The gut microbiota regulates bone mass in mice.

Abstract

The gut microbiota modulates host metabolism and development of immune status. Here we show that the gut microbiota is also a major regulator of bone mass in mice. Germ-free (GF) mice exhibit increased bone mass associated with reduced number of osteoclasts per bone surface compared with conventionally raised (CONV-R) mice. Colonization of GF mice with a normal gut microbiota normalizes bone mass. Furthermore, GF mice have decreased frequency of CD4(+) T cells and CD11b(+) /GR 1 osteoclast precursor cells in bone marrow, which could be normalized by colonization. GF mice exhibited reduced expression of inflammatory cytokines in bone and bone marrow compared with CONV-R mice. In summary, the gut microbiota regulates bone mass in mice, and we provide evidence for a mechanism involving altered immune status in bone and thereby affected osteoclast-mediated bone resorption. Further studies are required to evaluate the gut microbiota as a novel therapeutic target for osteoporosis.

Full text

Page 1

The Gut Microbiota Regulates Bone Mass in Mice
Klara Sjo ¨gren,1Cecilia Engdahl,1Petra Henning,1Ulf H Lerner,1,2Valentina Tremaroli,3
Marie K Lagerquist,1Fredrik Ba ¨ckhed,3and Claes Ohlsson1
1Centre for Bone and Arthritis Research, Institute of Medicine, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden
2Department of Molecular Periodontology, Umea ˚ University, Umea ˚, Sweden
3Center for Cardiovascular and Metabolic Research/Wallenberg Laboratory, Sahlgrenska Academy at University of Gothenburg,
Gothenburg, Sweden
ABSTRACT
The gut microbiota modulates host metabolism and development of immune status. Here we show that the gut microbiota is also a
major regulator of bone mass in mice. Germ-free (GF) mice exhibit increased bone mass associated with reduced number of osteoclasts
per bone surface compared with conventionally raised (CONV-R) mice. Colonization of GF mice with a normal gut microbiota normalizes
bonemass.Furthermore,GFmicehavedecreasedfrequencyofCD4þTcellsandCD11bþ/GR1osteoclastprecursorcellsinbonemarrow,
which could be normalized by colonization. GF mice exhibited reduced expression of inflammatory cytokines in bone and bone marrow
compared with CONV-R mice. In summary, the gut microbiota regulates bone mass in mice, and we provide evidence for a mechanism
involving altered immune status in bone and thereby affected osteoclast-mediated bone resorption. Further studies are required to
evaluate the gut microbiota as a novel therapeutic target for osteoporosis. ? 2012 American Society for Bone and Mineral Research.
KEY WORDS: GERM FREE; OSTEOPOROSIS; OSTEOIMMUNOLOGY; SEROTONIN; BONE METABOLISM
Introduction
T
genesthanourhumangenome.(1)Thegutmicrobiotaisacquired
at birth and, although a distinct entity, it has clearly coevolved
with the human genome and can be considered a multicellular
organ that communicates with and affects its host in numerous
ways.(2)There are several different populations of endocrine cells
in the gut that can be affected by the gut microbiota, and
hormones released by the bacteria-stimulated endocrine cells
caninturn influence host function by entering circulation and by
direct communication with terminals of visceral afferent nerves
and immune cells.(3)Accordingly, the gut microbiota has many
metabolic functions and contributes to host adiposity.(4)It is
also of importance for maintaining immune homeostasis, ie,
protecting the host from invading pathogens while avoiding
immune responses to harmless commensal microbes.(5)Studies
have shown that germ-free (GF) animals have immature mucosal
immune systems with hypoplastic Peyer’s patches containing
few germinal centers and reduced number of IgA-producing
plasma cells and lamina propria CD4þT cells.(6)Furthermore, GF
mice have reduced number of CD4þT cells in the spleen and
he human gut is populated by trillions of bacteria, known
as gut microbiota, which collectively contain 150-fold more
fewerandsmallergerminalcenterswithinthespleen,suggesting
that the gut microbiota is capable of shaping systemic
immunity.(7,8)
The skeleton serves as a niche for mesenchymal and
hematopoietic progenitors. T-cell precursors generated from
hematopoietic stem cells in the bone marrow migrate to the
thymus where T-cell development occurs. Mature T cells then
recirculate between blood and the secondary lymphoid organs
and a large proportion migrate back to the bone marrow where
they reside together with other plasma cells.(9)The bone marrow
is a major reservoir for memory B and T cells.(10)Bone-forming
osteoblasts (OB) are derived from pluripotent mesenchymal
stromal cells, whereas bone-resorbing osteoclasts (OCL) are
derived from hematopoietic stem cells that also generate
immune cells. OCLs are specifically derived from the myeloid
lineage of hematopoietic cells, and it is the local microenviron-
ment that determines whether the myeloid precursor cell will
differentiate into a macrophage, a myeloid dendritic cell, or an
OCL.(11)The presence of macrophage colony-stimulating factor
(M-CSF) leads to increased proliferation and survival as well as
upregulated expression of receptor activator ofnuclear factor-kB
(RANK) in precursor cells. This allows RANK ligand (RANKL)
to bind and start the signaling cascade that leads to OCL
ORIGINAL ARTICLE
J JBMRBMR
Received in original form October 31, 2011; revised form February 8, 2012; accepted February 17, 2012. Published online March 9, 2012.
Address correspondence to: Klara Sjo ¨gren, PhD, Associate Professor, Institute of Medicine, Centre for Bone and Arthritis Research, Vita Stra ˚ket 11, SE-413 45
Gothenburg, Sweden. E-mail: Klara.Sjogren@medic.gu.se
Re-use of this article is permitted in accordance with the Terms and Conditions set out at http://wileyonlinelibrary.com/onlineopen#OnlineOpen_Terms.
Additional Supporting Information may be found in the online version of this article.
Journal of Bone and Mineral Research, Vol. 27, No. 6, June 2012, pp 1357–1367
DOI: 10.1002/jbmr.1588
? 2012 American Society for Bone and Mineral Research
1357

