Serotonin Regulates Osteoclast Differentiation Through Its Transporter


  • The authors have no conflict of interest


5-HTT mediates antidepressant-sensitive clearance of 5-HT after its release into neural synapses. We found increased expression of 5-HTT in RANKL-induced osteoclast-like cells. Fluoxetine, an inhibitor of 5-HTT, reduced osteoclast differentiation but not activation. Reserpine, an inhibitor of 5-HT intracellular transport, potentiated differentiation. These results indicate a role for 5-HTT in osteoclast function and suggest that commonly used antidepressive agents may affect bone mass.

Introduction: Interactions between the serotonergic and skeletal systems are suggested by various clinical observations but are poorly understood.

Materials and Methods: Using gene microarrays, we found that the serotonin transporter (5-HTT) was strongly expressed in RANKL-induced osteoclasts. Using RANKL stimulation of RAW264.7 cells and mouse bone marrow cells as a model system for osteoclast differentiation, we studied the possible role/s of the different components of the serotonin (5-HT) system on the differentiation process.

Results: Osteoclast 5-HTT exhibited typical 5-HT uptake activity that was inhibitable by fluoxetine (Prozac). Fluoxetine reduced osteoclast differentiation but did not inhibit the activation of preformed osteoclasts, whereas the addition of 5-HT itself enhanced differentiation. Fluoxetine-treated osteoclast precursors had reduced NF-κB activation and elevated inhibitory protein κBα (IκBα) levels compared with untreated cells. 5-HT, on the other hand, resulted in activation of NF-κB. Reserpine inhibition of intracellular transport of 5-HT into cytoplasmic vesicles potentiated RANKL-induced osteoclast formation, suggesting the importance of intracellular 5-HT in regulating osteoclast differentiation. Reserpine also modestly enhanced the expression of the osteoclast marker TRACP in the absence of RANKL.

Conclusions: Taken together, these data suggest that the 5-HT system plays an important role in bone homeostasis through effects on osteoclast differentiation and implies that commonly used antidepressive agents may affect bone mass.


BONE RESORPTION BY osteoclasts is critical for normal bone development and remodeling. Excessive resorption is a key pathogenic component in osteopenic conditions such as osteoporosis, arthritis, periodontitis, and certain malignancies. Bone resorption is regulated by a complex system of hormones and locally produced cytokines that stimulate osteoblasts and stromal cells to express RANKL, which results in the differentiation and activation of osteoclasts.(1–8)

Although these key mechanisms of osteoclast formation and activation have now been elucidated, many gaps remain in a complete understanding of the regulation of this cell type. Evidence for neuroendocrine regulation of the immune system is mounting; however, whether the neural system also regulates bone is less certain. Neuroendocrine-immune system interactions are supported by a number of observations: (1) spleen, thymus, bone marrow, and lymph nodes are innervated by neurons of the autonomic nervous system; (2) changes in brain functions can affect different immune responses; (3) immune and neuroendocrine cells share receptors for serotonin and neuropeptides; (4) neuropeptides can alter the functional activity of immune system cells; (5) several hormones and neuropeptides can be synthesized by leukocytes; and (6) cytokines produced by leukocytes are able to modulate neuroendocrine system activity, behavior, sleep, and thermoregulation.(9–16)

Because osteoclasts derive from hematopoietic cell precursors and a relationship between bone and the immune system has been established,(17–20) it is possible that neuroendocrine mechanisms, in particular those related to the serotonin (5-HT) system, could also regulate osteoclast differentiation/activation. In this regard, fluoxetine (Prozac), a selective serotonin reuptake inhibitor (SSRI) that blocks 5-HT uptake by its specific transporter (5-HTT), has been reported to reduce bone resorption in mice with adjuvant-induced arthritis.(21) Such observations suggest that there may be a significant relationship between the 5-HT system and bone remodeling.

In this study, we investigated the regulation of osteoclasts by the 5-HT system. We report that 5-HT does indeed exert significant regulatory effects on osteoclast development and activation in both in vitro and in vivo systems.


Differentiation of RAW264.7 and bone marrow cells in vitro

RAW264.7 mouse macrophage/monocytes (TIB-71) were purchased from ATCC (Manassas, VA, USA). Cells were cultured in DMEM/1.5 g/liter sodium bicarbonate (JRH Biosciences, Lenexa, KS, USA) plus 10% non-heat-inactivated FBS (Invitrogen, Carlsbad, CA, USA). DMEM was supplemented with recombinant mouse soluble RANKL (a gift from Dr William Dougall, Immunex, Seattle, WA, USA) at a concentration of 10 ng/ml. Recombinant murine RANKL is an NH2-terminal fusion of a leucine zipper trimerization domain(22) with residues 134-316 of murine RANKL. Mouse bone marrow cells were obtained from long bones of 2-week-old C57Bl6/J mice and cultured for 7 days in αMEM/10% FBS supplemented with 20 ng/ml RANKL and 50 ng/ml human M-CSF (PeproTech, Rocky Hill, NJ, USA). In some experiments, 5-HT was stripped from the FBS by incubation twice in 0.25% dextran-coated charcoal (Sigma-Aldrich, St Louis, MO, USA) for 30 minutes at 45°C. When indicated, fluoxetine (Sigma-Aldrich) or serotonin receptor antagonists specific for 5-HT1B, 5-HT2B, and 5-HT4 (SB216641, SB204741, and CR113808, respectively; Tocris-Cookson, Ellisville, MO, USA) were added to the cultures. Cultures were also exposed to the following inhibitors: reserpine (VMAT1), clorgyline (MAO-A), or pargylline (MAO-B; all from Sigma-Aldrich) for 4 days.

