Full text of DOI:10.1359/jbmr.2002.17.10.1822

JOURNAL OF BONE AND MINERAL RESEARCH
Volume 17, Number 10, 2002
© 2002 American Society for Bone and Mineral Research
Dentin Matrix Protein 1, a Target Molecule for Cbfa1 in
Bone, Is a Unique Bone Marker Gene
JIAN Q. FENG,1,2 JIANGHONG ZHANG,1,2 SARAH L. DALLAS,1 YONGBO LU,1 SHUO CHEN,3
XIAOYU TAN,1 MICHAEL OWEN,4 STEPHEN E. HARRIS,1 and MARY MACDOUGALL3
ABSTRACT
Dentin matrix protein 1 (Dmp1), a phosphoprotein highly linked to dentin formation, has also been reported to be
expressed in the skeleton. However, the role of Dmp1 in skeletal tissues remains unclear. To clarify the role of Dmp1
in bone formation, we characterized the expression profile of Dmp1 in bone and cartilage and examined whether
Dmp1 expression was regulated by core-binding factor a1 (Cbfa1). Studies of fetal rat calvarial (FRC) cell cultures
showed that the expression of Dmp1 was associated closely with “bone nodule” formation and mineralization in
vitro. In situ hybridization studies were performed to examine the spatial and temporal expression patterns of
Dmp1 during development in mouse embryos from 12.5 day postcoitus (dpc) to 8 weeks postnatal; these studies
showed that Dmp1 first appeared in hypertrophic cartilage cells, followed by osteoblasts, and later was expressed
strongly in osteocytes. The expression profiles of Cbfa1 and Dmp1 overlapped in both cartilage and bone during
development, with Cbfa1 preceding Dmp1. Examination of Dmp1 expression in Cbfa1؊/؊ mice revealed that Dmp1
was absent in the developing bones of Cbfa1-null mice, whereas there was essentially no change in Dmp1 expression
in the arrested tooth bud. Transient transfection studies showed forced expression of Dmp1 under the control of
Cbfa1 and gel shift data indicated the presence of a functional osteocalcin-specific element (OSE)-2 response
element in the Dmp1 proximal promoter region. However, in vitro promoter studies suggested that regulation of
Dmp1 by Cbfa1 was not mediated by direct binding of Cbfa1 to this site and may be through indirect mechanisms.
These studies highlight Dmp1 as a unique marker gene for osteoblastic differentiation. The close association of
Dmp1 and Cbfa1 in the developing skeleton suggests that Dmp1 may play an important role in bone formation.
(J Bone Miner Res 2002;17:1822–1831)
Key words: Dmp1, development, bone, Cbfa1, cartilage
INTRODUCTION
teins localized within collagen gap zones that bind calcium and
phosphate ions in an appropriate conformation to nucleate
formation of apatite crystals.(1,2) These highly phosphorylated
proteins such as bone sialoprotein (BSP), as well as osteocalcin
(OC), and alkaline phosphatase (AP) are widely referred to as
bone marker genes.(3–7)
ONE AND dentin are mineralized tissues that have many
characteristics in common, especially those related to ini-
tiation of the mineralization event. It is believed that mineral-
ization is initiated by phosphorylated extracellular matrix pro-
B
Dentin matrix protein 1 (Dmp1) is an acidic phosphory-
lated extracellular matrix protein originally identified from a
The authors have no conflict of interest.
1Department of Oral Biology, School of Dentistry, University of Missouri-Kansas City, Kansas City, Missouri, USA.
2Jian Q. Feng and Jianghong Zhang contributed equally to this work.
3Department of Pediatric Dentistry, University of Texas Health Science Center at San Antonio, San Antonio, Texas, USA.
4Imperial Cancer Research Fund, London, United Kingdom.
1822

STUDY OF Dmp1 EXPRESSION AND REGULATION
1823
rat incisor cDNA library.(8) DNA sequence analysis has cDNA (0.6 kb) was cleaved by XhoI, a partial Cbfa1 cDNA
revealed that the protein is rich in aspartic acid, glutamic (270 bp; a gift from Dr. G. Karsenty, Baylor College of
acid, and serine. This protein has numerous potential phos- Medicine, Houston, TX, USA) cut by EcoRI, and a partial
phorylation sites for casein kinases I and II. A potential
Arg-Gly-Asp (RGD) cell attachment sequence and a con-
sensus sequence for N-glycosylation have also been identi-
fied. The Dmp1 gene has been mapped to mouse chromo-
some 5q21 and human chromosome 4q21:22.(9–11) These
regions contain a superfamily gene locus of related acidic
dentin and bone phosphoproteins that are of potential im-
portance during the process of biomineralization in dentin
and bone. This region of the human chromosome contains
the critical loci for dentinogenesis imperfecta types II and
III and dentin dysplasia type II,(12) all diseases of teeth that
display mineralization defects. Currently, the precise func-
tion of Dmp1 is unknown, although this protein can bind
tightly to hydroxyapatite and can mediate cell attachment
through the RGD domain.(13) Initial in situ hybridization
and Northern data suggested that this gene was odontoblast
specific.(14) More recent studies have shown that Dmp1 is
also expressed in bone and brain,(15,16) although data from
Toyosawa and colleagues(17) suggest that Dmp1 is a specific
marker gene for osteocytes and preosteocytes.
OC cDNA was liberated by HindIII. In situ hybridization on
frozen sections was carried out essentially as described by
Shamim at al.(22) The hybridization temperature was set at
55°C, and washing temperature was at 70°C, which inacti-
vates endogenous AP. DIG-labeled nucleic acids were de-
tected in an enzyme-linked immunoassay with a specific
anti–DIG-AP antibody conjugate and the color substrates
4-nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-
indolyl phosphate (BCIP) according to the manufacturer’s
instructions (Roche). The hybridization signals were photo-
graphed with a Nikon E800 microscope (Nikon, Tokyo,
Japan) and a MagnaFire camera (Optronics, Goleta, CA,
USA).
