Bone Vol. 25, No. 6
December 1999:623– 629
ORIGINAL ARTICLES
Evidence for a Dense and Intimate Innervation of the Bone
Tissue, Including Glutamate-Containing Fibers
C. M. SERRE, D. FARLAY, P. D. DELMAS, and C. CHENU
INSERM Unit 403, Faculté de Medecine RTH Laënnec and Hôpital E. HERRIOT, Lyon, France
The recent demonstration in bone cells of receptors for
glutamate (Glu), a major neuromediator, suggests that Glu
may also act as a signaling molecule in bone and regulate
bone cell metabolism. Although bone is known to be innervated, the distribution and characteristics of nerve fibers in
this tissue have not been well documented. We have studied
the anatomical distribution of nerve fibers and the presence
of glutamate-immunoreactive ones in sections of long bones
from neonatal, 15-, and 25-day-old rats, using immunocytochemistry with antibodies directed against several neuronal
markers and Glu. We showed by electron microscopy that
bone is rich in nerve-like processes running along vessels
adjacent to bone trabeculae, in the vicinity of hematopoı̈etic
cells and bone cells. Immunocytochemical studies at the
tissue and cellular level confirmed the presence of a dense
network of thin nerve processes immunolabeled for neurofilament 200, tyrosine hydroxylase, and microtubule associated protein-2, three markers of nerve fibers. Some of these
nerve processes showed local dilatations in contact with
medullary cells and bone cells that were immunolabeled for
synaptophysin, a nerve terminal marker. Glu was largely
expressed in these thin nerve processes in proximity to bone
cells. These findings show evidence for a dense and intimate
network of nerve processes in bone, some of which were
containing Glu, suggesting glutamatergic innervation in
bone. (Bone 25:623– 629; 1999) © 1999 by Elsevier Science
Inc. All rights reserved.
demonstrated in bone, indicating the presence of both sympathetic and sensory fibers.2,3,12,14,16,17,19,33 These studies have
shown a high degree of peptidergic innervation in bone tissue in
regions of high osteogenic activity. However, immunocytochemistry using antibodies directed against specific neuronal markers
has not been performed at the cellular level in bone, and the
distribution of nerve fibers in the direct vicinity of bone cells is
still not well known.
Glutamate (Glu) is a major neuromediator of both the central
and peripheral nervous system.26 The recent identification of a
neuronal glutamate transporter in bone,25 as well as the demonstration of the expression of different subtypes of Glu receptors
by both bone resorbing osteoclasts and bone forming osteoblasts,7,31 suggest that Glu might be an important local regulator
of bone cell functions. The origin of Glu in bone is unknown, and
one hypothesis is that its presence is due to innervation of this
tissue. Clinical observations,10,11,15 as well as experimental and
in vitro studies,4,18,20,22,23 have already established the involvement of the peripheral nervous system in the regulation of bone
development and bone remodeling.
The purpose of this study was to examine the anatomical
distribution of nerve fibers in bone and to identify Glu-immunoreactive fibers. We used immunocytochemistry on long bone
sections from neonatal, 15- and 25-day-old rats, with antibodies
directed against several specific neuronal markers and Glu. We
have demonstrated the presence of a dense network of nerve
processes in the vicinity of bone cells, some of which contained
Glu, suggesting the existence of a glutamatergic innervation in bone.
Key Words: Nerve fibers; Rat long bone sections; Immunocytochemistry; Electron microscopy; Neurofilament 200;
Glutamate.
Materials and Methods
Animals and Tissue Processing
Introduction
Neonatal (few hours), 15- and 25-day-old Wistar rats were used
for this study. Animals were killed, tibia and femurs were rapidly
excised, roughly cleaned of adherent tissues and immediately
immersed in 1% glutaraldehyde in 0.1 M sodium cacodylate
buffer, pH 7.3. Bones were cut transversely at the diaphysis
level, then longitudinally, and fixed for 24 h.27 After several
washings in sodium cacodylate buffer, all specimens were postfixed with potassium ferrocyanide-reduced osmium tetroxyde for
2 h.30 Undecalcified samples were immediately dehydrated
through a graded ethanol series and embedded in Epon, whereas
some samples were partially decalcified in 4.13% disodium
ethylenediamine-tetra-acetic acid (EDTA) for 5 days35 before
dehydration and embedding in Epon.