5-HT uptake

The method to determine the 5-HT uptake was previously described.(23,24) Briefly, RAW264.7 cells were plated in 48-well plates at a density of 104 cells/well and cultured for 4 days with or without RANKL (10 ng/ml). Cells were washed twice with HBSS and incubated with the indicated concentrations of [3H]5-HT (84 Ci/mmol; Amersham Pharmacia Biotech, Piscataway, NJ, USA) in 100 μl HBSS with or without 3 μM fluoxetine for 10 minutes at 37°C. Incubation was terminated with two washes of ice-cold HBSS and one wash of PBS. The accumulated [3H]5-HT was determined by solubilizing the cells with 100 μl 1% SDS and liquid scintillation spectrometry.

Cytochemical and immunochemical staining

Cells were stained for TRACP using a commercially available kit (387-A; Sigma-Aldrich). Undifferentiated and differentiated RAW264.7 and bone marrow cells were stained using an anti-5-HTT antibody as described.(25) Briefly, cells were fixed in 4% paraformaldehyde for 15 minutes, rinsed in PBS, and incubated overnight at 4°C in blocking solution (5% Sheep Serum/0.1% Triton X-100 in PBS). Cells were subsequently incubated with a mouse monoclonal antibody to the serotonin transporter (AB-N09; Advanced Targeting Systems, San Diego, CA, USA) for 90 minutes at room temperature, washed with 2.5% Sheep Serum/0.1% Triton X-100 in PBS, and incubated with a chicken anti-mouse/Rhodamine conjugated (sc-2861; Santa Cruz Biotech, Santa Cruz, CA, USA) for 60 minutes at room temperature. Finally, cells were incubated with the cathepsin K substrate (Z-Leu-Arg)2-rhodamine 110 (CalBioChem, La Jolla, CA, USA) for 10 minutes at room temperature, washed in PBS, and sealed under coverslip in 4-diazabicyclo(2,2,2)octane (Sigma-Aldrich). Visualization and photography were performed with a confocal microscope.

RT-PCR and Northern blot analysis

Total RNA was extracted from cells using the TRIAZOL reagent (Invitrogen). For RT-PCR, 2 μg of RNA were reversed-transcribed to cDNA using SuperScript II (Invitrogen) following the manufacturer's instructions. One-tenth (2 μl) of the cDNA was used as a template for the PCR reaction (35 cycles at 94°C for 1 minute, 58°C for 1 minute, and 72°C for 1 minute). For Northern blot analysis, 10 μg of RNA were size fractionated on a 1.2% agarose denaturing gel and transferred onto nylon membranes. The membranes were hybridized to32P radiolabeled probes overnight at 42°C in hybridization solution (50% formamide, 5× SSC, 5× Denhardt's solution, 0.5% SDS, and 100 μg/ml salmon sperm DNA), washed twice to remove unbound probe in 1× SSC, 0.2% SDS at 50°C for 20 minutes, and exposed to X-ray film. The DNA probes were generated by RT-PCR from osteoclast (OC) RNA, using the following primer-pairs: 5-htt (GenBank Accession NM_010484), sense 5′-GACCAGTGTGGTGAACTGCATGAC-3′, antisense 5′-GATGATGGCAAAGAATGTGGATGCTG-3′, amplicon size: 189 bp; Mao-a (GenBank Accession XM_217588); sense 5′-ACACAGTGGAGTGGCTACATG-3′; antisense 5′-GGGAGCTTCTTAATCTTGTAC-3′, amplicon size: 259 bp; Mao-b (GenBank Accession NM_013198); sense 5′-GAGAGAGAGCTGCCAGAGAG-3′; antisense 5′-GAGCTGTTGCTGACAAGATGG-3′, amplicon size: 190 bp; vmt-1 (GenBank Accession NM_013152); sense 5′-CTGGTGGACTTACGCCACACC-3′; antisense 5′-CGCTGCTAGGATCATCGCTG-3′, amplicon size: 320 bp; vmt-2 (GenBank Accession NM_013031); sense 5′-CGTGGCCTTTTGTATGGGCT-3′; antisense 5′-GGGTCTCAGTCACTTTCAGATTC-3′, amplicon size: 272 bp; TPH (GenBank Accession NM_009377); sense 5′-GTCCTGTGGCTGGTTACCTC-3′; antisense 5′-TGTTTGCACAGCCCAAACTC-3′, amplicon size: 280 bp; AADC (GenBank Accession NM_016672); sense 5′-GAGTTTGAGTCACTGGTACG-3′; antisense 5′-GATCACTGATGTGTTCCCAG-3′, amplicon size: 250 bp; β-actin (GenBank Accession NM_007393); sense 5′-GAAGAGCTATGAGCTGCCTG-3′; antisense 5′-CACAGAGTACTTGCGCTCAG-3′, amplicon size: 306 bp; tracp (GenBank Accession BC029644); sense 5′-ACACAGTGATGCTGTGTGGCAACTC-3′; antisense 5′-CCAGAGGCTTCCACATATATGATGG-3′, amplicon size: 465 bp; cathepsin K (GenBank Accession NM_007802); sense 5′-CTGAAGATGCTTTCCCATATGTGGG-3′; antisense 5′-GCAGGCGTTGTTCTTATT-CCGAGC-3′, amplicon size: 364 bp; MMP-9 (GenBank Accession BC046991); sense 5′-CGAGTGGACGCGACCGTAGTTGG-3′; antisense 5′-CAGGCTTAGAGCCA-CGACCATACAG-3′, amplicon size: 355 bp.