Preparation of nuclear extracts and electrophoretic
mobility shift assay
Nuclear extracts from fetal rat calvarial (FRC) cells with
osteoblast-like characteristics and an odontoblast cell
Core-binding factor a1 (Cbfa1) is a member of the runt line(23) were prepared by
domain family of transcription factors. Direct evidence that
a method previously de-
scribed.(24,25) The extraction buffer contained 0.5 mM of
dithiothreitol, 0.5 mM of phenylmethylsulfonyl fluoride,
and 2.5 ␮g/ml of leupeptin and pepstatin. The double-
stranded oligonucleotides GTCATCTGAGAACCACAA-
GAGCTTT (wild type) and GTCATCTGAGAAGAACAA-
GAGCTTT (mutant) were labeled with [␥-32P]adenosine
triphosphate (ATP) and purified on a polyacrylamide gel.
For competition binding reactions, the unlabeled competitor
DNA in 100-fold molar excess of the labeled probe was
included. Samples were separated by electrophoresis on a
5% polyacrylamide gel in 1ϫ Tris-borate-EDTA (TBE) at
160 V for 90 minutes. The gels were dried and autoradio-
graphed.
Cbfa1 is involved in bone and tooth development comes
from Cbfa1 gene knockout experiments.(18,19) There is a
complete absence of bone, as well as arrested tooth devel-
opment, in Cbfa1-deficient animals. In Cbfa1-null mice, the
membranous bones of the skull are replaced by fibrous
tissue and endochondral bone fails to replace the cartilagi-
nous skeleton. Heterozygous Cbfa1 mice express a pheno-
type that is similar to the clinical manifestations of cleido-
cranial dysplasia (CCD), in which functional mutations of
the Cbfa1 gene have been identified.(20,21)
In this study, we sought to determine the role of Dmp1 in
bone by first defining the spatial and temporal expression of
the gene in the developing skeleton and comparing its
expression with that of the osteoblast-specific transcription
factor Cbfa1 as well as the mature osteoblast marker OC.
Because of the observed overlap in expression of these two
genes and our identification of a Cbfa1 response element
osteocalcin-specific element (OSE)-2 in the Dmp1 pro-
moter, we next examined whether Dmp1 expression was
regulated by Cbfa1. This was done by in vitro approaches
using gel shift assays, mutation of OSE-2, and transient
transfection of Cbfa1, as well as in vivo studies using
Cbfa1-null mice. These studies show that Dmp1 is a marker
gene for chondrocytes and osteoblasts during early skeleto-
genesis and for osteocytes in the mature skeleton and sug-
gest that it may be an important target molecule for Cbfa1
during cartilage and bone development.
FRC cell culture
FRC cells, enriched for cells with an osteoblast pheno-
type, were isolated from timed pregnant Sprague–Dawley
rats using procedures described by Bellows et al.(26) and
Harris et al.(27) Briefly, first-passage cells were plated at a
density of 1.3 ϫ 104 cells/well in 24-well tissue culture
dishes (Costar, Corning, NY, USA). Cells were grown to
confluence in ␣-modified minimum essential medium (␣-
MEM; Gibco BRL, Carlsbad, CA, USA) supplemented with
7% FCS. The medium was then changed to ␣-MEM sup-
plemented with 7% FCS, 100 ␮g/ml of ascorbic acid
(Sigma, St. Louis, MO, USA), and
5 mM of
␤-glycerophosphate (Sigma). This was defined as day 0.
Subsequently, the medium was changed every 2 days until
cell cultures were stopped at days 2, 4, 7, 10, and 14
respectively. For each time point, one set of cells was used
for RNA extraction, while a parallel set was stained using
MATERIALS AND METHODS
In situ hybridization
Digoxigenin (DIG)-labeled cRNA probes were prepared the Von Kossa procedure.(26,27) A similar culture procedure
by using the RNA Labeling Kit (Roche, Indianapolis, IN, was used for studies of Cbfa1 and Dmp1 expression in 2T3
USA). To make antisense probes, a partial mouse Dmp1 (an osteoblast) cells.

1824
FENG ET AL.
Promoter activities were measured after 48-h transfection
in C3H10T1/2 cell lines. All data were normalized by the
dual-luciferase assay by cotransfection of thymidine kinase
(TK)-renilla luciferase vector (Promega). In this dual-
luciferase system, both a wild type and the mutated Dmp1
promoter fragments were linked to the firefly luciferase
gene while the cotransfected renilla luciferase gene was
driven by the TK promoter. The second reporter gene pro-
vided an internal control by which each value within an
experimental set could be normalized. Because both en-
zymes used different substrates, dual-luciferase measure-
ments were performed using TD-20/20 (Promega) in a
single tube, with firefly luciferase assayed first followed by
renilla luciferase. The promoter activity was reflected by the
ratio of firefly/renilla luciferase for each construct.
Northern and reverse transcription-polymerase chain
reaction analyses
Total RNA was extracted from FRC or 2T3 cells using
the Ultraspec isolation system (Biotecx, Houston, TX,
USA) based on the manufacturer’s protocol. One micro-
gram of total RNA from each time point was converted to
cDNA using reverse transcriptase (RT; Promega, Madison,
WI, USA) with 0.1 ␮g of oligo (dT) as a primer. Polymerase
chain reaction (PCR) amplification then was performed
using a 2-␮l aliquot of the target cDNA. The sites of Dmp1
primers, corresponding to the Dmp1 gene, are displayed in
Fig. 4A. A 5Ј primer (primer 1, 5Ј-CGTTCCTCTG
GGGGCTGTCC-3Ј) and a 3Ј primer (primer 2, 5Ј-
CGGGATCA TCGCTCTGCATC-3Ј) were used for Dmp1
RT-PCR. A 5Ј primer (5Ј-CCATGGAGAACGCTGGGG-
3Ј) and a 3Ј primer (5Ј-caaagttgtgatggatgacc-3Ј) for
GAPDH were used as internal controls. PCR reactions were
amplified for 30 cycles for each set of primers using an
annealing temperature between 56°C and 58°C, depending
on the specific primer set. Quantitative PCR was accom-
plished by adding 1 ␮l of [␣-32P]deoxycytidine triphosphate
(dCTP) into each Dmp1 reaction followed by electrophore-
sis, DNA transfer to nitrocellulose membrane, and quanti-
tation of signals using the AMBIS image acquisition and
analysis system (AMBIS, San Diego, CA, USA). Amplified
products were normalized to GAPDH.