Semithin sections (0.8 –1 ␮m) were cut on a Reichert Ultracut
E Microtome (Leica, Leitz, Austria). Some sections were stained
with azur-methylene blue to enable selection of samples for
Pioneer studies, using light microscopy with methylene blue
staining1,29 or silver impregnation5,6,24,34 have demonstrated that
bone is innervated. Nevertheless, the distribution of nerve fibers
in bone has not been well documented, and the discrimination of
fibers representing the sensory, sympathetic, and parasympathetic nervous system was not possible with conventional histological techniques. Duncan and Shim8 have later shown the
presence of sympathetic nerve fibers in bone tissue by visualizing noradrenaline. More recently, nerve fibers immunoreactive
for neuropeptides and tyrosine hydroxylase (TH) have been
Address for correspondence and reprints: Dr. C. Chenu, INSERM Unit
403, Hôpital E. HERRIOT, Pavillon F, 69 437 Lyon Cedex 03, France.
E-mail: chenu@lyon151.inserm.fr
© 1999 by Elsevier Science Inc.
All rights reserved.
623
8756-3282/99/$20.00
PII S8756-3282(99)00215-X
624
C. M. Serre et al.
Glu-containing nerve processes in bone
Bone Vol. 25, No. 6
December 1999:623– 629
Figure 1. Electron micrographs illustrating
nerve fibers and nerve processes in growing rat
long bones. (a) Distal metaphysis of 15-day-old
rat tibia in the vicinity of the afferent vascular
tree. Part of a large nerve bundle containing an
unmyelinated axon (arrow) lined with a Schwann
cell (S) is associated to the afferent vessel (AV).
Inset: enlargement showing the structure of the
axon; microtubules (arrow head) are observed.
(b) Endosteum of 25-day-old rat tibia. A nerve
fiber is observed close to an osteoblast (Ob).
Numerous unmyelinated axons (arrows) are surrounded by a Schwann cell (S1), whereas another
Schwann cell (S2) shows a unique myelinated
axon (asterisk). (c) Borderline between metaphysis and diaphysis of 15-day-old rat tibia. An
unmyelinated nerve fiber is present close to hematopoı̈etic cells (H) in bone marrow. A
Schwann cell (S) contains several axons (arrows). (d) Metaphysis of neonatal rat femur. A
thin nerve process is running between hematopoı̈etic cells (H) and shows an enlargement (large
arrow) in the vicinity of cells. (e) Metaphysis of
15-day-old rat tibia. A nerve process (large arrow) containing microtubules (arrows) and microfilaments (arrow head) is observed in contact
with an osteoblast (Ob).
immunocytochemistry at tissue or cellular levels as well as for
transmission electron microscopy (TEM).
For TEM, thin sections (80 nm to 100 nm) were cut and
stained with uranyl acetate and lead citrate. Sections were examined using a Jeol 1200EX electron microscope (Jeol Ltd.,
Akishina, Tokyo, Japan) at 80 kV.
Immunocytochemistry
Antibodies. The monoclonal antibodies directed against neurofilament 200 (NF 200), synaptophysin and the microtubule associated protein-2 (MAP2) were obtained from Sigma (St Louis,
MO). The rabbit anti-tyrosine hydroxylase (TH) and anti-glutamate polyclonal antibodies were purchased from Chemicon (Temecula, CA). The antibody used in this study recognizes free
Glu. The polyclonal antibody directed against human ␣I (1)
carboxy-telopeptide of type I collagen (LF-67) was provided by
Dr. L. Fisher (NIH, Besthesda, MD). The polyclonal antibody
directed against rat type III collagen was given by Dr. Hartmann
(Novotec, Lyon, France). The anti-VIP polyclonal antibody was
a gift from Dr. J. L. Saffar (Paris, France). Controls were
performed by using a monoclonal antibody directed against the
Kappa chains of immunoglobulins (MOPC21) or purified immunoglobulins G from nonimmune rabbit serum.
Semithin sections. Sections from decalcified samples were
laid on sylanized slides (aminopropyltriethoxyalsilane, 2% in
acetone) and dried overnight at 37°C. Epon was removed with
13.3% (w/v) potassium hydroxide in a methanol/propylene
oxyde (2/1) mixture and rehydrated. Sections were treated for 20
min with 100 mM glycine and 50 mM ammonium chloride in
Tris buffer, pH 7.6 in order to saturate free aldehydic groups.
