Arginine methyltransferase CARM1/PRMT4 regulates endochondral ossification.

Abstract

Chondrogenesis and subsequent endochondral ossification are processes tightly regulated by the transcription factor Sox9 (SRY-related high mobility group-Box gene 9), but molecular mechanisms underlying this activity remain unclear. Here we report that coactivator-associated arginine methyltransferase 1 (CARM1) regulates chondrocyte proliferation via arginine methylation of Sox9.
CARM1-null mice display delayed endochondral ossification and decreased chondrocyte proliferation. Conversely, cartilage development of CARM1 transgenic mice was accelerated. CARM1 specifically methylates Sox9 at its HMG domain in vivo and in vitro. Arg-methylation of Sox9 by CARM1 disrupts interaction of Sox9 with beta-catenin, regulating Cyclin D1 expression and cell cycle progression of chondrocytes.
These results establish a role for CARM1 as an important regulator of chondrocyte proliferation during embryogenesis.

Full text

Page 1

BioMed Central
Page 1 of 10
(page number not for citation purposes)
BMC Developmental Biology
Open Access
Research article
Arginine methyltransferase CARM1/PRMT4 regulates
endochondral ossification
Tatsuo Ito1, Neelu Yadav2, Jaeho Lee2, Takayuki Furumatsu1,
Satoshi Yamashita4, Kenji Yoshida1, Noboru Taniguchi1,
Megumi Hashimoto4, Megumi Tsuchiya4, Toshifumi Ozaki3, Martin Lotz1,
Mark T Bedford2 and Hiroshi Asahara*1,4
Address: 1Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, CA 92037, USA, 2Department of
Carcinogenesis, University of Texas M. D. Anderson Cancer Center, PO Box 389, Smithville, TX 78957, USA, 3Department of Orthopedic Surgery,
Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama 700-8558, Japan and 4Department of
Regenerative Medicine, National Research Institute for Child Health and Development, Tokyo 157-8535, Japan
Email: Tatsuo Ito - itot@mskcc.org; Neelu Yadav - neeluyadav@hotmail.com; Jaeho Lee - jhlee@mdanderson.org;
Takayuki Furumatsu - matino@md.okayama-u.ac.jp; Satoshi Yamashita - syamashita@nch.go.jp;
Kenji Yoshida - k_yoshida@research.otsuka.co.jp; Noboru Taniguchi - tanigu@scripps.edu; Megumi Hashimoto - gyu@med.showa-u.ac.jp;
Megumi Tsuchiya - mtsuchiya@fbs.osaka-u.ac.jp; Toshifumi Ozaki - tozaki@md.okayama-u.ac.jp; Martin Lotz - mlotz@scripps.edu;
Mark T Bedford - mtbedford@mdanderson.org; Hiroshi Asahara* - asahara@nch.go.jp
* Corresponding author
Abstract
Background: Chondrogenesis and subsequent endochondral ossification are processes tightly
regulated by the transcription factor Sox9 (SRY-related high mobility group-Box gene 9), but
molecular mechanisms underlying this activity remain unclear. Here we report that coactivator-
associated arginine methyltransferase 1 (CARM1) regulates chondrocyte proliferation via arginine
methylation of Sox9.
Results: CARM1-null mice display delayed endochondral ossification and decreased chondrocyte
proliferation. Conversely, cartilage development of CARM1 transgenic mice was accelerated.
CARM1 specifically methylates Sox9 at its HMG domain in vivo and in vitro. Arg-methylation of Sox9
by CARM1 disrupts interaction of Sox9 with beta-catenin, regulating Cyclin D1 expression and cell
cycle progression of chondrocytes.
Conclusion: These results establish a role for CARM1 as an important regulator of chondrocyte
proliferation during embryogenesis.
Background
The precise patterning of the developing skeletal frame-
work relies on the appropriate control of chondrogenesis,
a multistep process during which mesenchymal cells dif-
ferentiate into chondrocytes[1,2]. This process is tightly
regulated by transcription factors, including Sox9 [3-6].
Mice lacking Sox9 display distortion of numerous carti-
lage-derived skeletal structures[7]. In addition, mice over-
expressing Sox9 in chondrocytes show dwarfism with
decreased chondrocyte proliferation and delayed endo-
Published: 2 September 2009
BMC Developmental Biology 2009, 9:47doi:10.1186/1471-213X-9-47
Received: 17 December 2008
Accepted: 2 September 2009
This article is available from: http://www.biomedcentral.com/1471-213X/9/47
© 2009 Ito et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