RESULTS
Overlap in expression of Cbfa1 and Dmp1 in
hypertrophic cartilage and bone formation during
embryonic development
To study the expression patterns of Cbfa1 and Dmp1,
mouse embryos were harvested at embryonic stage E12.5
(cartilage formation), E14.5 (initiation of ossification), and
E16.5 (rapid bone formation). As shown in Fig. 1A, Dmp1
expression was virtually undetectable at E12.5, whereas
Cbfa1 was highly expressed in the developing cartilage
tissues. At E14.5, Cbfa1 continued to be expressed at a high
level in cartilaginous tissues. At this stage, Dmp1 expres-
sion could now be observed in the developing skeleton but
was confined to areas containing hypertrophic cartilage
(Fig. 1B and 1B inset). At E16.5, the expression of Dmp1
overlapped more closely with that of Cbfa1, with expression
seen in the hypertrophic cartilage and in osteoblasts (Figs.
1C and 1D). However, in contrast to Cbfa1, no Dmp1
expression was seen in areas of cartilage that were not
undergoing mineralization (Figs. 1C and 1D). Thus, the
expression of Cbfa1 and Dmp1 appeared to be linked
closely, with Cbfa1 preceding Dmp1 during early bone
formation.
To generate a Northern blot, 5 ␮g of poly(Aϩ) was
fractionated on a 1% formaldehyde agarose gel, transferred
to a Nytran filter (Schleicher & Schuell, Keene, NH, USA),
and UV cross-linked. The radioactive probes were prepared
by random priming from partial cDNAs of a 0.3-kb mouse
Cbfa1 and 0.6-kb mouse Dmp1. Prehybridization and hy-
bridization were carried out at 42°C in 5ϫ SSC containing
50% formamide and 150 mg/ml of denatured salmon sperm
DNA. Quantitation of the Northern blots was performed
using the method described previously.
Transient transfection
A Cbfa1 expression vector in which a Cbfa1 cDNA was
driven by the cytomegalovirus (CMV) promoter (a gift from
Dr. G. Karsenty, Baylor College of Medicine, Houston, TX,
USA) was transiently transfected into a nonosseous cell line
C3H10T1/2. LipofectAMINE PLUS-mediated DNA deliv-
ery was used for transfection in accordance with the proto-
col supplied by the manufacturer (Gibco BRL). Total RNA
was isolated after 48-h transfection for measurement of
Dmp1 expression by RT-PCR using an annealing tempera-
ture of 58°C and 30 cycles.
Expression profile of Dmp1 during bone nodule
formation and mineralization in vitro
The in vitro “bone” nodule formation during culture of
FRC cells is a well-established model system for the study
of expression profiles for bone marker genes.(26,27) To fur-
ther investigate whether Dmp1 is linked to bone nodule
formation and mineralization, Dmp1 expression was exam-
ined in long-term FRC cell cultures (Fig. 2). We have
previously reported the presence of two alternatively spliced
transcripts for the Dmp1 gene.(28) Because mouse Dmp1
Mutagenesis of Dmp1 promoter and
dual-luciferase assay
Mutated Cbfa1 binding sites were generated using the cDNA is ϳ3 kb and the size difference between transcripts
QuickChange Site-Directed Mutagenesis Kit (Stratagene, La A and B is only 45 bp,(28) we used RT-PCR to evaluate
Jolla, CA, USA) with primers 5Ј-CGTTTAGTCATCTGG expression of the two transcripts. The graph in Fig. 2B
GATCCTCAAGAG CTTTTTAGAG-3Ј and 3Ј-GCAAATC shows that although both transcripts were expressed at all
AGTAGACCCTAGGAGTTCTCGAAAAATCTC-5Ј based time points examined, transcript A was increased dramati-
on the manufacturer’s manual.
cally during nodule formation (day 10) and mineralization

STUDY OF Dmp1 EXPRESSION AND REGULATION
1825
FIG. 1. Coordinated expression of
Cbfa1 and Dmp1 in hypertrophic
chondrocytes and osteoblasts during
embryonic development. In situ hy-
bridization was performed on frozen
sections from (A) 12.5 dpc, (B) 14.5
dpc, and (C) 16.5 dpc embryos using
probes for Cbfa1 (upper panels of
A–D) and Dmp1 (lower panels of
A–D). (D) Enlarged views of 16.5
dpc embryos. D1–3 shows Cbfa1 ex-
pression and d1–3 shows Dmp1 ex-
pression in the calvaria (D1 and d1),
vertebral osteoblasts (D2 and d2),
and chondrocytes/hypertrophic chon-
drocytes from a rib (D3 and d3). The
red arrows indicate locations of the
cross-sections taken from the rib.
Hybridization signals are shown by
purple AP staining. Orange G was
used as the counterstain.
(day 14; Fig. 2B). In contrast, transcript B was expressed at
low levels and remained unchanged throughout the culture
period. Furthermore, immunocytochemistry (Fig. 2C) indi-
cated that Dmp1 was localized to nodule-forming areas,
suggesting that the secreted Dmp1 may be involved in
nodule formation. We also examined expression of Dmp1
and Cbfa1 during in vitro mineralization by Northern anal-
ysis using the osteoblast cell line 2T3. This cell line was
derived from the calvarial cells of a transgenic mouse in
which the T-antigen was expressed under control of the
bone morphogenetic protein (BMP) 2 promoter, thus direct-
ing expression of the immortalizing T-antigen to the osteo-
blast lineage.(29) This cell line has been extensively charac-
terized and shown to express osteoblast-specific markers
and to form mineralized bonelike nodules in vitro. Consis-
tent with our in vivo findings, Cbfa1 and Dmp1 expression
appeared to be linked closely in long-term mineralizing
cultures 2T3 osteoblasts (Fig. 3A).