Probe sequences were confirmed by sequencing.

Cell proliferation and apoptosis assays

Cell proliferation studies were carried out using the Quick Cell Proliferation Assay Kit (BioVision, Mountain View, CA, USA), following the manufacturer's instructions. DNA fragmentation studies were done as follows: RAW264.7 cells were seeded onto 24-well plates (5 × 104 cells/well) and cultured for 96 h, washed with PBS, and lysed with lysis buffer (5 mM Tris/HCL [pH 8.0]/20 mM EDTA/0.5% [vol/vol] Triton X-100/0.1% SDS). High molecular weight DNA was removed by centrifugation at 14,000 rpm for 30 minutes, and the supernatants were extracted with phenol/chloroform/isoamyl alcohol (25:24:1). The soluble DNA was precipitated with 2.5 volumes of ethanol in the presence of 0.3 M sodium acetate. The DNA was resuspended in TE, treated with DNase-free RNase (Roche Applied Science, Indianapolis, IN, USA) for 1 h at 30°C, and electrophoresed in a 1.5% agarose gel. Apoptosis studies were also done using the Vybrant Apoptosis Assay Kit 4 (Molecular Probes) following the manufacturer's instructions. This assay allows the detection of apoptosis based on changes that occur in the permeability of cell membranes. After staining live, unfixed cultures, apoptotic cells exhibit green fluorescence, and necrotic cells show red or red and green fluorescence, whereas viable cells show no fluorescence.

Resorption assay

RAW264.7 cells were cultured for 5 days on a 3D Collagen Cell Culture System matrix following the manufacturer's instructions (Chemicon International, Temecula, CA, USA) in DMEM/10% FBS with or without 20 ng/ml mouse RANKL. Cells were subsequently removed from the matrix by collagenase treatment and cultured on dentin slices (ALPCO Diagnostics, Windham, NH, USA) for 36 h in DMEM/10% FBS 20 ng/ml mouse RANKL. After incubation, the cells were removed by washing in 0.2 M ammonium hydroxide for 15 minutes. Resorption lacunae were visualized by scanning electron microscopy and counted.

Isolation of nuclear proteins

Cells were pelleted, washed twice with PBS, and resuspended in 800 μl ice-cold lysis buffer (HEPES, 10 mM; KCL, 10 mM; EDTA, 0.1 mM; EGTA, 0.1 mM; DTT, 1.0 mM; PMSF, 1.0 mM; aprotinin, 10 μg/ml; pepstatin, 10 μg/ml; leupeptin, 10 μg/ml). The samples were incubated on ice for 30 minutes, vortexed for 30 s after addition of 50 μl of 10% NP-40, and centrifuged for 10 minutes at 4°C. The pellets were resuspended in an ice-cold nuclear buffer (HEPES, 20 mM; NaCl, 400 mM; EDTA, 1.0 mM; EGTA, 1.0 mM; DTT, 1.0 mM; PMSF, 1.0 mM; aprotinin, 10 μg/ml, pepstatin, 10 μg/ml; leupeptin, 10 μg/ml), incubated on ice for 2 h, and centrifuged for 10 minutes at 4°C. The supernatants were collected as nuclear extract and stored at −70°C. Protein concentration was measured using the Pierce protein assay reagent (Pierce, Rockford, IL, USA).

Electrophoretic mobility shift assay

Double-stranded oligonucleotides containing the binding sites for NF-κB (5′-AGTTGAGGGGATTTCCCAGGC-3′, 3′-TCAACTCCCCTGAAAGGGTCCG-5′) and activator protein 1 (AP-1) (5′-CGCTTGATGAGGCAGCCGGAA-3′, 3′-GCGAACTACTCAGTCGGCCTT-5′) were end-labeled with [32P]-ATP and incubated with the nuclear extract for 20 minutes at room temperature. The samples were loaded on a 4% nondenaturing polyacrylamide gel. The gel was dried and exposed to X-ray film. For competition experiments, nuclear extracts were incubated with unlabeled double-stranded NF-κB or AP-1 oligonucleotides for 20 minutes. Labeled oligonucleotides were added to the reaction mixture. For supershift assays, nuclear extracts were incubated with either p65 or p50 antibodies (Santa Cruz Biotechnology) overnight at 4°C, followed by the addition of the labeled oligonucleotides.