BMC Developmental Biology 2009, 9:47 http://www.biomedcentral.com/1471-213X/9/47
Page 2 of 10
(page number not for citation purposes)
chondral bone formation[1]. However, the precise mech-
anism how Sox9 regulates chondrogenesis both spatially
and temporally is still largely unknown.
CARM1 belongs to a family of arginine-specific protein
methyltransferases (PRMTs), which includes at least eight
members (PRMT1-8) [8]. All PRMT family members share
a core arginine methyltransferase region composed of a
conserved Ado-Met binding domain and a more divergent
C-terminal domain. CARM1 has been shown to synergis-
tically activate transcription with nuclear receptors in
combination with other coactivators, such as p160 family,
p300/CBP and SRC-2/TIF2/GRIP1[9,10]. After recruit-
ment to promoters of steroid-responsive genes, CARM1
methylates specific residues (Arg17 and Arg26) at the N-
terminus of histone H3 resulting in transcriptional activa-
tion[11,12].
Results and Discussion
Skeletal phenotype of CARM1 null mice
To examine the potential role of CARM1 in skeletal devel-
opment, we analyzed CARM1 null embryos, which die
immediately after birth as reported[13]. At E14.5, before
ossification starts, embryos did not exhibit significant dif-
ferences by alcian blue and alizarin red staining. However,
at E16.5, ossification in null mutants was remarkably
delayed, and the size of null mutant was smaller com-
pared with heterozygotes (Figure 1A, B: SafraninO). At
E18.5, the difference between null and wild type gets rel-
atively weak (data not shown). Using von Kossa staining,
we observed that absorption of calcified cells in null
mutants at E16.5 was delayed (Figure 1C: von Kossa). In
situ hybridization (ISH) of, Col2a1, Col10a1, Bone Sialopro-
tein (BSP), Osteopontin (Op), Osteocalcin (Oc) and Runx2
in mutant embryos supported the conclusion that endo-
chondral bone formation was significantly delayed (Addi-
tional file 1). Importantly, BrdU pulse-labeling of cells in
E16.5 CARM1-null and wild type mouse embryos
revealed a marked reduction in the number of BrdU-pos-
itive chondrocytes in mutant compared with wild type
embryos (Figure 1D: áBrdU), indicating that chondrocyte
proliferation in mutant embryos was inhibited.
Bone development of CARM1-transgenic mice
For gain-of-function analysis, we generated transgenic
mice in which CARM1 expression is driven by the ubiqui-
tously expressed beta-actin promoter (Additional file 2).
In contrast to null mutants, CARM1-transgenic mice ana-
lyzed at E18.5 were larger than controls (Figure 1E: Dou-
ble stain, Tg). Alizarin red positive regions appeared at the
shaft of humerus in E14.5 CARM1-transgenic mice but
were not visible in wild type mice (Figure 1E: Double
stain, Tg limb, arrow heads). von Kossa staining showed
absorption of calcified regions was also accelerated in
E14.5 transgenic compared with wild type mice (Figure
1F: von Kossa), while SafraninO staining was unchanged
and chondrocyte differentiation marker, Col2a1 and
Col10a1 expression was not significantly altered in carti-
lage in E14.5 transgenic compared to wild type mice (Fig-
ure 1F: SafraninO, Tg, Additional file 1). These data
indicate that endochondral bone formation in CARM1
transgenic mice is accelerated relative to wild type mice,
although we could not exclude the possibility that
CARM1 may also directly regulate osteoblasts differentia-
tion.
CARM1 expression in growth plates
To evaluate a potential functional link between Sox9 and
CARM1, we examined their expression patterns during
bone formation. CARM1 mRNA expression was high in
proliferating chondrocytes of growth plates. Immunohis-
tochemistry of Sox9 and ISH analysis of Col2a1 show that
Sox9 and Col2a1 expression overlaps with that of CARM1
at the proliferating zone of wild type E15.5 growth plates
(Figure 2A), although chondrocytes at the prehyper-
trophic zone express lower levels of CARM1 but relatively
abundant Sox9 and Col2a1[14].
Sox9 and CARM1 interaction
We next asked whether CARM1 and Sox9 interact. In a
GST-pull down assay, full length GST-Sox9 protein bound
to in vitro translated CARM1 protein, suggesting direct
interaction (Figure 2B: GST-pull down). To define the
interaction domain of Sox9 with CARM1, we performed
the assay using in vitro translated CARM1 and bacterially
expressed GST-Sox9 fragments. Full length GST-Sox9 (1-
507aa) and 328-508aa and 423-508aa fragments inter-
acted with CARM1, while 1-327aa and 1-422aa fragments
did not (Figure 2B GST-pull down). Taken together,
CARM1 likely interacts with Sox9 via the Sox9 carboxy-
terminal domain, which was previously reported to be a
transcription activation domain[15].
CARM1 methylates Sox9
Since CARM1 methylates some non-histone proteins,
such as CBP/p300[16] and splicing factors[17], we asked
whether Sox9 can serve as a substrate for CARM1. Recom-
binant Flag-Sox9 was incubated with recombinant
CARM1/PRMT4 or PRMT1 in the presence of 3H-labeled
S-adenosyl-L-methionine ([3H]AdoMet) as a methyl
donor. Histone H3 was assayed as a positive control
because H3 can be methylated by CARM1 and
PRMT1[18]. We observed that Sox9 could be methylated
by CARM1/PRMT4, but not by PRMT1 (Figure 2C).
Arginine methylation was also seen in endogenous Sox9
immunoprecipitated from mouse (Figure 2D, left) or
human (data not shown) primary cultured chondrocytes.
A recent report showing that the HMG domain of
HMGA1a can be arginine-methylated[19] prompted us to
determine whether arginine residues in the HMG domain