An OSE-2 response element exists in the Dmp1
promoter, but this response element is not critical for
Dmp1 promoter activity in vitro
To understand the mechanism by which Cbfa1 regulates
Dmp1 expression, we identified a putative Cbfa1 response
element or OSE-2, (AACCACA) in the Dmp1 promoter at
Ϫ165 bp. Cbfa1 binding to this response element was
examined by electrophoretic mobility shift assay (EMSA),
using classical OSE-2 as a positive control. The response
element identified in Dmp1 bound to a specific protein
complex in nuclear extracts from odontoblasts, which mi-
grated with a similar electrophoretic mobility as the com-
plex formed with classical OSE-2 (Fig. 4A, lanes 1 and 4,
black arrow). An excess of the unlabeled Dmp1 element
abolished binding of the complex to both the labeled Dmp1
response element and the classical OSE-2 element (Fig. 4A,
lanes 2 and 6). Similarly, an excess of the unlabeled OSE-2
element abolished binding of the complex to both the la-
beled Dmp1 response element and the classic OSE-2 (Fig.
4A, lanes 3 and 5). Binding of the Dmp1 response element
to the same protein complex also occurred in FRC osteo-
blasts (Fig. 4A, lane 7, black arrow). An excess of unlabeled
response element or OSE-2 completely abolished binding
(Fig. 4A, lanes 8, 9, 12, and 13) while a mutated element in
which the AACCACA site was disrupted did not compete
for binding (Fig. 4A, lanes 10 and 11). In nuclear extracts
from odontoblasts, an additional complex was observed that
bound to the Dmp1 response element (Fig. 4A, lanes 1 and
3, white arrow). This binding was competed by the unla-
beled Dmp1 response element (Fig. 4A, lane 2) but not by
classical OSE-2 (Fig. 4A, lane 3). At present, the nature of
this complex is unknown.
Forced expression of Cbfa1 induces Dmp1 expression
in a nonosteoblastic cell line
The overlap in expression of Cbfa1 and Dmp1 suggested
that Dmp1 may be regulated by the transcription factor
Cbfa1. To explore this possibility, Cbfa1 expression plas-
mids were transfected into
a nonosseous cell line
C3H10T1/2 that has undetectable endogenous Dmp1 ex-
pression to determine whether Dmp1 expression could be
induced (Fig. 3B). In cells transfected with empty vector,
Dmp1 expression was undetectable after 30 cycles of RT-
PCR. In contrast, transfection with the Cbfa1 expression
vector resulted in a dose-related induction of Dmp1 expres-
sion in C3H10T1/2 cells, suggesting that Cbfa1 functionally
regulates Dmp1 expression in vitro.
Although we had identified a Cbfa1 response element in
Dmp1 that bound to nuclear extracts from odontoblasts and
osteoblasts, we next wanted to determine whether this response

1826
FENG ET AL.
FIG. 2. Dmp1 expression is associated with bone nodule formation in vitro. (A) Diagrammatic representation of mouse genomic Dmp1. The
positions of the primers used for PCR determination of Dmp1 transcripts A and B are indicated. The 5Ј primer (primer 1) is located on exon 2
and a 3Ј primer (primer 2) located on exon 6. (B) Expression profiles of Dmp1 transcripts in FRC cell culture. Cultured FRC cells were harvested
on days 0, 2, 4, 7, 10, and 14 and expression of Dmp1 transcripts A and B were analyzed by RT-PCR. The expression level of Dmp1 was
normalized to that of GAPDH. The experiments were carried out four times, and a representative blot is shown. The closed bars indicate transcript
A, and the open bars indicate transcript B. The RT-PCR products were confirmed by DNA sequencing. (C) Dmp1 is localized at sites of bone
nodule formation. Cultured FRC cells (day 7) were fixed and stained by immunofluorescence using a rabbit polyclonal anti–Dmp1 antibody. The
left panel shows a phase contrast image and the right panel shows the same field stained with mouse Dmp1 polyclonal antibody. The lines indicate
the location of Dmp1 in the nodules. (D) Photograph of FRC bone nodule formation in vitro. Cultured FRC cells, initially plated at 13,000
cells/cm2, were fixed and stained with Van Giesen stain, which gives a pink color for collagen and brown-yellow color for unmineralized nodules,
and Von Kossa stain, which stains mineralized structures black. Cultured FRC cells were fixed and stained on days 0, 2, 4, 7, 10, and 14.
Representative images from days 7, 10, and 14 are shown. Note that formation of nonmineralized nodules begins at day 7 and mineralization
begins at day 10 and peaks at day 14.
element was critical for the regulation of Dmp1 by Cbfa1 in (Figs. 5C–5E) was highly expressed both in the developing
C3H10T1/2 cells. Cbfa1 was cotransfected into C3H10T1/2 bones and in the developing teeth. In contrast, Dmp1 ex-
cells either with a reporter construct consisting of a 0.9-kb pression was absent in the undifferentiated bone tissues of
mouse Dmp1 promoter fused to luciferase, a reporter construct Cbfa1-null embryos. Interestingly, the expression level of
in which the OSE-2 putative Cbfa1 binding sites were mutated, Dmp1 in the arrested tooth bud was similar to that in the
or a positive control reporter construct consisting of the OC normal tooth (Figs. 5D and 5E), suggesting that the control
promoter with six repeats of OSE2-pGL3 (Fig. 4B). Surpris- mechanisms regulating expression of Dmp1 may be differ-
ingly, mutation of the OSE-2 site in the Dmp1 promoter ent in bone compared with teeth. OC, a marker of mature
construct had no effect on its basal activity. Furthermore, osteoblasts, was absent in both the undifferentiated bone
forced expression of Cbfa1 did not increase Dmp1 promoter and the arrested tooth tissues of Cbfa1 mice (Figs. 5A and
activity in C3H10T1/2 cells. However, as expected, Cbfa1 5B), indicating that regulation of OC by Cbfa1 is likely
dramatically enhanced OC promoter activity (Fig. 4B). These under the same control mechanism in both bones and teeth.
data suggest that Cbfa1 has no direct effect on OSE-2 in the
Dmp1 promoter in vitro.