Page 2

formation.(12)M-CSF and RANKL are expressed by bone marrow
stromal cells as well as by periosteal and endosteal OBs in
response to hormones and cytokines stimulating bone resorp-
tion.(13)Tumor necrosis factor (TNF) is an inflammatory cytokine
produced by myeloid cells that promotes osteoclastogenesis.(14)
Studies have shown that interleukin 1 (IL-1) is a downstream
regulator of the effects of TNFa on inducing osteoclastogenesis
and joint damage.(15,16)The association between inflammation
and bone loss is well established, and in autoimmune disease
such as arthritis, osteoclastic bone resorption is driven by
cytokine-producing activated T cells.(17,18)Mice lacking T cells
(nude mice) have high, normal, or low bone mass depending on
the age of the mice when studied.(19–21)Conflicting data exist on
whether mice lacking T cells are protected against ovariectomy
(ovx)-induced bone loss.(19,21)To overcome compensatory
mechanisms that might be present in mice that lack T cells
resulting from a mutation, the effect of ovx in WT mice that were
depleted of T cells in vivo by treatment with anti-CD4 and anti-
CD8 antibodies were recently investigated. These mice were
foundtobeprotectedagainstovx-inducedboneloss,arguingfor
a role of T cells in this process.(22)
Serotonin (5-hydroxytryptamine, or 5-HT) is a hormone and a
neurotransmitter, and the majority of the circulating serotonin is
synthesized in the gut by the enterochromaffin cells.(23)Bone
cells express functional serotonin receptors, and gut-derived
serotonin has been demonstrated to have negative effects on
bone formation in mice.(24,25)The rate-limiting step in serotonin
synthesis in the gut is catalyzed by the enzyme tryptophan
hydroxylase-1 (Tph1).(26,27)The action of serotonin is limited by
reuptake into epithelial cells of the intestinal mucosa and
serotonergicneurones,whereitisbrokendown.(28)Thisuptakeis
mediated by the serotonin transporter SERT.(29)Recent studies
have reported conflicting data on the effects of serotonin on
bone. By treating mice with an inhibitor to Tph1 to decrease gut-
derived serotonin synthesis, two studies demonstrated that this
led to prevention and treatment of ovx-induced bone loss,
whereas one other study showed no effect on bone after similar
treatment.(30–32)
The importance of the gut microbiota for development of the
host’s immune system and the suggested connection between
gut-derived serotonin and bone led us to investigate the impact
of gut microbiota on bone mass. For this purpose, we used the
GF mouse as a model.
Materials and Methods
Mice
Female GF C57Bl6/J mice were maintained in flexible plastic film
isolators under a strict 12-hour light cycle (lights on at 6:00 a.m.).
Sterility was routinely confirmed by culturing and PCR analysis
from feces using universal bacterial primers amplifying the 16S
rRNA gene (8F, AGAGTTTGATCCTGGCTCAG; 338R, TGCTGCCT-
CCCGTAGGAGT). Age-matched female CONV-R C57Bl6/J mice
were transferred to identical isolators at weaning. Both groups of
mice were fed an autoclaved chow diet (Labdiet, St. Louis, MO,
USA) ad libitum. To produce colonized control mice, we
colonized 3-week-old GF mice with gut microbiota from
C57Bl6/J donor mice as previously described.(4)Blood was
collected from the axillary vein under anesthesia with Ketalar/
Domitor vet, and the mice were subsequently killed by cervical
dislocation. Tissues for RNA preparation were immediately
removed and snap-frozen in liquid nitrogen for later analysis.
Boneswereexcisedandfixedin4%paraformaldehydeorflushed
with PBS for isolation of bone marrow cells. The study protocols
were approved by the University of Gothenburg Animal Studies
Committee.
Peripheral quantitative computed tomography (pQCT)
Computed tomographic scans were performed with the pQCT
XCT RESEARCH M (version 4.5B, Norland, Fort Atkinson, WI, USA)
operating at a resolution of 70mm, as described previously.(33)
Trabecular volumetric bone mineral density (vBMD) was
determined ex vivo with a metaphyseal pQCT scan of the
proximal tibia. The scan was positioned in the metaphysis at a
distance distal from the proximal growth plate corresponding to
3%ofthetotallengthofthetibia,andthetrabecularboneregion
was defined as the inner 45% of the total cross-sectional area.
Cortical bone parameters were analyzed in the mid-diaphyseal
region of the femur.(34)
MicroCT (mCT)
mCT analyses were performed on the proximal tibia by using
Skyscan1072scanner(SkyscanN.V.,Aartselaar,Belgium),imaged
with an X-ray tube voltage of 100kV and current of 98mA, with a
1-mm aluminium filter.(35)The scanning angular rotation was
1808 and the angular increment 0.908. The voxel size was
6.51mm isotropically. Data sets were reconstructed by using a
modified Feldkamp algorithm and segmented into binary
images using adaptive local thresholding.(36,37)Trabecular bone
distal of the proximal growth plate was selected for analyses
within a conforming volume of interest (cortical bone excluded)
commencing at a distance of 338.5mm from the growth plate,
and extending a further longitudinal distance of 488mm in the
proximal direction. Trabecular thickness and separation were
calculated by the sphere-fitting local thickness method.(38)
Cortical measurements were performed in the diaphyseal region
of femur starting at a distance of 5221mm from the growth plate
and extending a further longitudinal distance of 163mm in the
proximal direction.
Histomorphometry
Bone histomorphometry was applied to analyze trabecular
bone in distal femur. Boneswere fixed in 4% paraformaldehyde,
dehydrated in 70% EtOH, and embedded in plastic. The
trabecular bone was analyzed using longitudinal plastic
sections obtained from three standardized sites of marrow
cavity with the site distance of 100mm. These three sites were
named as site A, site B, and site C. In each site, the trabecular
bone was analyzed in one field. Masson-Goldner Trichrome–
stained section with the thickness of 4mm was used to measure
static parameters and unstained section with the thickness of
8mm to measure dynamic parameters.(39,40)For the measure-
ment of dynamic parameters, the femora have been labeled
with calcein twice, by intraperitoneal injection at eight days,
1358
SJO¨GREN ET AL.
Journal of Bone and Mineral Research