Page 3

BMC Developmental Biology 2009, 9:47http://www.biomedcentral.com/1471-213X/9/47
Page 3 of 10
(page number not for citation purposes)
Analysis of skeletal phenotypes in CARM1-null mutant embryos
Figure 1
Analysis of skeletal phenotypes in CARM1-null mutant embryos. (A) Appearance of skeletons of E14.5 and E16.5
embryos stained with alcian blue (cartilage) followed by alizarin red (bone). Forelimbs from E16.5 wild type (wt) (top) and null
(lower) siblings show shortened bones. (B) Histological analysis of limbs of CARM1-null mutant embryos. SafraninO staining of
humerus of E16.5 embryos. (B: a & b) Boxed regions are shown at higher magnification. (C) Staining by von Kossa's method in
the humerus of E16.5 heterozygous and mutant embryos. (D) BrdU incorporation of humerus of E16.5 embryos. Differences,
assessed by one-way analysis of variance and an unpaired Student's t-test (*), are significant p < 0.001. (E) Whole skeletal prep-
arations of E14.5 and E18.5 CARM1 transgenic (Tg) and wt littermates. Arrow indicates a calcificated region in the shaft of the
humerus of E14.5 Tg embryos. (F) Histological analysis of limbs in E14.5 wt and Tg embryos. SafraninO staining of the humerus
in E14.5 embryos. Staining using von Kossa's method visualizes mineral deposition in the humerus of E14.5 Tg embryos.

Page 4

BMC Developmental Biology 2009, 9:47http://www.biomedcentral.com/1471-213X/9/47
Page 4 of 10
(page number not for citation purposes)
CARM1 interacts with and methylates Sox9
Figure 2
CARM1 interacts with and methylates Sox9. (A) Histological analysis of E15.5 wt embryos. Sections of humerus were
stained with CARM1 and Col2a1 mRNA probes. Sox9 is stained by ISH and immunohistochemistry (IHC). (B) Sox9 interacts
with CARM1 in vitro. Recombinant CARM1 protein and GST-Sox9 fragments were mixed and subjected to a GST-pull down
assay. (C) Sox9 recombinant proteins and Histone H3 were incubated with PRMT1 and CARM1 in the presence of [3H]
AdoMet. (D) Endogenous Sox9 methylation was detected in mouse primary cultured chondrocytes (left panel). Sox9 multiple
point mutants a (mt(a)) and b (mt(b)), wt Sox9, Flag-tagged PABP1 (as a positive control) and Flag-tagged Runx2 (as a negative
control) expression plasmids were transfected into 293T cells (right panels). (E) Schematic representation of Sox9 mutants
used in the methylation assay.