Comparison of Dmp1 expression during embryonic
versus postnatal bone development
Dmp1 expression is absent in the undifferentiated bone
tissues of Cbfa1-null mice
Bone resorption and formation occur continually through-
out life to remodel the bone for maintenance of calcium
Cbfa1-null mutant mice lack mature bones(18) and display homeostasis and for adaptation to mechanical loading, as
arrested tooth development,(19) suggesting that Cbfa1 may well as to repair microdamage to the skeleton. This process
function as a critical regulatory factor during the develop- of bone remodeling can be viewed as distinct from the bone
ment of bones and teeth. To further define the role of Cbfa1 formation that occurs during embryonic development of the
in the regulation of Dmp1 expression, we examined Dmp1 skeleton. To determine whether Dmp1 may play a role in
expression in mice that lacked the gene for Cbfa1. In bone remodeling during postnatal development, we exam-
16.5-day and 19.5-day embryos from wild-type mice, Dmp1 ined expression of Dmp1 during embryonic development as

STUDY OF Dmp1 EXPRESSION AND REGULATION
1827
FIG. 3. Expressions of Cbfa1 and Dmp1 overlap in prolonged cultures of osteoblast cells and forced expression of Cbfa1 induces Dmp1
expression in nonosseous cells in vitro. (A) Overlap of expression of Cbfa1 and Dmp1 in osteoblasts in vitro. A Northern hybridization showing
Cbfa1 and Dmp1 expression in 17-day cultures of 2T3 osteoblasts is shown in the upper panel, with the quantified data, normalized to GAPDH,
shown in the lower panel. Day 0 corresponds to when the cells reach confluences. (B) Transient transfection of a Cbfa1 expression plasmid induces
Dmp1 expression in C3H10T1/2 fibroblasts. C3H10T1/2 cells cultured in 12-well plates were transfected with a Cbfa1 expression plasmid (0.1
␮g/well and 0.5 ␮g/well). RNA was harvested 48 h after transfection, and Dmp1 expression was analyzed by RT-PCR using an annealing
temperature of 58°C, with 30 amplification cycles. GAPDH expression was used as an internal control.
FIG. 4. There is an OSE-2 in the mouse Dmp1 promoter, but this response element is not critical for Dmp1 promoter activity in vitro. (A) The
proximal Dmp1 promoter contains an OSE-2, and gel shift assays were performed using labeled double-stranded oligonucleotides corresponding
to the putative OSE-2 response element from Ϫ157 to Ϫ175 bp in the proximal Dmp1 promoter region (1–3 and 7–13). Osteoblast and odontoblast
nuclear extracts were used as a source of nuclear proteins. The sequences of oligonucleotides are shown in Materials and Methods. Classical
OSE-2 oligonucleotides(24) were used as controls (lanes 4–6). Unlabeled oligonucleotides used for competition were Dmp1 (lanes 2, 6, 8, and 9),
mutated Dmp1 (mDmp1, lanes 10 and 11), and classical OSE-2 (lanes 3, 5, 12, and 13). Competitor oligonucleotides were used at a 100-fold
excess concentration. The data shown are representative of three separate experiments. (B) Forced expression of Cbfa1 has no direct effect on the
OSE-2 response element in Dmp1 promoter in C3H10T1/2 fibroblasts. Cbfa1 was cotransfected into C3H10T1/2 cells either with a reporter
construct consisting of a 0.9-kb mouse Dmp1 promoter fused to luciferase (Dmp1), a similar reporter construct in which the OSE-2 putative Cbfa1
binding sites were mutated (mDmp1), a positive control reporter construct consisting of the OC promoter with six repeats of OSE2-pGL3 (OC),
or the control empty luciferase vector without a promoter (pGL3). The data shown are representative of three separate experiments.
well as in postnatal animals (Fig. 6). At embryonic stage
E16, Dmp1 was highly expressed in the newly formed
hypertrophic cartilage (primary ossification center). By
stage E18, when the mineralized bone collar is starting to
Expression of Cbfa1 and Dmp1 is disassociated during
postnatal bone development
We next compared the postnatal expression of Dmp1 with
form on the outside of the diaphysis, Dmp1 was expressed Cbfa1 to determine whether there is link between Cbfa1 and
in the hypertrophic cartilage as well as in osteoblasts in the Dmp1 expression in mature bones. The expression pattern
newly mineralizing bone. Interestingly, by 8 weeks of post- of OC was also examined for comparison. In 2-week-old
natal life, Dmp1 was actively expressed in the hypertrophic postnatal animals, Dmp1 was actively expressed in the
chondrocytes and was highly expressed in osteocytes em- hypertrophic zone of the growth plate and was also highly
bedded in the bone matrix.
expressed in osteocytes (Fig. 7B). High expression in the

1828
FENG ET AL.
FIG. 5. Lack of Cbfa1 is associated with no Dmp1 expression in developing bone tissues, and lack of Cbfa1 has little effect on Dmp1 expression
in the arrested tooth bud. In situ hybridization using Dmp1 and OC probes were carried out on sagittal sections from frozen embryos. Results from
wild-type embryos are shown in the upper panel and results from Cbfa1-null embryos are shown in the lower panels. (A) OC expression in the
developing femur of the 16.5-dpc embryo; (B) OC expression in developing tooth of the 18-dpc embryo; (C) Dmp1 expression in developing
femur of the 16.5-dpc embryo; (D) Dmp1 expression in developing tooth of the 16.5-dpc embryo; (E) Dmp1 expression in developing tooth of
the 19.5-dpc embryo. Note that Dmp1 is still expressed in the arrested teeth of Cbfa1-null mice but is not expressed in the developing bones. In
contrast, OC expression is absent in both bones and teeth in Cbfa1Ϫ/Ϫ mice.
osteocytes was also observed in 8-week-old postnatal bones postnatal bone development, at which time Cbfa1 was vir-
(Fig. 6, e2). In contrast, in the long bone, Cbfa1 expression
was virtually undetectable (Fig. 7A). OC, as a control, was
actively expressed in osteoblasts but was not detected in
chondrocytes or osteocytes (Fig. 7C).
tually undetectable. These data suggest that expression of
Dmp1 is related not only to bone formation during embry-
onic development, but may also play a role in the bone
formation associated with skeletal remodeling. Thus, Dmp1
may represent a unique bone-specific marker, because its
expression occurs after Cbfa1 expression but before expres-
sion of other well-characterized markers such as BSP and
OC, which are expressed in differentiated osteoblasts.