Page 3

and at one day before sacrifice. The parameters were measured
using OsteoMeasure histomorphometry analysis system with
software version 2.2 (Osteometrics, Atlanta, GA, USA) and
following the guidelines of the American Society for Bone and
Mineral Research.(41)
Blood analysis
Analyses were performed according to the manufacturer’s
instructions for serum testosterone (RIA, MP Biomedicals/ICN
Biomedicals, Costa Mesa, CA, USA), serum calcium (QuantiChrom
Calcium Assay Kit [DICA-500], Hayward, CA, USA), serum 25-
Hydroxy Vitamin D (EIA, Immunodiagnostic Systems, Herlev,
Denmark),PlasmaPTH (MouseIntactPTHELISA,Immutopics, San
Clemente, CA, USA), and serum serotonin (ELISA, IBL-America,
Minneapolis, MN, USA). As a marker of bone resorption, serum
levelsoftypeI collagenfragmentswereassessedusing aRatLaps
ELISA kit (Nordic Bioscience Diagnostics, Herlev, Denmark).
Serum levels of osteocalcin, a marker of bone formation, were
determined with a mouse osteocalcin immunoradiometric assay
kit (Immutopics).
Quantitative real-time PCR (qRT-PCR) colon
The proximal 2cm of the colon was used for RNA extraction
using the RNeasy Mini Kit (Qiagen, Sollentuna Sweden) with on-
column DNase I (Qiagen) digestion according to the manufac-
turer’s instructions. Random hexamer-primed cDNA templates
were synthesized from 500ng of RNAs using the High Capacity
cDNA Reverse Transcription Kit (Applied Biosystems, Foster City,
CA, USA) according to the manufacturer’s instructions. cDNAs
were diluted to a final volume of 140mL, and 2mL were used for
qRT-PCR assays in 25-mL reactions containing 1X SYBR Green
Master Mix buffer (Thermo Scientific, Gothenburg, Sweden) and
900 nM gene-specific primers (450 nM primer concentrations
were used to assess L32 housekeeping gene expression). A
melting curve was performed for each primer pair to identify
a temperature where only amplicon, and not primer dimers,
accounted for SYBR Green-bound fluorescence. Assays were
performed in analytical triplicates using a 7900HT Fast Real-Time
PCR System (Applied Biosystems) or CFX96 Real-Time System
(Bio-Rad Laboratories, Stockholm, Sweden) and normalized to
the level of RNA encoding the L32 ribosomal protein using the
DDCT method.(42)Primer sequences used in this study: Tph1-F,
AACAAAGACCATTCCTCCGAAAG; T1-R, TGTAACAGGCTCACAT-
GATTCTC; L32-F, CCTCTGGTGAAGCCCAAGATC; L32-R, TCTGGG-
TTTCCGCCAGTTT. The primer
6678411a1) were obtained from the PrimerBank website
(http://pga.mgh.harvard.edu/primerbank/). We used a prede-
signed RT-PCR assay from Applied Biosystems for the analysis
of SERT (Mm00439391_m1), IL-6 (Mm00446190_m1), and TNFa
(Mm00443258_m1). The mRNA abundance was calculated using
the ‘‘standard curve method’’ (User Bulletin 2; PE Applied
Biosystems, Carlsbad, CA, USA) and adjusted for the expression
of L32 (Mm00777741_sH) ribosomal RNA. RT-PCR analyses for
SERT, IL-6, TNFa, and L32 were performed using the ABI Prism
7000 Sequence Detection System (PE Applied Biosystems).
sequencesforTph1 (ID:
qRT-PCR femur and bone marrow
Total RNA was prepared from femur and bone marrow using