Osteoclasts express a functional 5-HTT

We generated osteoclasts by stimulating murine RAW264.7 macrophage-like cells with RANKL. After 4 days in culture, numerous multinucleated giant cells (>5 nuclei) formed.(26) These cells strongly expressed osteoclast-specific markers, including TRACP, cathepsin K, calcitonin receptor, matrix metalloproteinase-9 (MMP-9), c-src, and the proton pump subunit ATP6I, and formed resorption pits on bone slices and submicron calcium phosphate films.(26)

We used the RAW264.7 differentiation system and gene arrays to identify genes that are differentially expressed as a result of RANKL-induced osteoclast formation. A radiolabeled mixed cDNA probe was prepared by reverse transcribing total RNA from both undifferentiated and 4-day differentiated osteoclasts. The probes were sequentially hybridized to the Mouse 1.2 array (Atlas cDNA Expression Arrays; Clontech Laboratories). Several differentially expressed genes were identified based on their positive hybridization patterns with RANKL-induced osteoclast cDNA and low or absent reactivity with undifferentiated RAW264.7 cDNA. Of note, the sodium-dependent serotonin transporter (5-HTT) was strongly upregulated in RANKL-induced cells (Fig. 1A). 5-htt mRNA was detected by RT-PCR as early as 6 h after RANKL stimulation (Fig. 1B). Northern analysis (Fig. 1C) showed 5-htt expression only after day 2, likely because of low abundance at earlier time-points, and with increasing expression thereafter. 5-HTT was also detected by immunohistochemistry in both RAW264.7 and bone marrow cell-derived osteoclasts, (Fig. 1D, center and right panels, respectively). 5-HTT-expressing cells also expressed cathepsin K (Fig. 1D, left panel), indicating that 5-HTT expression is restricted to differentiated osteoclasts.

Figure FIG. 1..

RANKL-induced osteoclasts express 5-HTT. (A) Gene arrays were hybridized to mixed cDNA probes from (left) undifferentiated and (right) RANKL-induced RAW264.7 cells. Upregulation of 5-HTT is indicated by an arrow. (B) RT-PCR of (right) 5-HTT expression and (left) β-actin in RAW264.7 cells. M, standard; −, no RANKL; +, RANKL stimulation for 4 days; 6, 12, and 24, RANKL stimulation for 6, 12, and 24 h. (C) Northern analysis of 5-HTT expression. U, no RANKL; lanes 2-4, RANKL stimulation for 2, 3, and 4 days, respectively. (D) Immunochemistry of 5-HTT expression. (Left) Differentiated RAW264.7 cells stained with preimmune mouse IgG and rhodamine-conjugated chicken anti-mouse IgG. (Middle) Differentiated RAW264.7 cells stained with (Z-Leu-Arg)2-rhodamine-110. Green fluorescence indicate the presence of rhodamine-110, a product of cathepsin K-mediated cleavage of the substrate. (Right) Differentiated (top) RAW264.7 and (bottom) bone marrow cells stained with anti-5-HTT monoclonal antibody and rhodamine-conjugated chicken anti-mouse IgG. 5-HTT-expressing multinucleated osteoclasts are indicated by arrows.

To determine if the osteoclast-expressed 5-HTT was functional, we performed 5-HT uptake assays. Differentiated osteoclasts transported [3H] 5-HT in a concentration-dependent manner, indicating that the 5-HTT can be saturated (Fig. 2A). Uptake was completely blocked by the selective serotonin reuptake inhibitor fluoxetine at concentrations of ≥1 μM. Kinetic studies revealed that 5-HT uptake activity could be seen as early as 24 h and peaked by 48 h after RANKL stimulation (Fig. 2B).

Figure FIG. 2..

5-HTT is functional in osteoclasts. (A) Uptake of 5-HT by the osteoclast-expressed transporter. RAW264.7 cells were cultured in the presence/absence of RANKL. The uptake of [3H]5-HT was assessed in the presence/absence of fluoxetine. *p < 0.01. (B) Kinetics of acquisition of 5-HTT uptake activity in RAW264.7-derived osteoclasts. Fluoxetine (3 μM) was added to the uptake assay to show 5-HTT-specific uptake.

5-HTT blockade inhibits osteoclast formation and activation

The differential expression of a functional 5-HTT in osteoclasts suggested that 5-HT transport could play a role in the differentiation of this cell type. Accordingly, we determined the effect of the 5-HTT inhibitor fluoxetine on osteoclast formation. As shown in Fig. 3A, fluoxetine inhibited the formation of TRACP+ osteoclasts from RAW264.7 cells in a dose-dependent manner (432, 266, and 122 osteoclasts/culture in untreated, 1 μM, and 3 μM fluoxetine, respectively). Densitometric analysis of Northern blots from parallel cultures showed a significant reduction in the expression of the late osteoclast markers cathepsin K and MMP-9 in fluoxetine-treated cells (∼70% and 80%, respectively; Fig. 3B), whereas a smaller effect was seen on the expression of an early marker, TRACP (∼20% reduction). To confirm this finding in normal cells, we also generated osteoclasts from normal mouse bone marrow cells in the presence of M-CSF and RANKL. We have observed that bone marrow cell-derived osteoclasts are more efficiently generated in the presence of 50 ng/ml RANKL. Fluoxetine also significantly inhibited bone marrow cell-derived osteoclast differentiation (Fig. 3C; 2354 and 178 osteoclasts/culture in untreated and 3 μM fluoxetine, respectively). Bone marrow cell-derived osteoclasts expressed osteoclast markers. Expression of TRACP, cathepsin K, and MMP-9 mRNA was also reduced in fluoxetine-treated bone marrow cell cultures, as shown by RT-PCR analysis (Fig. 3D, compare lanes F and +).

Figure FIG. 3..