Endogenous peroxydase activity was inhibited by incubation for
15 min with 1% sodium azide and 1.5% H2O2 in 50% methanol,
and nonspecific immunoreactions were blocked for 30 min with
10% normal goat serum (NGS) in Tris buffer saline (TBS)
containing 0.01% bovine serum albumin (BSA), according to
Ingram et al.21 Sections were incubated overnight with primary
antibodies diluted in TBS containing 1% NGS (dilutions of
antibodies: anti-NF200: 30 ␮g/mL; anti-MAP2: 1 ␮g/mL; antisynaptophysin: 6 ␮g/mL, anti-Glu: 1/1000; anti-TH: 1/15000;
anti-VIP: 1/2000; LF-67: 1/5000; anti-type III collagen (batch
217): 1/1000). Control sections were incubated with nonimmune
rabbit serum or with the monoclonal antibody MOPC21 which
has no known hapten or antigen binding activity. Antigenantibody complexes were detected with a peroxydase streptavidin system (Dako, Copenhagen, Denmark) and revealed with
3-3⬘ diaminobenzidine tetrahydrochloride (DAB) (Sigma) in Tris
buffer containing 0.01% H2O2. All washings were done with
Bone Vol. 25, No. 6
December 1999:623– 629
C. M. Serre et al.
Glu-containing nerve processes in bone
625
Figure 2. Immunohistochemical localization of several nerve markers in long
bones of growing rats. (a) Metaphysis of
neonatal rat femur. NF 200-positive nerve
processes (arrows), showing local enlargements (arrow heads), are running
along bone trabaeculae (t) and vessels
(V). They are close to bone cells (Oc:
osteoclast, Ob: osteoblast) and hematopoı̈etic cells (H). Original magnification:
⫻1000. (b) Deep metaphysis of neonatal
rat femur at the level of the afferent vessel. Tyrosine hydroxylase is present in
these nerve processes (arrows) showing
local dilatations (arrow heads) in contact
with hematopoı̈etic cells (H). Oc: osteoclast. Original magnification: ⫻1500.
(c) Deep metaphysis of neonatal rat femur. MAP2-positive nerve processes
showing dilatations (arrow heads) in contact with hematopoı̈etic cells (H) and osteoclasts (Oc). Original magnification:
⫻1000. (d) Diaphysis of 15-day-old rat
tibia. NF200 is present in nerve endings
(arrow heads) in contact with hematopoı̈etic cells (H). Original magnification:
⫻1000.
TBS containing 0.02% Triton X100. Sections were counterstained with Meyer’s hematoxylin (diluted 1:3 in distilled water),
dehydrated, and mounted in Xam (Gurr-BDH Laboratory, Poole,
UK).
Ultrathin sections. Sections from decalcified and undecalcified samples were collected on nickel grids covered with a
formwar film. Grids were floated for 10 min on a drop of 50 mM
glycine in 0.05 M Tris buffer pH 7.4, transferred for 30 min on
a drop of Tris buffer containing 1% BSA and 1% NGS, then
Figure 3. Immunolabelings for NF 200
and type I or III collagen are mutually
exclusive in long bone sections. (a) and
(b) Consecutive sections of neonatal rat
femur (metaphysis). Type I collagen (a)
is visualized in bone matrix, osteoblasts
(Ob) and preosteoblasts (POb) (arrows).
NF200-positive nerve processes (b) are
running in bone marrow close to osteoblasts (Ob), hematopoı̈etic cells (H) and
vessels (arrow heads). The labelling
profile obtained in (b) is completely distinct from that obtained in (a). Original
magnification: ⫻1000. (c) and (d) Consecutive sections of neonatal rat femur
(metaphysis). Type III collagen (c) is
observed in the remaining cartilage matrix of the bone trabeculae (double arrow) and in collagen fibrils in the bone
marrow (arrows). NF200-positive nerve
processes (arrow heads) are located near
bone cells and vessels. The labeling profile obtained in (d) is completely distinct
from that obtained in (c). Original magnification: ⫻400.
incubated for 90 min at room temperature on a drop of the
specific primary antibodies (dilutions of antibodies: antiNF200: 60 ␮g/mL; anti-MAP2: 1 ␮g/mL; anti-TH: 1/5000;
anti-Glu: 1/500). After incubation, sections were rinsed on
drops of Tris buffer/BSA, placed on biotinyled anti-mouse or
anti-rabbit antibodies (dilution: 1/300 in Tris buffer/BSA, pH
7.4), then laid on gold streptavidin (Jansen, Tebu, France) at
pH 8.2. After careful washing, grids were transferred to 1%
glutaraldehyde, rinsed in distilled water, and conventionally
626
C. M. Serre et al.
Glu-containing nerve processes in bone
Bone Vol. 25, No. 6
December 1999:623– 629
stained with uranyl acetate and lead citrate before examination
by TEM.