TriZol Reagent (Invitrogen, Lidingo ¨, Sweden). The RNA was
reverse transcribed into cDNA using High-Capacity cDNA
Reverse Transcription Kit (#4368814, Applied Biosystems, Stock-
holm, Sweden). RT-PCR analyses were performed using the ABI
Prism 7000 Sequence Detection System (PE Applied Biosystems).
WeusedpredesignedRT-PCRassaysfromAppliedBiosystemsfor
the analysis of IL-6 (Mm00446190_m1), IL-1 (Mm00434228_m1),
and TNFa (Mm00443258_m1) mRNA levels. The mRNA abun-
dance of each gene was calculated using the ‘‘standard curve
method’’ (User Bulletin 2; PE Applied Biosystems) and adjusted
for the expression of 18S (4308329) ribosomal RNA.
Flow cytometry
Bone marrow cells were harvested by flushing 5mL PBS through
thebonecavityofonefemurusingasyringe.Aftercentrifugation
at 515g for 5 minutes, pelleted cells were resuspended in Tris-
buffered 0.83% NH4Cl solution (pH 7.29) for 5 minutes to lyse
erythrocytes and then washed in PBS. Bone marrow cells
were resuspended in RPMI culture medium (PAA Laboratories,
Pasching, Austria) before use. The total number of leucocytes in
bone marrow was calculated using an automated cell counter
(Sysmex, Hamburg, Germany). For flow cytometry analyses, cells
were stained with allophycocyanin (APC)-conjugated antibodies
to CD4 for detection of T helper cells (Becton Dickinson,
Stockholm, Sweden) and fluorescein isothiocyanate (FITC)-
conjugated antibodies to CD8 cytotoxic T cells (Becton
Dickinson) or Peridinin-chlorophyll proteins (PerCP)-conjugated
antibodies to Gr-1/Ly-6G (BioLegend, Cambridge, UK) to elimi-
nate granulocytes and FITC-conjugated antibodies to CD11b for
detection of OCL precursor cells (Becton Dickinson). The cells
werethen subjected to fluorescence activated cell sorter analysis
(FACS) on a FACSCalibur (BD Pharmingen, Franklin Lakes, NJ,
USA) and analyzed using FlowJo software. Results are expressed
as cell frequency (%).
In vitro culture of osteoclasts from bone marrow cells
Bone marrow cells were flushed from femur and tibias from
8-week-old GF and CONV-R female mice and washed once in
a-MEM culture medium (Gibco, Stockholm, Sweden). Bone
marrowcellswereseededat1?106cells/cm2in96-wellplatesin
a-MEM medium supplemented with 10% fetal bovine serum
(Sigma, St. Louis, MO, USA), 2mM glutamax (Gibco), 50mg/mL
gentamicin (Gibco), 100U/mL penicillin, 100mg/mL streptomy-
cin (PEST, Gibco), 30ng/mL recombinant murine M-CSF (rmM-
CSF, cat no. 416-ML; R&D Systems, Minneapolis, MN, USA), and
2ng/mL recombinant mouse RANKL (cat no. 462-TEC, R&D
Systems).After4days, withamediumchangeatday3,cellswere
stained for tartrate-resistant acid phosphatase (TRAP, Sigma cat
no. 387A). TRAP-positive cells containing three or more nuclei
were counted as TRAPþOCL. Because the osteoclasts generated
from bone marrow cultures from CONV-R mice were consider-
ably larger and contained more nuclei than osteoclasts from
GF mice, we also counted osteoclasts that were spread and
contained more than five nuclei.
Journal of Bone and Mineral Research
GUT MICROBIOTA REGULATES BONE MASS
1359

