypes in escape latency in the visible platform trials indicating competence in motor and sensory systems in Wsh/Wsh mice. Performance during the probe trial indicates impairments in spatial learning and memory in Wsh/Wsh mice. There is a difference between genotypes in the amount of time spent in each quadrant (Wsh/+ vs. Wsh/Wsh: t15=2.45, p<0.05; Fig 4D). Wsh/+ mice spent significantly more of their time in the target quadrant compared to the other three (42%; F3,32=20.81, p<0.01). In contrast, Wsh/Wsh mice spent only 28% of their time in the target quadrant, the same as the amount of time spent in any other quadrant (F3.28=0.94, p>0.05; Fig 4E) indicating a lack of learning and/or memory of the location of the platform.
The volume of the granule cell layer (GCL) of the DG in Wsh/Wsh mice (n=8) was reduced by 12.9% compared to littermate Wsh/+ controls (n=10; t16=3.22; p<0.01). There was no difference in the volume of CA1-CA3 regions (t16=0.94; p>0.05; Table 1).
Cell proliferation was assessed by incorporation of BrdU into dividing cells (Fig 5A). There was a 43.5% reduction in the number of proliferating cells in the GCL of the DG in Wsh/Wsh mice (n=9) compared to Wsh/+ littermates (n=12) over several rostral-caudal levels (75.2±15.0 vs. 133.1±14.7; t19=2.73, p<0.05; Fig 5B, C). There were no significant effects of genotype on cell proliferation in another neurogenic niche, the SVZ, where no mast cells are found. Wsh/+ mice had 356.1±34.0 while Wsh/Wsh mice had 371.8±20.6 BrdU+ cells in the SVZ (Fig 5D,E; t9=0.49, p>0.05).
We stained for DCX, a marker of immature neurons, to determine if the reduction in cell proliferation applied to cells of neuronal fate (Fig 6A). Wsh/Wsh mice (n=8) had reduced DCX immunoreactivity compared to Wsh/+ controls (n=8), with a total area of staining of 0.08±0.01mm2 in Wsh/Wsh mice compared to 0.11±0.01 mm2 in Wsh/+ controls (Fig 6B; t14=3.06, p<0.01).
There was a 26.8% reduction in the number of BrdU+ cells in the GCL of the DG in Wsh/Wsh mice (n=9) compared to Wsh/+ controls (n=8) 28 days after BrdU injection (t15=2.38; p<0.05). Given the reduction in cell proliferation in Wsh/Wsh mice, there was no significant difference in cell survival between genotypes (χ2=0.53, p>0.05; Fig 7A).
BrdU and NeuN co-localization in animals used in the cell survival study was examined to determine if differentiation of new cells in the dentate gyrus was affected in Wsh/Wsh mice (Fig 7B). In Wsh/Wsh mice (n=9), 63.4±2.8% of BrdU labeled cells were NeuN positive, compared to 51.9±4.6% in Wsh/+ mice (n=8; t15=1.76; p=0.11). Although this difference was not statistically significant, this suggestive increase in BrdU+NeuN+ cells in Wsh/Wsh mice may represent a compensatory increase in neuron fate due to the reduced cell proliferation.
There were no significant effects of genotype on cell survival in the olfactory bulb (Fig 7C, D). Wsh/+ (n=9) and Wsh/Wsh (n=8) mice had similar number of BrdU+ cells per 100 micron cube of tissue within the granular cell layer of the olfactory bulb, 34.67±1.53 and 34.29±0.98 respectively (t15=0.2, p>0.05).
Given the known effects of chronically elevated corticosterone on hippocampal neurogenesis (Gould et al., 1991) and the anxiety phenotype of Wsh/Wsh mice (Nautiyal et al., 2008), we measured plasma corticosterone to assess whether the reduction in cell proliferation in Wsh/Wsh mice was associated with stress. There were no significant differences in baseline levels of plasma corticosterone between Wsh/Wsh (n=7) and Wsh/+ (n=7) mice (Table 2; t12=0.05, p>0.05).
The cellular localization of c-kit receptor in the wt mouse brain has previously been described (Zhang and Federoff, 1997). To examine c-kit receptor expression in the brain, we performed immunohistochemistry in hippocampus and cerebellum, and found no differences in the brain of Wsh/Wsh mice compared to Wsh/+ littermates (Fig 8).