Some semithin sections from decalcified samples were immunolabeled for NF200 without Meyer’s hematoxylin counterstaining. These sections were dehydrated in ethanol, impregnated
with a mixture of propylene oxyde-Epon (v/v), and re-embedded
in Epon. Ultrathin sections were cut and observed without
counterstaining.
Results
Localization of Nerve Fibers in Long Bones of Growing Rats
Electron microscopy studies of long bones from neonatal, 15and 25-day-old rats revealed the presence of nerve fibers either in
the vicinity of the periosteum and surrounding connective tissues
or in the muscles attached to it. Some of them were running along
veins and arteries. The diaphyseal nerve trunk, entering bone
along nutrient vessels, comprised mainly nonmyelinated fibers in
15-day-old rats (Figure 1a). In the periosteum of 15- and
25-day-old rats, nerve fibers were generally unmyelinated,
whereas in the endosteum, some myelinated fibers were observed
in proximity to osteoblasts, more numerous in 25-day-old rats
(Figure 1b). In bone marrow, only unmyelinated nerve fibers
running between hematopoı̈etic cells were demonstrated (Figure
1c). Myelinated nerve fibers were never present in the endosteum
or bone marrow of neonatal rats.
In addition to these sparse morphologically clearly identified
nerve fibers, abundant thin nerve-like processes were observed in
long bone sections. These ran along vessels adjacent to bone
trabeculae in the vicinity of hematopoı̈etic and bone cells and
often showed enlarged endings in close contact with these cells.
They were always devoid of basement membrane. Neither
Schawnn cells nor myelin sheath were observed at their periphery. The structural features of these nerve-like processes were
slightly different for neonatal and 15-day-old rats. In neonatal
rats, nerve like processes contained only hyaloplasm and rare
organelles (Figure 1d). In older rats, more organelles were
observed, such as small mitochondria, glycogen particles, and
dense granules. Microfilaments, microtubules, or microvesicles,
usually abundant in nerve fibers, were often visible in these cell
processes (Figure 1e). In contrast, they contained no golgi
apparatus and no endoplasmic tubular structures, generally described in stromal cells.
Immunoreactivity of these Nerve Processes For Several
Specific Neuronal Markers
An antibody directed against NF200, used as a general nerve
marker, revealed that the epiphysis of neonatal rats consisted
only of cartilage at this stage of development, and was devoid of
innervation. In contrast, in all the metaphysis of neonatal rats and
in the diaphysal area containing the nutrient vessels, a dense
innervation was observed that appeared as a prominent network
of thin cytoplasmic cell processes in close relation with blood
vessels (Figure 2a). Nerve endings, corresponding to dilatations
of these nerve processes, were found in contact with medullary
cells and bone cells (Figure 2a). Hematopoı̈etic cells, endothelial
cells, and bone cells were always negative for this marker in any
part of the bone. These nerve processes were also labeled for TH
(Figure 2b) and MAP2 (Figure 2c), two other markers of nerve
fibers. Control sections, in which specific antibodies were replaced by MOPC21 or by purified immunoglobulins G from
rabbit, showed no specific immunoreactivity (data not shown). In
femurs and tibia of 15- and 25-day-old rats, immunolabeling for
neuronal markers stained the same network of nerve processes
Figure 4. Electron micrographs of rat long bone sections after immunocytochemical labeling for various nerve markers. (a,b,c,d) Immunolabelings of neonatal rat femur (a,c,d) or 15 day-old rat tibia (b) metaphysis, followed by biotin streptavidin-gold complex staining. For NF200
(a,b), TH (c), and MAP2 (d), gold particles (arrows) are localized in
nerve processes running in bone marrow close to preosteoblasts (pOb), or
in nerve endings (containing microvesicle: asterisk) in contact with
hematopoı̈etic cell (H). The inset in 4c points out gold particules (arrow).
Bars: 200 nm. (e) Low magnification electron micrograph of a reembedded semithin section immunostained for NF200. Positive profiles (arrows) contain few organelles and were morphologically similar to those
observed on ultrathin sections immunolabeled with gold particles (Fig.
4a,b).
adjacent to blood vessels, bone marrow and bone cells. Its
distribution was slightly different for these later stages of development, because nerve processes were observed in the epiphysis
in the secondary ossification center. They were localized in
proximity to bone cells lining trabeculae facing the growth plate.