Long-term FRC cell primary cultures are a well-accepted
in vitro model for the study of bone formation and miner-
alization events.(26,27) The chronological expression se-
quence of different bone marker genes during proliferation
and differentiation of the mature osteoblast phenotype in
this culture system has been described previously.(6) Our
data using this culture system show that Dmp1 protein is
localized in areas undergoing nodule formation. Thus, the in
vitro Dmp1 expression profile is in agreement with our in
vivo data. In addition, our in vitro studies using the FRC
culture system confirmed the presence of two transcripts for
Dmp1.(29) Expression of the full-length transcript A was
correlated with bone nodule formation and mineralization.
In contrast, transcript B, which is missing exon 5, did not
appear to be related to bone formation because the expres-
sion level was low and was largely unchanged during nod-
ule formation and mineralization. At present, the signifi-
cance of transcript B is unknown.
DISCUSSION
Although Dmp1 was originally isolated as a dentin-
specific gene, our studies provide strong evidence that it is
highly expressed in hypertrophic cartilage and bone and that
its expression is regulated by the osteoblast-specific tran-
scription factor Cbfa1 in early bone formation. Previous
studies of Dmp1 expression have focused mainly on the
craniofacial region. In addition, there has been no detailed
analysis of Dmp1 expression during bone development. The
results from our in situ hybridization studies of Dmp1
expression at different developmental stages clearly show
that Dmp1 is expressed in both intramembranous and en-
dochondral bones. This is interesting in view of a recent
report suggesting that cells that form intramembranous bone
may differ from cells that form endochondral bone in their
osteogenic mechanisms.(30) In our developmental studies,
Dmp1 expression was first detected in hypertrophic chon-
drocytes at ϳE14.5, which is before the initiation of trabec-
ular bone formation. These data are in agreement with
RT-PCR data reported by D’Souza et al., which showed
detection of Dmp1 expression at E13.(16) Taken together,
these data suggest that Dmp1 is an early marker gene for
bone formation. The signal for Dmp1 intensified as ossifi-
cation began and remained strong in osteoblasts. However,
Dmp1 expression was observed mainly in the hypertrophic
The recent discovery of the function of Cbfa1 has had a dra-
matic impact on our understanding of osteogenesis.(18,19,24,31–37)
This gene is expressed in cells of the osteoblast lineage and
serves as a master gene regulating osteoblast-specific gene
expression. For example, deletion of the Cbfa1 gene results
chondrocyte zone of the growth plate and osteocytes during in the complete lack of a mineralized skeleton,(18) while

STUDY OF Dmp1 EXPRESSION AND REGULATION
1829
FIG. 7. Expression patterns of Cbfa1, Dmp1, and OC in cartilage and
bone during postnatal development. In situ hybridization was per-
formed on frozen sections from mouse 2-week-old femurs using probes
for (A and enlarged inserts) Cbfa1, (B and enlarged inserts) Dmp1, and
(C and enlarged inserts) OC. Nuclei are stained with methyl green and
hybridization signals are shown by the purple AP staining. The white
arrow, Dmp1 in an osteocyte; the red arrow, Dmp1 in a hypertrophic
chondrocyte; the blue arrow, OC in an osteoblast.
FIG. 6. Expression of Dmp1 during embryonic versus postnatal bone
development. The upper panels (A–E) are a model showing the se-
quential stages in endochondral bone formation and the lower panels
show in situ hybridization for Dmp1 in developing bones at various
developmental stages corresponding to the schematic representations
above. (A) Mesenchymal cell condensation, (B) cartilage formation,
(C) hypertrophic cartilage formation in the center of the diaphysis
expression of Cbfa1 and Dmp1 was linked closely in 2T3
osteoblast cultures, which mineralize in vitro. Third, forced
expression of Cbfa1 in a nonosteoblast cell line that had low
corresponding to stage E16 below (c and c1, an enlarged insert), (D) endogenous expression of Dmp1 induced Dmp1 expression
initial formation of a mineralized bone collar in the diaphysis corre-
sponding to stage E18 below (d and d1, an enlarged insert), and (E)
advanced mineralization of the diaphysis with formation of osteocyte
lacunae corresponding to the 8-week-old postnatal stage shown below
(e and enlarged inserts equivalent to hypertrophic chondrocytes [e1]
osteoblasts in vitro, which bound to nuclear extracts from
and osteocytes [e2]). Note that in developing bones, Dmp1 is expressed
in the hypertrophic cartilage at stage E16, followed by expression in
by the cells and, finally, Dmp1 expression was completely
absent in the bones of Cbfa1Ϫ/Ϫ mice. Interestingly, we
identified a putative Cbfa1 response element (OSE-2) in the
Dmp1 promoter region (Ϫ165 bp) in both odontoblasts and
these cells. However, our transient promoter studies sug-
gested that Cbfa1 did not directly regulate the mouse Dmp1
osteoblasts at stage E18. In 8-week-old postnatal bones, Dmp1 expres-
promoter and that mutation of OSE-2 did not change the
basal Dmp1 promoter activity. Therefore, it seems likely
that regulation of Dmp1 by Cbfa1 is not mediated directly
through this OSE-2 site but may be via an indirect pathway.