Given the known effects of SSRI treatment on serotonin signaling and neurogenesis in mice (Malberg et al., 2000), we asked if fluoxetine treatment in Wsh/Wsh mice would reverse the deficit in neurogenesis (Fig 9). Wsh/Wsh (n=17) and Wsh/+ (n=16) mice were assessed for cell proliferation by BrdU labeling following 35 days of treatment with fluoxetine or saline (Fig 9A). The deficit in cell proliferation in Wsh/Wsh mice was replicated in these saline-treated animals. Wsh/Wsh mice had a ~40% reduction in the number of BrdU+ cells compared to Wsh/+ mice (t13=4.05, p<0.01). Fluoxetine treatment resulted in an increase of BrdU+ cells, regardless of genotype (F1,29=15.9, p<0.01). Fluoxetine increased the number of BrdU+ cells in the dentate gyrus from 62.0±4.4 to 90.2±13.5 in Wsh/+ mice (Mann-Whitney, p<0.05) and from 36.4±4.5 to 107.8±17.4 in Wsh/Wsh mice (t15=3.76, p<0.01). This indicates that Wsh/Wsh mice can sustain normal cell proliferation in the DG. There was no significant effect of genotype on the number of BrdU+ cells in fluoxetine-treated mice (t16=0.80, p>0.05). There was a suggestive, but non-significant interaction of genotype and treatment (F1,29=2.99; p=0.09). A similar reversal of the deficits in neurogenesis was also seen with the assessment of DCX immunoreactivity (Fig 9B). Fluoxetine treatment increased DCX staining regardless of genotype (F1,29=8.43; p<0.01) from 0.10±0.01 to 0.18±0.03 mm2 in Wsh/+ mice and from 0.07±0.01 to 0.19±0.04mm2 in Wsh/Wsh mice (t15=2.71, p<0.05). SSRI treatment increased DCX staining in Wsh/Wsh mice more than in Wsh/+ mice resulting in no significant genotype effects following treatment (t16=0.41, p>0.05).
The present studies indicate that mast cells contribute to hippocampal physiology and function. We show that mast cells are a source of serotonin in the hippocampus. Mast cell deficient Wsh/Wsh micee have deficits in hippocampus-dependent spatial learning and memory as assessed in the Morris water maze and radial arm maze. Additionally, mast cell deficient Wsh/Wsh mice have reduced neurogenesis in hippocampus, but not in SVZ. Finally, elevating serotonin by treatment with the SSRI, fluoxetine, restores neurogenesis in the deficient animals.
Mast cells lie on the brain side of the blood-brain barrier, and are poised to act on adjacent neurons, glia and blood vessels (Silverman et al., 2000). They are found in the brains of all mammals that have been studied (Olsson, 1969; Dropp, 1976, 1979; Persinger, 1979; Goldschmidt et al., 1985; Gill & Rissman, 1998; Kriegsfeld et al., 2003). Some fraction of CNS resident mast cells are always activated, releasing granules by piecemeal partial degranulation (Kraeuter Kops et al., 1990; Dvorak et al., 1992) even in animals at rest with no activation of other immune system components (Wilhelm et al., 2000). Mast cells granular contents can diffuse a great distance, and their granule remnants are seen up to 500μm away from the cell of origin (Ruoss et al., 1991; Marszalek et al., 1997; Nautiyal et al., 2008). Their location in leptomeninges also enables release of mast cells mediators into the CSF, allowing access to hippocampal parenchyma via diffusion (Proescholdt et al., 2000). In summary, the data indicate that mast cells are present in the mouse brain throughout development and adulthood, that they contribute serotonin and possibly other mediators to nearby hippocampal tissue and that they affect cell proliferation and behavior.
Altered serotonin signaling in the hippocampus is associated with reduced neurogenesis and deficits in spatial learning (Lucki, 1998; Gould, 1999; Benninghoff et al., 2010). Additionally, mast cells are known to synthesize serotonin in both rodents (Ringvall et al., 2008) and humans (Kushnir-Sukhov et al., 2007) and to release serotonin following application of C48/80 (Lambracht-Hall et al., 1990; Marathias et al., 1991). Here we demonstrate mast cells can contribute serotonin to the caudal hippocampus (Fig 3). Our data show that serotonin levels in the hippocampus Wsh/+, but not Wsh/ Wsh mice are elevated following stimulation of mast cells with C48/80. This serotonin is likely contributed both by mast cells within the hippocampus, as well as those lying in the nearby leptomeninges. As a control, we examined Tyr levels and found that this does not change following C48/80 stimulation. The lack of differences in resting serotonin levels, between Wsh/+ and Wsh/Wsh, may be due either to limits in assay sensitivity or to variability in mast cell numbers, rather than to the absence of differences between the groups. Overall, the results confirm a significant contribution of mast cells to the neuronal pool of serotonin in the hippocampus. That said, while the present studies focus on serotonin, it is possible that other mast cell mediators including histamine, nerve growth factor and cytokines also contribute to hippocampal physiology and hippocampus dependent behavior.