The deeper part of the epiphysis appeared less innervated. In the
metaphysis immediately under the growth plate, nerve processes
were sparse. Farther away from the growth plate, they formed an
expanded network between bone trabeculae, although the intensity of the labeling was not homogeneous from one trabecula to
another. In the diaphysis, nerve processes were seen under
cortical bone as well as in periosteocytic lacunae; a few nerve
Bone Vol. 25, No. 6
December 1999:623– 629
C. M. Serre et al.
Glu-containing nerve processes in bone
627
Figure 5. Immunocytochemical localizations of synaptophysin and VIP
in long bones of growing rats. (a) Metaphysis of neonatal rat femur.
Synaptophysin is present in enlargements of nerve processes in contact
with blood vessels (arrows), osteoclasts (asterisk) and hematopoı̈etic
cells (arrow heads). (b) Diaphysis of 25-day-old rat tibia. Synaptophysin
is present in nerve endings (arrow heads) in contact with osteoblasts
underlaying cortical bone (c). Some periosteocytic lacunae contain positive
endings (arrow) in contact with osteocytes. (c) and (d) Consecutive sections
of neonatal rat femur metaphysis, near afferent vessels. Synaptophysin (c) is
visualized in nerve processes close to hematopoı̈etic cells (arrows) and in the
wall of the artery (a) (arrow heads). VIP (d) is coexpressed in the same nerve
profiles. Original magnification: ⫻1000.
endings were also evident in the center of the diaphysis, in
contact with hematopoı̈etic cells (Figure 2d).
The distribution of nerve markers was different from those of
type I and III collagens. Immunolabelings for NF200 and type I
or type III collagen, performed on consecutive bone sections,
were found to be mutually exclusive. Type I collagen was
visualized in bone matrix, in osteoblasts, in some stromal cells
probably preosteoblasts (Figure 3a), but never in cell processes
immunolabeled for NF200 (Figure 3b). Type III collagen was
present in the remaining cartilage matrix of the bone trabeculae
and in extracellular collagen fibrils abundant at stromal cell
periphery (Figure 3c), but was not expressed in the network of
cell processes containing NF200 (Figure 3d).
The electron-microscopic analysis, using colloidal gold immunolabeling for neuronal markers, confirmed that nerve processes were expressing NF200 (Figure 4a,b), TH (Figure 4c),
and MAP2 (Figure 4d). The labeling was abundant in nerve
processes running in bone marrow and in their widened parts in
contact with hematopoı̈etic cells (Figure 4a) and bone cells
(Figure 4b,c,d). A low magnification electron micrograph of a
reembedded semithin section immunostained for NF200 (Figure
4e) showed that the immunolabeled profiles contained rare organelles and were morphologically similar to those observed on
ultrathin sections immunostained for NF200 with colloidal gold
(Figure 4a,b).
Synaptophysin, one of the major proteins present in synaptic
vesicles, was detected in long bone sections. The tissue distribution of this protein was similar to other nerve markers, but not
identical. Labeling for synaptophysin was more restricted and
mainly localized to short nerve structures in contact to bone cells,
endothelial, and bone marrow cells (Figure 5a,b,c). VIP immunoreactivity was also found in bone and the tissue distribution of
this neuropeptide was identical to synaptophysin (Figure 5d).
Figure 6. Immunocytochemical localization of glutamate in long bones
of growing rats. (a) Metaphysis of neonatal rat. Glu (arrows) is present in
nerve processes running in bone marrow and showing enlargements in
contact with hematopoı̈etic cells (H). Original magnification: ⫻800. (b,c)
Electron micrographs of 15-day-old rat tibia sections immunolabeled for
Glu. Glu (arrows) is observed in nerve processes in the vicinity of
osteoblasts (Ob) as well as in the afferent nerve trunk. S: Schwann cell;
A: unmyelinated axon.
628
C. M. Serre et al.
Glu-containing nerve processes in bone
Expression of Glutamate in Long Bones of Growing Rats
In the metaphysis and diaphysis of neonatal rat femurs, nerve
processes, running in bone marrow close to bone cells and blood
vessels, contained Glu (Figure 6a). Immunolabeling for Glu was
particularly abundant in the endosteal part of the metaphysis and
was localized in the same expanded network of thin processes
previously shown to be immunoreactive for nerve markers. A
few other cells, mainly osteoblastic cells, also contained Glu in
varying amounts, whereas osteoclasts, endothelial, and hematopoı̈etic cells were always negative.
In 15- and 25-day-old rat tibia, Glu had also the same tissue
distribution as neuronal markers, highly expressed in nerve
processes in proximity to blood vessels and bone cells.