Fluoxetine inhibits RANKL-induced osteoclast formation by RAW264.7 and normal bone marrow macrophages (BMMs). (A) RAW 264.7-derived osteoclasts were quantified by TRACP staining after 4 days. *p < 0.01. (B) Fluoxetine inhibits the expression of osteoclastic genes cathepsin K and MMP-9 in RANKL-stimulated RAW264.7 cells. Northern analysis. (C) Fluoxetine inhibits RANKL-induced osteoclast formation by normal BMMs. BMMs were cultured in the presence of M-CSF and RANKL, and the number of osteoclasts was quantified after 7 days by TRACP staining. *p < 0.001. (D) Expression of osteoclast markers TRACP, cathepsin K, and MMP-9 is assessed by RT-PCR analysis. BMMs were differentiated in the presence of receptor antagonists specific for 5-HT1B (0.5 μM, lane 1B), 5-HT2B (3 μM, lane 2B), 5-HT4 (3 μM, lane 4), a combination of the three receptor antagonists (lane All), or fluoxetine (lane F). Lane +, control cultures (no antagonists).

To determine if fluoxetine was inhibiting proliferation or inducing apoptosis and to rule out a possible toxicity of fluoxetine, a proliferation assay was carried out. As shown in Fig. 4A, the proliferation of RAW264.7 cells in the presence of RANKL was unaffected by up to 3 μM fluoxetine. We also performed DNA fragmentation analysis and a fluorescence-based apoptosis assay in RANKL-induced fluoxetine-treated cells. Control cultures were treated with 1 mM sodium nitroprusside (SNP), a drug that induces apoptosis in RAW264.7 cells (Fig. 4B, +SNP). Fluoxetine did not induce either fluorescent staining (Fig. 4B, Fluo/RANKL) or DNA fragmentation (Fig. 4C, lane 2) in RAW264.7 cells, indicating that no apoptosis or necrosis was occurring.

Figure FIG. 4..

Fluoxetine does not affect RAW264.7 cell proliferation nor does it induce apoptosis. (A) Fluoxetine has no effect on the proliferation of RAW264.7 cells, as assessed by a modified MTT assay. (B) RAW264.7 cells stained with the Vybrant Apoptosis Assay Kit 4. Cultures treated with 1 mM sodium nitroprusside show numerous apoptotic (green) and necrotic (red and yellow) cells (+SNP, fluorescence). Cultures treated with RANKL in the absence (−Fluo/+RANKL, fluorescence) or presence (+Fluo/+RANKL, fluorescence) of fluoxetine show no apoptotic cells. (C) Effects of fluoxetine on DNA fragmentation. DNA was extracted from RANKL-induced (lane 1) and RANKL-induced/fluoxetine-treated (lane 2) cells. There is no evidence of DNA fragmentation as a result of the fluoxetine treatment. DNA marker (lane 3).

To determine whether fluoxetine was also affecting activation of preformed osteoclasts, we stimulated RAW264.7 cells with RANKL in a collagen matrix and isolated the differentiated osteoclasts by enzymatic degradation of the gel. Recovered osteoclasts were plated onto dentin slices to assess resorption. In preliminary control experiments, we found that RANKL was required not only for differentiation but also for osteoclast activation (Fig. 5A). Interestingly, there was no significant difference between the control and fluoxetine-treated cells (Fig. 5B). These results suggest that fluoxetine affects primarily the differentiation of precursors rather than the activation of preformed osteoclasts.

Figure FIG. 5..

Effect of fluoxetine on osteoclast resorption activity. (A) RANKL-stimulated RAW264.7 cells form resorption pits. (B) Preformed osteoclasts were plated on dentin slices in the presence of fluoxetine. The number of resorption pits was not affected by fluoxetine treatment (179 and 205 pits in control and fluoxetine-treated cells, respectively).

5-HT stimulates osteoclast formation

The requirement for intact 5-HTT function in differentiating osteoclasts suggested that 5-HT is needed and must be transported into the cell for differentiation to occur. The 10% FBS-supplemented medium used to support cultures contained high levels of 5-HT from platelet lysis (∼300 ng/ml by ELISA). Not unexpectedly, the addition of 5-HT (3-300 ng/ml) had no effect on osteoclast formation in cultures containing FBS. We then depleted the 5-HT from the FBS ∼80-fold by two cycles of adsorption to dextran-activated charcoal. When we added serotonin back (3-300 ng/ml) to RANKL-stimulated RAW264.7 cultures, the expression of osteoclast markers cathepsin K and MMP-9 were both upregulated in a 5-HT dose-dependent manner (Fig. 6A). 5-HT also had a dose-dependent stimulatory effect on the ability of normal bone marrow cells to generate osteoclasts (Fig. 6B). To determine if RAW264.7 cells could produce 5-HT, we performed RT-PCR analysis using specific primers for tryptophan hydroxylase (TPH) and aromatic-L-amino acid decarboxylase (AADC). These two enzymes are essential for 5-HT synthesis. Our results showed (Fig. 6C) that neither undifferentiated (lane −) nor differentiated (lane +) RAW264.7 cells could express these enzymes, suggesting that they must obtain 5-HT from an external source in vivo.

Figure FIG. 6..