The effects of SSRI administration suggests a deficit in serotonin signaling in the Wsh/Wsh mice. Administration of the SSRI fluoxetine, does not correct the mutation in Wsh/Wsh mice, yet does increase hippocampal serotonin signaling (Malberg et al., 2000) and augments cell proliferation by stimulating division of early progenitor cells (Encinas et al., 2006). The levels of cell proliferation seen in Wsh/Wsh mice following chronic fluoxetine treatment are comparable to that of their Wsh/+ littermates (Fig 8) and C57BL/6 WT mice in general (Santarelli et al., 2003; Navailles et al., 2008). Thus, the reduction in neurogenesis in Wsh/Wsh mice is not due to a defect in progenitor cells themselves or in their responses to serotonergic signaling. Following SSRI treatment, Wsh/Wsh mice show somewhat greater increases in cell proliferation than Wsh/+ mice, perhaps mediated by an increased sensitivity to the elevated serotonin levels. This may result from an increase in 5-HT1A or 5-HT2C receptor expression, as these subserve serotonin-mediated increases in neurogenesis (Benninghoff et al., 2010). As previously reported for the C57BL/6 strain (Crowley et al., 2005; Sugimoto et al., 2008), we found no effect of SSRI treatment on depressive behaviors (data not shown).
While we provide evidence in the SSRI studies that the observed hippocampal deficits in Wsh/Wsh mice may be accounted for by the mast cell deficit, the c-kit mutation may also contribute. Thus, stem cell factor - c-kit receptor signaling in the hippocampus are implicated in spatial learning and memory and neurogenesis (Katafuchi et al., 2000; Zhang & Sieber-Blum, 2009). However, c-kit receptor immunostaining in the brain reveals no difference between Wsh/Wsh and Wsh/+ mice (Fig 8).
Adult hippocampal neurogenesis is implicated in the regulation of emotionality and memory (Shors et al., 2001; Snyder et al., 2005; Sahay and Hen, 2007) and genetic ablation of neurogenesis throughout the brain produces major deficits in performance on spatial learning and memory tasks (Garthe et al., 2009). This raises the possibility that the deficits in hippocampal neurogenesis may be causally related to deficits in spatial learning and memory in Wsh/Wsh mice. However, in the data presented here, the deficits of Wsh/Wsh mice on these tasks are more profound than previously reported following reduced neurogenesis. For example, following ablation of adult hippocampal neurogenesis, there is no effect on acquisition of hippocampal dependent spatial learning and memory tasks such as the Morris water maze, radial arm maze and Y-maze (Snyder et al., 2005; Saxe et al., 2006; Saxe et al., 2007; Clark et al., 2008). In studies where deficits in performance are reported (Zhang et al., 2008), they are much smaller in comparison to those seen here. These prior studies focused on the effects of reduced neurogenesis in adulthood. Possibly, the profound deficits in the Wsh/Wsh mouse deficits in spatial learning and memory (this study) and in emotionality behaviors (Nautiyal et al., 2008), result from the absence of mast cell-dependent neurogenesis and decreased serotonin of mast cell origin during development. In support, decreased serotonin during development contributes to mood disorders such as anxiety and depression (reviewed in (Daubert and Condron, 2010).
Overall these data suggest a striking instance of an immune system cell contribution to normal brain development and extend the existing body of work that shows that immune cells and their signaling molecules contribute to normal hippocampal function (Raison et al., 2006; Ziv et al., 2006; Goshen et al., 2008; Koo and Duman, 2008). The consequences of inadequate immune activation during development may include an increased prevalence of hippocampus-related behavioral disorders, including anxiety and depression (reviewed in Rook and Lowry, 2008).
We thank Drs. Matthew Butler, Michael Drew, and Lance Kriegsfeld for their comments on earlier drafts of the manuscript, Dr. Rene Hen for helpful discussions, and Dr. Joseph LeSauter and Sharmin Ferdaus for technical assistance.
Funding sources: F31 NIMH 084384 (to KMN), NSF IOS 05-54514 (to RS), NSF DBI 320988 (to Barnard College), R21 NIMH 067782 (to RS), P30DA018310 (to JVS) and R01 DE018866 (to JVS); the content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Any Conflict of Interest: No conflict of interest