The electron microscopic immunocytochemistry analysis
confirmed that Glu was largely present in nerve processes in the
vicinity of or in contact with bone cells (Figure 6b), as well as in
some axons of the afferent nerve trunk (Figure 6c).
Discussion
The present study has demonstrated the existence, in long bones
of neonatal, 15- and 25-day-old rats, of an extensive network of
cytoplasmic cell processes immunostained for several nerve
markers. NF200 is a general neuronal marker specific of neurofilaments present in cells and tissues of neuronal origin. Immunoreactivity for NF200 was found in thin cell processes showing
some dilatations that were not distributed uniformly and were
more numerous when close to blood vessels. To further confirm
the identity of these nerve processes, antibodies directed against
other neuronal markers such as TH, synaptophysin, and MAP2
were used. Immunostaining for TH, a commonly accepted
marker for noradrenergic fibers,3 revealed the occurrence of
sympathetic nerve processes in this network, confirmed by the
presence of VIP-immunoreactive nerve profiles, as shown previously.19 Synaptophysin, a nerve terminal marker, was mainly
demonstrated in varicosities of some short nerve processes generally located in the vicinity of, or in contact with, blood vessels,
bone cells, or hematopoı̈etic cells. MAP2-immunoreactive nerve
processes were also present in rat long bones and could be
sensory nerve processes because this neuronal marker is primarily expressed in the dendrosomatic compartment and excluded
from the axons of most nerve fibers.13 Our results confirm the
presence of both sensory and sympathetic nerve fibers previously
documented in bone and suggest that these two types of innervation may form parallel networks with a close distribution in
bone tissue. Immunoreactivities for these specific nerve markers
were only observed in this network of particular cell processes
and in well-defined nerve fibers. Stromal cells expressing type I
and III collagens, osteoblastic cells, hematopoı̈etic cells, endothelial cells, and osteoclasts were always negative for each of
these neuronal markers, as demonstrated by light and
electron microscopy.
This dense network of thin varicose nerve processes running
with medullary sinusoids deeply in metaphysis and ending in
contact with bone cells or medullary cells constitutes a novel
observation not emphasized in previous studies. Morphologically
well-defined unmyelinated and myelinated nerve fibers, as described previously,6,28 were also observed in 15- and 25-day-old
rat long bone sections, but not in neonatal bones. In contrast, the
network of nerve-like processes was already present in neonatal
rats and the identification of TH, VIP, and MAP2-immunoreactive processes in this network confirmed that rat bone is already
supplied at birth with sympathetic and sensory nerve fibers.32
This localization of nerve processes at proximity to bone and
medullary cells is indicative of a possible role of nerve supply in
Bone Vol. 25, No. 6
December 1999:623– 629
the regulation of bone cell activities, in agreement with clinical
observations in patients with neurological disorders and experimental denervation studies.10,11,15 However, although some of
these nerve terminals were in contact with bone or bone marrow
cells, our electron-microscopic observations did not detect typical synapses in the nerve endings which did not contain many
vesicles. Hara-Irie et al.14 have also demonstrated the presence of
CGRP-positive nerve fibers in close contact with osteoclasts, at
epiphysal trabeculae facing the growth plate of rat femurs; these
structures were not typical synapses in agreement with our own
findings. Further studies will be needed to clarify the presence of
synaptophysin in bone in the absence of synaptic structures.
The nature of neurotransmitters present in bone has not been
clearly identified. In addition to noradrenergic sympathetic nerve
fibers, various neuropeptides used by either sensory or postganglionic sympathetic nerve fibers have been demonstrated in
bone, suggesting a possible role for these neuromediators in
skeletal metabolism.19,20,22,23
This is the first study to reveal glutamate-immunoreactive
nerve processes in bone. Glu is the primary excitatory neuromediator in both the central nervous system and the peripheral
nervous system.26 We have shown that Glu was widely expressed in nerve processes running in bone marrow in the
vicinity of bone or bone marrow cells, suggesting that it might
also act as a neurotransmitter in bone. In addition, Glu was
detected in some osteoblasts and preosteoblasts, probably in
relation to cell metabolism. Our observations have clearly demonstrated some contacts between Glu-immunoreactive nerve terminals and bone cells, but no synaptic vesicles were observed in
these structures. Although our study did not show any evidence
for a release of Glu from these nerve terminals, we can consider
that significant local concentrations of Glu might be present in
the vicinity of bone cells.