5-HT dose-dependently increases osteoclast differentiation. (A) RAW264.7 cells were cultured in serotonin-depleted serum (FBS−) or in depleted serum supplemented with 3, 30, or 300 ng/ml added 5-HT. Cells were stimulated with RANKL for 3 or 4 days, and the expression of osteoclastic genes cathepsin K and MMP-9 was assessed by Northern blotting. (B) 5-HT stimulates osteoclast formation by normal BMMs. BMMs were cultured in FBS− supplemented with 3, 30, and 300 ng/ml added 5-HT and 300 ng/ml 5-HT plus 3 μM fluoxetine (300/F) in the presence of M-CSF and RANKL. The number of multinucleated osteoclasts was quantified after 7 days by TRACP staining. *p < 0.05, **p < 0.001. (C) Undifferentiated (lane −) and differentiated (lane +) RAW264.7 cells do not express TPH or AADC mRNA. Brain RNA (lane B) was used as positive control for RT-PCR.

Blockade of OC-expressed 5-HT receptors does not affect osteoclast formation

In addition to 5-HTT, RAW264.7-derived osteoclasts expressed mRNA transcripts for 5-HT receptors 5-HT1B, 5-HT2B, and 5-HT4 (Fig. 7A). 5-HT could therefore theoretically modulate osteoclast differentiation and activation through 5-HTT, 5-HT receptors, or both. To distinguish among these possibilities, osteoclast differentiation was assessed in the absence/presence of fluoxetine and receptor antagonists specific for 5-HT1B, 5-HT2B, and 5-HT4. As shown in Fig. 7B, fluoxetine suppressed differentiation of RAW264.7 cells into osteoclasts (752 and 242 in control and fluoxetine-treated cells, respectively). In contrast, none of the three 5-HT receptor antagonists tested had any effect on differentiation when tested individually or in combination (795 for 1B, 846 for 2B, 874 for 4, and 876 for the 1B/2B/4 combination; Fig. 7B). Interestingly, somewhat different results were obtained using normal bone marrow cells (Fig. 7C, top). Treatment with antagonists to receptors 1B and 4 significantly reduced the ability of bone marrow cells to form osteoclasts (434, 67, 256, and 15 osteoclasts/culture in untreated, 1B-, 4-, and 1B/2B/4-treated cells, respectively). Treatment with the same antagonists also resulted in a reduction in the expression of osteoclast markers TRACP, cathepsin K, and MMP-9 (Fig. 3D). A parallel experiment was done in which the cells were cultured for 2 days before starting the antagonist treatment (Fig. 7C, bottom). In this context, we observed that treatment with antagonists 1B and 4 also inhibited osteoclast formation, although to a lesser extent. The difference was particularly of note in the case of antagonist 1B (67 and 234 osteoclasts at days 0 and 2, respectively). This result suggested that the antagonists' effect was exerted on early osteoclast precursors present in bone marrow cell preparations. Bone marrow cells stimulated for 2 days with RANKL and M-CSF are more similar to RAW264.7 cells, where the antagonists have no effect on RANKL-induced differentiation.

Figure FIG. 7..

Effect of 5-HTT and 5-HT receptor modulation on osteoclast differentiation. (A) RT-PCR analysis of 5-HT receptor expression in RANKL-induced RAW264.7 osteoclasts after 4 days. Brain, positive control. (B) RAW264.7 cells were induced with RANKL in medium containing normal FBS. Cells were exposed to fluoxetine and/or receptor antagonists specific for 5-HT1B (0.5 μM), 5-HT2B, and 5-HT4 (3 μM each). Osteoclasts were quantified by TRACP staining. *p < 0.01.

Reserpine stimulates osteoclast multinucleation

In neurons, 5-HT taken up through 5-HTT is either degraded by monoamine oxidase (MAO) or sequestered in intracellular vesicles, the latter action mediated by a vesicular monoamine transporter (VMAT). Both undifferentiated and differentiated RAW264.7 cells express MAO-A, whereas only differentiated cells express VMAT-1 (Fig. 8A). MAO-B expression was not detected. The addition of 3 μm reserpine, a VMAT-1 inhibitor, promoted osteoclast differentiation in the presence of RANKL (410 osteoclasts compared with 181 osteoclasts in untreated cells; Fig. 8B). On the other hand, addition of clorgyline, a specific MAO-A inhibitor, and pargyline (an MAO-B inhibitor) had no effect on osteoclast formation. When the inhibitors were added together with reserpine, the overall effect was that of reserpine alone (380 ± 42 and 410 ± 14 osteoclasts, respectively; Fig. 8B).

Figure FIG. 8..

Expression and function of vesicular monoamine transporter and monoamine oxidase in osteoclasts. (A) RT-PCR analysis of MAO-A, MAO-B, VMAT-1, and VMAT-2 expression in RAW264.7-derived osteoclasts. Lanes: B, brain (positive control); −, no RANKL; +, RANKL stimulation. (B) Blockade of the vesicular monoamine transporter promotes osteoclast formation. Cultures were exposed to the VMAT1 inhibitor reserpine at 0, 0.3, 1, and 3 μM. TRACP+ multinucleated osteoclasts were quantified. *p < 0.05.

In addition to increasing osteoclast numbers, reserpine treatment also stimulated the formation of much larger cells with greater multinucleation that covered the entire culture dish. Of interest, reserpine treatment in the absence of RANKL induced TRACP expression by a number of mono- and binucleated cells (Table 1), although large multinucleated osteoclasts were not formed. Taken together, these data indicate that elevated intracellular 5-HT levels synergize with RANKL-induced signals and may by themselves have some ability to induce the expression of early osteoclastic genes in a RANKL-independent manner.