Page 4

Statistical analyses
All the statistical results are presented as the means?SEM.
Between-group differences were calculated using unpaired
t tests. Comparisons between multiple groups were calculated
using a one-way analysis of variance (ANOVA) followed by
Tukey’s post hoc test. A two-tailed p ? 0.05 was considered
significant.
Results
Absence of gut microbiota leads to increased bone mass
in mice
To investigate whether the gut microbiota regulates bone mass,
we determined the proximal tibial trabecular vBMD in 7-week-
old female GF and CONV-R mice. We found that GF mice had
increased trabecular vBMD compared with CONV-R mice
(277?14 versus 208?10mg/cm3, p ? 0.01) as determined
by pQCT. Further analysis with mCT showed that trabecular bone
volume/tissue volume (BV/TV) was increased by 39.4% in the
distal femur of GF compared with CONV-R mice (p ? 0.01,
Fig. 1A, B). The increased BV/TV was associated with increased
trabecular number (þ39.5%, p ? 0.01) and decreased trabecular
separation (þ26.2%, p ? 0.01), whereas trabecular thickness was
unchanged in GF compared with CONV-R mice (Fig. 1C–E).
Histomorphometric analysis of the distal femur from 9-week-old
mice showed similar results, with increased BV/TV (þ39.7%, p ?
0.01) in the distal metaphyseal region of femur, increased
trabecular number (þ36.5%, p ? 0.01), and reduced trabecular
separation (þ29.2%, p ? 0.01), whereas trabecular thickness was
unchanged in GF compared with CONV-R mice (Supplemental
Fig. S1). Histomorphometric analysis showed a decrease in the
number of OCLs per bone perimeter, ie, the number of OCLs
relative to the external perimeter of the trabecular bone in GF
compared with CONV-R mice (–11%, p ? 0.05, Table 1). Dynamic
histomorphometry was used to analyze bone formation, and
mineralizing surface per trabecular bone surface was increased
in GF compared with CONV-R mice (þ17%, p ? 0.05, Table 1).
However, there was no change in mineral apposition rate
(Table 1). Analysis of cortical bone in the mid-diaphyseal region
of femur by mCT showed that cortical bone area was increased in
GF compared with CONV-R mice (CONV-R 0.56?0.01mm2; GF
0.61?0.01 mm2, p ? 0.01). We measured serum osteocalcin as a
marker of bone formation, and although there was a tendency of
increased levels in GF compared with CONV-R mice, it did not
reach statistical significance (CONV-R 304.2?29.5ng/mL; GF
350.4?29.3ng/mL, p¼0.32). As a marker of bone resorption,
we measured serum RatLaps (C-terminal telopeptides), and
although there was a tendency of decreased levels in GF
compared with CONV-R mice, it did not reach statistical
significance (CONV-R 38.2?4.8ng/mL; GF 30.5?3.0ng/mL,
p¼0.27).
GF mice have decreased levels of serum serotonin