Glu acts on a variety of different ligand-gated ion cell surface
receptors, classified in two categories, mainly, NMDA receptors
and non-NMDA receptors. These receptors, as well as glutamate
transporters, have been shown to be expressed by bone
cells.7,25,31 Furthermore, we have recently demonstrated that
NMDA subtype Glu receptors expressed by osteoclasts are
functional and that Glu elicits current in these cells.7,9
The cellular origin of Glu and the mechanisms leading to its
release in bone are unknown. One proposed hypothesis is that
Glu is released from bone cells and has a role in paracrine
intercellular communications in bone.25 Another possibility is
that Glu, which is the most largely distributed neuromediator in
the nervous system, is released from nerve fibers present in the
vicinity of bone cells. Our results, which demonstrate in long
bones of growing rats an extensive network of nerve processes
containing Glu in proximity to bone cells, strengthen this last
hypothesis, and provide initial data for the existence of a glutamatergic innervation of bone.
Acknowledgments: The authors would like to thank G. Boivin and L.
Malaval for helpful comments on the manuscript. Electron microscopic
studies were performed in the Centre de Microscopie Electronique et de
Pathologie Ultrastructurale, Faculté de Medecine RTH Laënnec, Lyon,
France. Part of these results were presented as an abstract at the second
joint meeting of the American Society For Bone and Mineral Research
and the International Bone and Mineral Society, December 1998, San
Francisco, CA.
References
1. Arthur, R. P. and Shelley, W. B. The technology of in vitro staining of nerves
in human skin with thiazin dyes. J Invest Dermatol 33:121–144; 1959.
Bone Vol. 25, No. 6
December 1999:623– 629
2. Bjurholm, A., Kreicbergs, A., Brodin, E., and Schultzberg, M. Substance Pand CGRP-immunoreactive nerves in bone. Peptides 9:165–171; 1988.
3. Bjurholm, A., Kreicbergs, A., Terenius, L., Goldstein, M., and Schultzberg, M.
Neuropeptide Y-, tyrosine hydroxylase- and vasoactive intestinal polypeptideimmunoreactive nerves in bone and surrounding tissues. J Auton Nerv Syst
25:119 –125; 1988.
4. Bjurholm, A., Kreicbergs, A., Schultzberg, M., and Lerner, U. H. Neuroendocrine regulation of cyclic AMP formation in osteoblastic cell lines (UMR
106-01, ROS 17/2.8, MC3T3-E1 and SAOS-2) and primary bone cells. J Bone
Miner Res 7:1011–1019; 1992.
5. Calvo, W. The innervation of the bone marrow in laboratory animals. Am J
Anat 123:315–328; 1968.
6. Calvo, W. and Forteza-Vila, J. On the development of bone marrow innervation in new-born rats as studied with silver impregnation and electron microscopy. Am J Anat 126:355–372; 1969.
7. Chenu, C., Serre, C. M., Raynal, C., Burt-Pichat, B., and Delmas, P. D.
Glutamate receptors are expressed by bone cells and are involved in bone
resorption. Bone 22:295–299; 1998.
8. Duncan, C. P. and Shim, S. S. The autonomic nerve supply of bone. J Bone Jt
Surg 53:323–330; 1977.
9. Espinosa, L., Itzstein, C., Delmas, P. D., and Chenu, C. Active NMDA
glutamate receptors are expressed by mammalian osteoclasts. J Physiol 518:
47–53; 1999.
10. Freehafer, A. A. and Mast, W. A. Lower extremity fractures in patients with
spinal-cord injury. J Bone Jt Surg 47:683– 694; 1965.
11. Gillespie, J. A. The nature of bone changes associated with the nerve injuries
and diseases. J Bone Jt Surg 36:464 – 473; 1963.
12. Goto, T., Yamaza, T., Kido, M. A., and Tanaka, T. Light-and electron
microscopic study of the distribution of axons containing substance P and the
localization of neurokinin-1 receptor in bone. Cell Tissue Res 293:87–93;
1998.
13. Guo, X., Rueger, D., and Higgins, D. Osteogenic protein-1 and related bone
morphogenetic proteins regulate dendritic growth and the expression of microtubule-associated protein-2 in rat sympathetic neurons. Neurosci Lett 245:131–
134; 1998.
14. Hara-Irie, F., Amizuka, N., and Ozawa, H. Immunocytochemical and ultrastructural localization of CGRP-positive nerve fibers at the epiphyseal trabecules facing the growth plate of rat femurs. Bone 18:29 –39; 1996.
15. Hardy, A. G. and Dickson, J. W. Pathological ossification in traumatic
paraplegia. J Bone Jt Surg 45:76 – 87; 1963.