Perhaps the most surprising finding was that although
sion is high in both hypertrophic chondrocytes in the growth plate and
osteocytes.
forced expression of the gene in nonskeletal cells results in
activation of genes characteristic of the osteoblast pheno- deletion of the Cbfa1 gene completely abolished Dmp1
type such as ALP, OC, and BSP.(24) We now provide expression in the developing bones, it had very little effect
multiple lines of evidence suggesting that Dmp1 is regu- on Dmp1 expression in odontoblasts. Indeed, the level of
lated by Cbfa1. First, the expression of both genes was Dmp1 expression in arrested tooth buds was similar to that
found to overlap in hypertrophic chondrocytes and bone, in the normal tooth. Odontoblasts have been shown to
with expression of Cbfa1 preceding that of Dmp1. Second, express Cbfa1 by Western blotting, RT-PCR, and super gel

1830
FENG ET AL.
shift assays.(38) The reason for the difference in the expres-
sion of Dmp1 in odontoblasts and osteoblasts in the absence
of Cbfa1 is not clear at this time and these observations
indicate that the different mechanisms may control Dmp1
expression in bone and tooth cells. Interestingly, our gel
shift assays using an oligonucleotide corresponding to
Ϫ157–Ϫ175 bp of the Dmp1 promoter, which contained a
putative Cbfa1 response element, showed a DNA-protein
complex consistent with binding to Cbfa1, which could be
competed with classical OSE-2. This complex was observed
using nuclear extracts from both osteoblasts and odonto-
blasts. However, an additional lower molecular weight com-
plex, which was not competed by OSE-2 was also observed
exclusively in odontoblasts. We believe that this complex
may represent an odontoblast-specific nuclear regulatory
protein, which could explain, in part, the difference in
Dmp1 regulation in bone and teeth. The further character-
ization of this protein currently is underway in our labora-
tory.
In conclusion, the results of our study indicate that the
Dmp1 gene is highly active both early in embryonic bone
development (hypertrophic chondrocytes and osteoblasts)
and later during postnatal bone formation (osteocytes) and
that its expression is indirectly regulated by Cbfa1. There-
fore, Dmp1 may represent a unique new marker gene for
osteoblastic differentiation. The close association of Dmp1
and Cbfa1 in the developing skeleton suggests that Dmp1
may play an important role in early bone formation.
6. Lian JB, Stein GS, Canalis E, Roby PG, Boskey AL 1999
Bone formation: Osteoblast lineage cells, growth factors, ma-
trix proteins, and the mineralization process. In: Favus MJ
(ed.) Primer on the Metabolic Bone Diseases and Disorders of
Mineral Metabolism, 4th ed. Lippincott Williams & Wilkins
Press, New York, NY, USA, pp. 14–29.
7. Ganss B, Kim RH, Sodek J 1999 Bone sialoprotein. Crit Rev
Oral Biol Med 10:79–98.
8. George A, Sabsay B, Simonian PA, Veis A 1993 Character-
ization of a novel dentin matrix acidic phosphoprotein. Impli-
cations for induction of biomineralization. J Biol Chem 268:
12624–12630.
9. George A, Silberstein R, Veis A 1994 In situ localization and
chromosomal mapping of the AG1 (Dmp1) gene. J Histochem
Cytochem 42:1527–1531.
10. MacDougall M, DuPont BR, Simmons D, Leach RJ 1996
Assignment of DMP1 to human chromosome 4 band q21 by in
situ hybridization. Cytogenet Cell Genet 74:189.
11. Hirst KL, Simmons D, Feng J, Aplin H, Dixon MJ, MacDou-
gall M 1997 Elucidation of the sequence and the genomic
organization of the human dentin matrix acidic phosphoprotein
1 (DMP1) gene: Exclusion of the locus from a causative role
in the pathogenesis of dentinogenesis imperfecta type II.
Genomics 42:38–45.
12. Aplin HM, Hirst KL, Crosby AH, Dixon MJ 1995 Mapping of
the human dentin matrix acidic phosphoprotein gene (DMP1)
to the dentinogenesis imperfecta type II critical region at
chromosome 4q21. Genomics 30:347–349.
13. Butler WT, Ritchie HH 1995 The nature and functional sig-
nificance of dentin extracellular matrix proteins. Int J Dev Biol
39:169–179.
14. George A, Silberstein R, Veis A 1995 In situ hybridization
shows Dmp1 (AG1) to be a developmentally regulated dentin-
specific protein produced by mature odontoblasts. Connect
Tissue Res 33:67–72.
15. Hirst KL, Ibaraki-O’Connor K, Young MF, Dixon MJ 1997
Cloning and expression analysis of the bovine dentin matrix
acidic phosphoprotein gene. J Dent Res 76:754–760.
16. D’Souza RN, Cavender A, Sunavala G, Alvarez J, Ohshima T,
Kulkarni AB, MacDougall M 1997 Gene expression patterns
of murine dentin matrix protein 1 (Dmp1) and dentin sialo-
phosphoprotein (DSPP) suggest distinct developmental func-
tions in vivo. J Bone Miner Res 12:2040–2049.
17. Toyosawa S, Shintani S, Fujiwara T, Ooshima T, Sato A,
Ijuhin N, Komori T 2001 Dentin matrix protein 1 is predom-
inantly expressed in chicken and rat osteocytes but not in
osteoblasts. J Bone Miner Res 16:2017–2026.
ACKNOWLEDGMENTS
The authors acknowledge Dr. Lynda Bonewald, Depart-
ment of Oral Biology, University of Missouri-Kansas City,
and Dr. David L. Carnes, Department of Periodontics, Uni-
versity of Texas Health Science Center (UTHSC) at San
Antonio, TX, for critical review and assistance in the prep-
aration of this study. This work was supported in part by the
UM Research Board from the University of Missouri, MO,
USA, and U.S. Public Health Service (USPHS) research
grants (to J.Q.F, DE12057, DE13480, and DE00455; and to
M.M., DE13221) from the National Institutes of Health
(NIH), Bethesda, MD, USA.
18. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K,
Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, Sato
M, Okamoto R, Kitamura Y, Yoshiki S, Kishimoto T 1997
Targeted disruption of Cbfa1 results in a complete lack of bone
formation owing to maturational arrest of osteoblasts. Cell
89:755–764.
REFERENCES
19. D’Souza RN, Aberg T, Gaikwad J, Cavender A, Owen M,
Karsenty G, Thesleff I 1999 Cbfa1 is required for epithelial-
mesenchymal interactions regulating tooth development in
mice. Development 126:2911–2920.
1. Hunter GK, Hauschka PV, Poole AR, Rosenberg LC, Gold-
berg HA 1996 Nucleation and inhibition of hydroxyapatite
formation by mineralized tissue proteins. Biochem J 317:59–
64.
20. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC,
Rosewell IR, Stamp GW, Beddington RS, Mundlos S, Olsen
BR, Selby PB, Owen MJ 1997 Cbfa1, a candidate gene for
cleidocranial dysplasia syndrome, is essential for osteoblast
differentiation and bone development. Cell 89:765–771.