Table Table 1.. Reserpine-Treated Mono- and Binucleated RAW 264.7 Cells Express TRACP in the Absence of RANKL
original image

Fluoxetine inhibits NF-κB activation

Activation of the transcription factor NF-κB is stimulated by RANKL and is essential for osteoclast formation.(8) The inhibition of differentiation by blocking the 5-HTT might therefore be caused, in part, by an inhibition of NF-κB activation. To test this hypothesis, we performed electrophoretic mobility shift assays (EMSAs) for NF-κB. As shown in Fig. 9A, RANKL induced a 20-fold increase in NF-κB activation compared with unstimulated RAW264.7 cells. In cells treated with RANKL and fluoxetine, activation was inhibited by ∼80%. Addition of a 100-fold excess cold NF-κB oligodeoxynucleotide (ODN) blocked binding, but an unrelated ODN (AP-1) had no effect. We also tested whether 5-HT could stimulate the activation of NF-κB. As shown in Fig. 9B, 5-HT stimulated a modest activation of NF-κB in the absence of RANKL (Fig. 9B, lane 3), whereas the addition of both 5-HT and RANKL had an additive effect. The presence of fluoxetine reduced NF-κB activation (Fig. 9B, lane 5). In fluoxetine-treated RANKL-stimulated cells, levels of the inhibitor inhibitory protein κBα (IκBα) were elevated compared with cells without fluoxetine by Western analysis and were similar to the levels seen in unstimulated cells (Fig. 9C). The reduction in intracellular 5-HT after fluoxetine treatment could therefore impinge on pathways that regulate IκBα phosphorylation and degradation.

Figure FIG. 9..

NF-κB activation is inhibited by fluoxetine and stimulated by serotonin. (A) NF-κB activity was assessed by EMSA in nuclear extracts of RANKL-stimulated RAW264.7 cells after 24 h. Activity was inhibited in cells cultured with fluoxetine (Fluo) and was blocked by 100-fold excess cold NF-κB oligodeoxynucleotide (ODN) but not cold AP-1 ODN. The arrows indicate the NF-κB complex. (B) RAW264.7 cells were cultured for 24 h in 5-HT-depleted medium with different stimuli. Nuclear extracts were made and assayed for NF-κB activity. NF-κB activation was stimulated by RANKL, 5-HT, and RANKL plus 5-HT (RANKL/5-HT). Fluoxetine abolished NF-κB activation (RANKL/5-HT/FLUO). (C) Western blot of IκBα levels in RANKL-induced RAW264.7 osteoclasts after 24 h. Fluoxetine increased IκBα levels ∼2-fold.


5-HT has been known for one-half of a century to participate in neurotransmission, nociception, and vasoactivity. Extracellular levels of 5-HT are regulated in part by its uptake by 5-HTT, thus limiting its availability for cell surface receptor stimulation. However, growing evidence indicates a broader role for 5-HTT, and for the serotonergic system in general, in the development and regulation of both neural and non-neural tissues. 5-HTT mRNA is widely expressed before organogenesis and throughout the second half of gestation, particularly in developing sensory ganglionic neurons and neuroepithelial cells, in neural crest and neural crest-derived tissues, and in the craniofacial and cardiac regions, suggesting a role in regulation of peripheral synaptogenesis(27) and tissue and organ formation. Embryos exposed to SSRIs and receptor ligands exhibit craniofacial malformations, possibly through effects on epithelial-mesenchymal interactions.(28–30)

In addition to these effects, our data now provide evidence for another role of the 5-HT system in the development of osteoclasts and in bone modeling and remodeling. Blocking 5-HTT results in an inhibition of osteoclast formation as well as a reduction in the expression of osteoclast markers TRACP, cathepsin K, and MMP-9.

Another interesting finding was that elevations in cytoplasmic rather than extracellular levels of 5-HT seem be responsible for the regulation of osteoclast development. Because both the addition of 5-HT to 5-HT-depleted medium and reserpine blockade of intracellular VMAT uptake increase differentiation, elevations in cytoplasmic levels of 5-HT may be required to enhance NF-κB activation through unknown mechanisms. In line with this hypothesis, we found that 5-HT resulted in an activation of NF-κB even in the absence of RANKL. To our knowledge, this is the first report of such an activity for 5-HT. In addition, stimulation of RAW264.7 cells with both 5-HT and RANKL resulted in an enhanced activation of NF-κB, which suggests that a common mechanism might be involved. The finding that reserpine by itself can induce some TRACP expression is especially intriguing and suggests that 5-HT, like TNFα,(31) may be able to bypass RANK signaling, at least to a limited extent.