16. Hill, E. L. and Elde, R. Calcitonin gene-related peptide-immunoreactive nerve
fibers in mandibular periosteum of rats: Evidence for primary afferent origin.
Neurosci Lett 85:172–178; 1988.
17. Hill, E. L. and Elde, R. Distribution of CGRP-, VIP-, D beta H-, SP-, and
NPY-immunoreactive nerves in the periosteum of rats. Cell Tissue Res 264:
469 – 480; 1991.
18. Hill, E. L., Turner, R., and Elde, R. Effects of neonatal sympathectomy and
capsaicin treatment on bone remodelling in rats. Neuroscience 44:747–755;
1991.
19. Hohmann, E. L., Elde, R. P., Rysavy, J. A., Einzig, S., and Gebhard, R. L.
Innervation of periosteum and bone by sympathetic vasoactive intestinal
peptide-containing nerve fibers. Science 232:868 – 871; 1986.
20. Hukkanen, M., Kontinnen, Y. T., Santavirta, S., et al. Rapid proliferation of
C. M. Serre et al.
Glu-containing nerve processes in bone
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
629
CGRP-immunoreactive nerves during healing of rat tibial fracture suggests
neural involvement in bone growth and remodeling. Neuroscience 54:969 –979;
1993.
Ingram, R. T., Clarke, B. L., Fisher, L. W., and Fitzpatrick, L. A. Distribution
of noncollagenous proteins in the matrix of adult human bone: Evidence of
anatomic and functional heterogeneity. J Bone Miner Res 8:1019 –1027; 1993.
Konttinen, Y. T., Imai, S., and Suda, A. Neuropeptides and the puzzle of bone
remodeling. Acta Orthop Scand 67:632– 639; 1996.
Lerner, U. H. Kinins and neuropeptides. In: Bilezikian, J. P., Raisz, L. G., and
Rodan, G. A., Eds. Principles of Bone Biology. San Diego, CA: Academic;
1996; 581–596.
Linder, J. E. A simple and reliable method for silver impregnation of nerves in
paraffin sections of soft and mineralized tissues. J Anat 127:543–551; 1978.
Mason, D. J., Suva, L. J., Genever, P. G., Patton, A. J., Steuckle, S., Hillam,
R. A., and Skerry, T. M. Mechanically regulated expression of a neural
glutamate transporter in bone: A role for excitatory amino acids as osteotropic
agents. Bone 20:199 –205; 1997.
Mayer, M. L. and Westbrook, G. L. The physiology of excitatory amino acids
in the vertebrate central nervous system. Prog Neurobiol 28:197–276; 1987.
McKee, M. D., Nanci, A., Landis, W. J., Gotoh, Y., Gerstenfeld, L. C., and
Glimcher, M. J. Effects of fixation and demineralization on the retention of
bone phosphoproteins and other matrix components as evaluated by biochemical analyses and quantitative immunocytochemistry. J Bone Miner Res 6:937–
945; 1991.
Milgram, J. W. and Robinson, R. A. An electron microscopic demonstration of
unmyelinated nerves in the haversian canals of the adult dog. Bull John Hopk
Hosp 117:163–173; 1965.
Miller, M. R. and Kasahara, M. Observations on the innervation of human long
bones. Anat Rec 145:13–23; 1963.
Neiss, W. F. Electron staining of the cell surface coat by osmium-low
ferrocyanide. Histochemistry 80:231–242; 1984.
Patton, A. J., Genever, P. G., Birch, M. A., Suva, L. J., and Skerry, T. M.
Expression of an N-Methyl-D-Aspartate-type receptor by human and rat
osteoblasts and osteoclasts suggests a novel glutamate signaling pathway in
bone. Bone 22:645– 649; 1998.
Sisask, G., Bjurholm, A., Ahmed, M., and Kreicbergs, A. The development of
autonomic innervation in bone and joints of the rat. J Auton Nerv Syst
59:27–33; 1996.
Tabarowski, Z., Gibson-Berry, K., and Felten, S. Y. Noradrenergic and peptidergic innervation of the mouse femur bone marrow. Acta Histochem (Jena)
98:453– 457; 1996.
Thurston, T. J. Distribution of nerves in long bones as shown by silver
impregnation. J Anat 134:719 –728; 1982.
Warshawsky, H. and Moore, G. A technique for the fixation and decalcification
of the rat incisors for electron microscopy. J Histochem Cytochem 15:542–549;
1967.
Date Received: December 1, 1998
Date Revised: August 2, 1999
Date Accepted: August 3, 1999