21. Mundlos S, Otto F, Mundlos C, Mulliken JB, Aylsworth AS,
Albright S, Lindhout D, Cole WG, Henn W, Knoll JH, Owen
MJ, Mertelsmann R, Zabel BU, Olsen BR 1997 Mutations
involving the transcription factor CBFA1 cause cleidocranial
dysplasia. Cell 89:773–779.
2. Glimcher MJ 1989 Mechanism of calcification: Role of colla-
gen fibrils and collagen-phosphoprotein complexes in vitro
and in vivo. Anat Rec 224:139–153.
3. Hauschka P, Lian J, Cole D, Gundberg C 1989 Osteocalcin
and matrix Gla protein: Vitamin K-dependent proteins in bone.
Physiol Rev 69:990–1047.
4. Young MF, Kerr JM, Ibaraki K, Heegaard AM, Robey PG
1992 Structure, expression, and regulation of the major non-
collagenous matrix proteins of bone. Clin Orthop 281:275–
294.
5. Denhardt DT, Guo X 1993 Osteopontin: A protein with di- 22. Shamim H, Mahmood R, Mason I 1999 In situ hybridization to
verse functions. FASEB J 7:1475–1482. RNA in whole embryos. In: Sharpe PT, Mason I (eds.) Mo-

STUDY OF Dmp1 EXPRESSION AND REGULATION
1831
lecular Embryology: Methods and Protocols, vol 97. Humana
Press, Clifton, NJ, USA, pp. 623–633.
mesenchymal interactions regulating tooth development in
mice. Development 126:2911–2920.
33. Karsenty G, Ducy P, Starbuck M, Priemel M, Shen J, Geoffroy
V, Amling M 1999 Cbfa1 as a regulator of osteoblast differ-
entiation and function. Bone 25:107–108.
34. Hirst KL, Simmons D, Feng J, Aplin H, Dixon MJ, MacDou-
gall M 1997 Elucidation of the sequence and the genomic
organization of the human dentin matrix acidic phosphoprotein
1 (DMP1) gene: Exclusion of the locus from a causative role
in the pathogenesis of dentinogenesis imperfecta type II.
Genomics 42:38–45.
35. Ducy P, Starbuck M, Priemel M, Shen J, Pinero G, Geoffroy
V, Amling M, Karsenty G 1999 A Cbfa1-dependent genetic
pathway controls bone formation beyond embryonic develop-
ment. Genes Dev 13:1025–1036.
23. MacDougall M, Thiemann F, Ta H, Hsu P, Chen LS, Snead
ML 1995 Temperature sensitive simian virus 40 large T anti-
gen immortalization of murine odontoblast cell cultures: Es-
tablishment of clonal odontoblast cell line. Connect Tissue Res
33:97–103.
24. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G 1997
Osf2/Cbfa1: A transcriptional activator of osteoblast differen-
tiation. Cell 89:747–754.
25. Feng JQ, Chen D, Cooney A, Tsai MJ, Harris MA, Dallas SJ,
Tsai SY, Feng M, Mundy GR, Harris SE 1995 The mouse
bone morphogenetic protein 4 gene: Analysis of promoter
utilization in fetal rat calvarial osteoblasts. J Biol Chem 270:
28364–28373.
36. Schinke T, Karsenty G 1999 Characterization of Osf1, an
osteoblast-specific transcription factor binding to a critical
cis-acting element in the mouse osteocalcin promoters. J Biol
Chem 274:30182–30189.
37. Javed A, Gutierrez S, Montecino M, van Wijnen AJ, Stein JL,
Stein GS, Lian JB 1999 Multiple Cbfa/AML sites in the rat
osteocalcin promoter are required for basal and vitamin
D-responsive transcription and contribute to chromatin orga-
nization. Mol Cell Biol 19:7491–7500.
38. MacDougall M, Uterbrink A, Rani S, Li S, Dodds A, Karsenty
G, Feng J 2000 Cbfa1/OSF2 down-regulates tooth-specific
gene dentin sialophosphoprotein (DSPP) during dentinogene-
sis. J Dent Res 79:339.
26. Bellows CG, Aubin JE, Heersche JN, Antosz ME 1986 Mineral-
ized bone nodules formed in vitro from enzymatically released rat
calvaria cell populations. Calcif Tissue Int 38:143–154.
27. Harris SE, Sabatini M, Harris MA, Feng JQ, Wozney J,
Mundy GR 1994 Expression of bone morphogenetic protein
messenger RNA in prolonged cultures of fetal rat calvarial
cells. J Bone Miner Res 9:389–394.
28. MacDougall M, Gu TT, Luan X, Simmons D, Chen J 1998
Identification of a novel isoform of mouse dentin matrix pro-
tein 1: Spatial expression in mineralized tissues. J Bone Miner
Res 13:422–431.
29. Ghosh-Choudhury N, Windle JJ, Koop B, Harris MA, Guer-
rero DL, Wozney JM, Mundy GR, Harris SE 1996 Immortal-
ized murine osteoblasts derived from BMP 2-T-Antigen ex-
pressing transgenic mice. Endocrinology 137:331–339.
30. Gorski JP 1998 Is all bone the same? Distinctive distributions
and properties of non-collagenous matrix proteins in lamellar
vs. woven bone imply the existence of different underlying
osteogenic mechanisms. Crit Rev Oral Biol Med 9:201–223.
31. Frendo JL, Xiao G, Fuchs S, Franceschi RT, Karsenty G, Ducy
P 1998 Functional hierarchy between two OSE2 elements in
the control of osteocalcin gene expression in vivo. J Biol Chem
273:30509–30516.
Address reprint requests to:
Jian Q. Feng, M.D., Ph.D.
Department of Oral Biology
University of Missouri-Kansas City
650 East 25th Street
Kansas City, MO 64108, USA
32. D’Souza RN, Aberg T, Gaikwad J, Cavender A, Owen M, Received in original form March 7, 2001; in revised form January
Karsenty G, Thesleff I 1999 Cbfa1 is required for epithelial- 23, 2002; accepted February 6, 2002.