Page 5

BMC Developmental Biology 2009, 9:47 http://www.biomedcentral.com/1471-213X/9/47
Page 5 of 10
(page number not for citation purposes)
of Sox9 (74-179aa) are CARM1 targets. Based on in vitro
methylation assays, simultaneous mutation of five
arginine residues within the HMG domain (mt(b)) com-
pletely abolished methylation signals, whereas mutants
exhibiting R177K, R178K and R179K (mt(a)) were meth-
ylated similarly to wild type (Figure 2D, right &2E). These
data indicate that CARM1 methylates multiple arginine
residues within the HMG domain of Sox9.
CARM1 regulates chondrocytes proliferation
Decreased chondrocyte proliferation phenotypes seen in
Sox9 transgenic mice are partly explained by the model
that Sox9 competes with Tcf/Lef for binding to beta-cat-
enin and regulates chondocyte proliferation via Cyclin D1
expression. The beta-catenin/Tcf complex binds the Tcf/
Lef consensus site in the Cyclin D1 promoter, transactivat-
ing Cyclin D1[20]. Sox9 reportedly inhibits transcriptional
beta-catenin activity, an inhibition that does not result
from competition by Sox9 with Tcf/Lef for Tcf/Lef DNA-
binding sites[21]. To determine whether arginine methyl-
ation of Sox9 modulates its interaction with beta-catenin,
we transfected wild type and methylation point mutant
forms of Flag-tagged Sox9 expression plasmids into
SW1353 cells, a chondrosarcoma cell line. Immunopre-
cipitation analysis showed that CARM1 overexpression
decreased the interaction between Flag-tagged Sox9 and
endogenous beta-catenin proteins (Figure 3A: top). By
contrast, protein with mutation of all Sox9 methylation
sites (residues 74, 152, 177, 178, and 179, R to K) strongly
interacted with beta-catenin, even when CARM1 was over-
expressed, suggesting that methlyation of CARM1 inhibits
the interaction between Sox9 and beta-catenin or that
CARM1 binding to Sox9 blocks beta-catenin binding. We
also checked whether these mutations on Sox9 may effect
on cellular localization. Mutants Sox9 R177K, R178K and
R179K were mainly located in the nucleus as well as wild
type Sox9, when they were orverexpressed in the SW1353
cell line (data not shown).
CARM1 regulates Cyclin D1 expression
Cyclin D1 expressed in proliferating chondrocytes of
growth plates is required for normal chondrocyte prolifer-
ation[22,23]. In growth plates of E16.5 CARM1-null
mutants, Cyclin D1 mRNA levels were reduced (Figure 3B:
ISH of CyclinD1 in CARM1-null), whereas Col2a1 was
expressed in the limb bud of CARM1-null mutants (Figure
3B: RT-PCR). By contrast, Cyclin D1 mRNA levels were
remarkably increased in E14.5 CARM1-transgenic mice
(Figure 3B: ISH of Cyclin D1 in CARM1-Tg). These data
suggest that CARM1 promote Cyclin D1 gene expression.
To determine how CARM1 regulates Cyclin D1 expression,
we performed Cyclin D1 promoter assays in human
embryonic kidney 293 cells, which express endogenous
Lef1. On the -963CD1 Cyclin D1 promoter reporter plas-
mid and on a Tcf mt reporter plasmid with a mutation in
the Tcf binding site[20], Sox9 inhibited basal promoter
activity, as previously reported (Figure 3C). CARM1 over-
expression rescued inhibition by Sox9 in this assay, sug-
gesting that CARM1 may inhibit Sox9 and beta-catenin
complex formation and thus increase Cyclin D1 expres-
sion. In contrast CARM1 and Sox9 did not activate tran-
scription using the Tcf mt reporter plasmid, which lacks
beta-catenin reactive sites. With the mutation at all
74,152,177,178,179 shows strong suppression at
CyclinD1 promoter, however, this suppression did not
relieved by overexpression of CARM1 (data not shown).
This suggests that CARM1 and Sox9 may cooperatively
regulate CyclinD1 promoter activity.
Evidence of CARM1-dependent Cyclin D1 regulation
prompted us to ask whether reduced chondrocyte prolif-
eration may partly explain bone development phenotypes
seen in CARM1-null mutant and transgenic embryos.
Consistent with BrdU experiments (Figure 1D: BrdU
stain), CARM1-null chondrocytes showed remarkably
lower Cyclin D1 mRNA than that seen in CARM1-hetero-
zygotes and wild type chondrocytes (Figure 4A: RT-PCR).
CARM1-heterozygotes chondrocytes showed quicker dis-
appearance of Cyclin D1 mRNA than wild type. Further-
more, 10.60% of CARM1-null chondrocytes were in S-
phase, compared with 16.85% in heterozygous chondro-
cytes, a typical shift seen following Cyclin D1 downregula-
tion (Figure 4B: FACS) [22]. Taken together, these data
support the idea that reduced chondrocyte proliferation
in CARM1-null mutant embryos is partly due to inhibi-
tion of Cyclin D1 expression (Figure 4C).
Conclusion
Sox9 function is likely regulated by post-translational
modifications such as phosphorylation, ubiquitination or
sumoylation [24-26]. Here we demonstrate that Sox9 is
Arg-methylated by CARM1. CARM1 is involved in the epi-
genetic programming of early embryo development[27],
and also early T cell development[28]. Our study shows
that CARM1 also plays a key role in cartilage develop-
ment. Further analysis of the role of CARM1 on chromat-
inized and non-chromatinized substrates during different
developmental stages should indicate the physiological
role of arginine methylation signals.
Methods
Tissue Culture Methods and Transfections
The human chondrosarcoma cell line (SW1353) was
grown in Dulbecco's modified Eagle's medium (Cellgro,
Mediatech) supplemented with 10% fetal calf serum and
penicillin/streptomycin (Sigma). Cells were transfected
using GeneJammer (Stratagene) according to the manu-
facturer. Amounts of transfected plasmids were as follows:
for Sox9-dependent activation, 50 ng of the Cyclin D1 pro-
moter plasmids, -963CD1, Tcf mt, and 100 ng of