Gut-derived serotonin has been reported to have negative
effects on bone mass, and GF mice had decreased serum
serotonin levels compared with CONV-R mice (?67.8%, p ? 0.01,
Fig. 2A). We analyzed the mRNA expression of Tph1 and SERT in
Fig. 1. Absence of gut microbiota leads to increased bone mass in mice.
Trabecular bone parameters were analyzed by mCT in the distal meta-
physeal region of femur from 7-week-old germ-free (GF) and conven-
tionally raised (CONV-R) female mice. (A) Representative mCT images of
one trabecular section from each group. (B) BV/TV (%), trabecular bone
volume as a percentage of tissue volume. (C) Tb.N (mm?1), trabecular
number. (D) Tb.Sp (mm), trabecular separation. (E) Tb.Th (mm), trabecular
thickness. Values are given as mean?SEM, n¼8 to 14.?p ? 0.05;??p ?
0.01 versus CONV-R, Student’s t test.
Table 1. Histomorphometric Analysis of Trabecular Bone in the
Distal Metaphyseal Region of Femur From 9-Week-Old Germ-
Free (GF) and Conventionally Raised (CONV-R) Female Mice
CONV-R GF
N.OCL/B.Pm (mm?1)
MS/BS (%)
MAR (mm/day)
BFR/BS (mm3/mm2/year)
2.92?0.09
27.9?1.3
1.95?0.15
197?15
2.60?0.11?
32.6?0.3?
1.59?0.02
190?1
N.OCL/B.Pm¼number of OCLs per trabecular bone perimeter; MS/
BS¼mineralizing surface per trabecular bone surface; MAR¼mineral
apposition rate; BFR/BS¼bone formation rate per trabecular bone
surface. Values are given as mean?SEM; n¼5 to 6 for static parameters
(N.OCL/B.Pm) and n¼3 to 6 for dynamic parameters in each group.
Student’s t test,?p ? 0.05 versus CONV-R.
1360
SJO¨GREN ET AL.
Journal of Bone and Mineral Research

Page 5

Fig. 2. GFmicehavedecreasedperipheralserotonin synthesis. (A) Serumserotoninlevelsin 7-week-oldGFandCONV-Rfemalemice.(B) qRT-PCRanalysis
of the expression of tryptophan hydroxylase-1 (Tph1) in proximal colon of 9-week-old GF and CONV-R female mice. (C) qRT-PCR analysis of the expression
of the serotonin transporter (SERT) in proximal colon of 9-week-old GF and CONV-R female mice. Values are given as mean?SEM, n¼5 to 14.??p ? 0.01
versus CONV-R, Student’s t test.
Fig. 3. GF mice have a decreased frequency of CD4 T cells and osteoclast (OCL) precursor cells (CD11bþ/Gr1–) in bone marrow. Femur bone marrow cells
from7-week-oldGFandCONV-RfemalemicewerestainedwithantibodiesrecognizingCD8,CD4,CD11b,andGr1andanalyzedbyflowcytometry,gating
as shown. (A) Values represent the percentage of CD8þand CD4þcells in the total bone marrow population. Data are representative of two independent
experiments. (B) Values represent the percentage of CD11bþ/Gr1–cells (OCL precursor cells; see text for details) in the total bone marrow population.
Data are representative of two independent experiments.
Journal of Bone and Mineral Research
GUT MICROBIOTA REGULATES BONE MASS
1361