We also found expression of three 5-HT receptors in differentiated RAW264.7 cells: 1B, 2B and 4. Selective blockade of these receptors in RAW264.7 cultures had no effect on osteoclast differentiation. However, a different result was obtained with bone marrow cells. Treatment with antagonists to receptors 1B and 4 inhibited formation of multinucleated TRACP+ cells and resulted in reduced expression of mRNA for the osteoclast markers TRACP, cathepsin K and MMP-9. These apparently conflicting results might be explained, in part, by the fact that RAW264.7 cells, equivalent to TRACP and CD14+ pro-osteoclasts, are more differentiated than the osteoclast precursors present in bone marrow cell preparations. 5-HT receptors 1B and 4 may therefore be important in mediating a differentiation step previous to pro-osteoclast formation. We tested this hypothesis by stimulating the bone marrow cells for two days with RANKL and M-CSF before adding the antagonists. We found that, indeed, although the antagonists still had some inhibitory effect on osteoclast formation, the effect was less pronounced, suggesting that functional receptors 1B and 4 are required for an early step in differentiation.

In addition, 5-HT has been detected in the nucleus as well as the cytoplasm of various cells during embryogenesis,(32) suggesting novel pathways of intracellular transport, as well as possible direct modulatory effects of 5-HT on signaling or gene transcription. However, it is more likely that 5-HT exerts indirect effects. In this regard, it has recently been found that some 5-HT receptors are expressed both on the neuronal cell surface and in the cytoplasm,(33) where they may associate with smooth endoplasmic reticulum-like organelles.(34) These receptors are presumably capable of transducing signals induced by intracellular 5-HT. Because different receptors transmit inhibitory versus stimulatory signals (e.g., 5-HT1B decreases, whereas 5-HT4 increases, cyclic adenosine monophosphate [cAMP]), the expression of some but not all receptors intracellularly could result in a different complex of signals being generated in response to intracellular 5-HT that is transported through 5-HTT compared with extracellular 5-HT that interacts with cell surface receptors. If only one or two of these receptors are present intracytoplasmically, the complex of signals they transduce in response to 5-HTT transport of 5-HT would be different from those on the cell surface where all three receptors subtypes are present. At present, the subcellular distribution of the 5-HT1B, 5-HT2B, and 5-HT4 receptors in osteoclasts is unknown, but will provide clues to the mechanism of 5-HT action.

In addition to its well-established role as a neurotransmitter, 5-HT modulates the proliferation and activity of a variety of cells, including vascular smooth muscle, T- and B-lymphocytes, natural killer (NK) cells, and macrophages.(35–37) 5-HT is also an important morphogen during fetal development. 5-HTT is transiently expressed during mouse embryogenesis in craniofacial epithelia and mesenchyme, including the tooth germ. Prominent expression of 5HT3 receptor transcripts occurs in association with regions of active chondrogenesis in the vertebral column, limbs, and craniofacial region.(38) Differential and overlapping spatio-temporal expression patterns of 5-HT2A-C receptors were found in olfactory receptor neurons, teeth, and genitalia.(39) Mouse embryos exposed to specific 5-HT reuptake inhibitors and receptor ligands exhibited several craniofacial malformations.(40–42)

There are a number of sources of 5-HT in vivo that could modulate osteoclast differentiation. Although modest levels are present in the circulation, 5-HT is released in locally high concentrations from sensory nerve plexuses, paraneurons, platelets, and mast cells.(43) Interestingly, mast cells are present in bone marrow, and their numbers increase dramatically in ovariectomized rats, which suffer rapid bone loss.(44) The human disorder mastocytosis is also significantly associated with osteoporosis.(45)

Neural-skeletal system links have also been described that are mediated by neuropeptides. Calcitonin gene-related peptide (CGRP) is released by sensory nerves and is a weak inhibitor of osteoclast formation and bone resorption.(46) Vasoactive intestinal peptide (VIP) is produced by sympathetic nerves and has a more complex action, with an initial inhibition of resorption through direct effects on osteoclasts followed by stimulation. In the presence of osteoblasts, increased resorption is observed.(47,48) These neuropeptides, as well as 5-HT, have also been shown to modulate immune function through effects on cell proliferation and activation.(49,50) Because osteoclasts are developmentally related to immune cells, it is plausible that multiple neuroendocrine mechanisms also are operative in the regulation of bone resorption.(20)

Although our results provide clear evidence of serotoninergic enhancement of osteoclast development, because 5-HTT and 5-HT receptors are expressed on osteoblasts(51,52) as well as osteoclasts, and these cells interact to regulate bone mass,(1,6) the net effect of the serontonergic system on bone mass is likely to be complex in vivo. In this regard, leptin potently inhibits appetite as well as bone formation through ventromedial hypothalamic neurons and β-adrenergic receptors on osteoblasts.(53,54) Elevated leptin levels correlate with elevations in brain 5-HT levels in humans,(55) and inhibition of 5-HT synthesis or blockade of the 5-HT (2B/2C) receptors reduces the anorexigenic effects of leptin.(56) Further studies are clearly needed to determine the role of the various components of the serotoninergic system as well as interactions with other central regulatory systems on bone in intact animals.

Functional polymorphisms in the 5-HTT gene, (5-HTTLPR, VNTR) have been associated with personality traits,(57) hyperkinetic disorders, schizophrenia and unipolar depression,(58) and clinical response to SSRIs.(59) Clinically, the use of SSRIs for the treatment of depression and other psychological disturbances is widespread and is increasing. Reductions in BMD have been reported in patients suffering from depression syndromes that seem to be related to derangements in serotoninergic function.(60–65) An evaluation of possible links between 5-HTT polymorphisms, the use of antidepressants, and effects on skeletal homeostasis is therefore timely.


The authors thank Dr Leslie Rae Morse and Dr Mike Levin for insightful comments on the manuscript.