Abstract
Free full text
This article has been retracted. Retraction in:
This article has been retracted. Retraction in:
Vitamin K Induces Osteoblast Differentiation through Pregnane X Receptor-Mediated Transcriptional Control of the Msx2 Gene
Abstract
Vitamin K is a fat-soluble vitamin that serves as a coenzyme for vitamin K-dependent carboxylase. Besides its canonical action, vitamin K binds to the steroid and xenobiotic receptor (SXR)/pregnane X receptor (PXR) and modulates gene transcription. To determine if the osteoprotective action of vitamin K is the result of the PXR/SXR pathway, we screened by two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis the PXR/SXR target genes in an osteoblastic cell line (MC3T3-E1) treated with a vitamin K2 (menaquinone 4 [MK4]). Osteoblastic differentiation of MC3T3-E1 cells was induced by MK4. Msx2, an osteoblastogenic transcription factor, was identified as an MK4-induced gene. Functional analysis of the Msx2 gene promoter mapped a vitamin K-responsive element (PXR-responsive element [PXRE]) that was directly bound by a PXR/retinoid X receptor α heterodimer. In a chromatin immunoprecipitation analysis, PXR was recruited together with a coactivator, p300, to the PXRE in the Msx2 promoter. MK4-bound PXR cooperated with estrogen-bound estrogen receptor α to control transcription at the Msx2 promoter. Knockdown of either PXR or Msx2 attenuated the effect of MK4 on osteoblastic differentiation. Thus, the present study suggests that Msx2 is a target gene for PXR activated by vitamin K and suggests that the osteoprotective action of MK4 in the human mediates, at least in part, a genomic pathway of vitamin K signaling.
The K vitamins are a group of fat-soluble vitamins that occur in two natural forms: phyloquinones (K1) and menaquinones (K2). Vitamin K (VK) has classically been associated with blood coagulation (31). In its canonical role, VK serves as a coenzyme for VK-dependent carboxylase. This enzyme converts glutamate residues into γ-carboxyglutamate (Gla) residues in VK-dependent proteins, such as prothrombin, and factors IX and X (6, 10, 29). Such VK-induced protein modification also occurs in osteocalcin (7, 21) and matrix Gla protein (MGP) (22). Thus, VK may exert beneficial effects on bone formation and remodeling. In fact, animal studies suggest that VK deficiency results in a reduction in bone mass together with hypocarboxylation of osteocalcin (25).
Clinically, the most common form of K2, menaquinone 4 (MK4), has been shown to prevent bone fractures (3). This osteoprotective effect is more pronounced in K2 than in K1, and hence MK4 has been used to treat osteoporotic patients in Japan (9, 11, 28). However, the bone phenotypic abnormalities of mice deficient in osteocalcin and MGP do not fully support the classical view that the osteoprotective action of VK is the result of the modification of skeletal proteins. These mice, which are genetically deficient for osteocalcin or MGP, exhibited bone mass increases instead of losses (4). This suggests that the osteoprotective action of VK is mediated by another pathway.
MK4 recently has been shown to act as a ligand for the steroid and xenobiotic receptor (SXR) in human osteoblastic cells (33). It transcriptionally regulates gene expression and represents a new pathway of VK action. The SXR and its mouse homolog, the pregnane X receptor (PXR), respond to xenobiotics and pregnenones. PXR and SXR are members of the nuclear receptor (NR) gene superfamily and bind to specific DNA elements (PXR-responsive elements [PXRE]) as heterodimers with one of the retinoid X receptor (RXR) subtypes (α, β, and γ) (1, 2, 15, 16, 18). Like the other NR members, ligand binding to PXR/SXR induces dissociation of corepressors and recruitment of coactivators for ligand-induced transactivation in the target gene promoters (17). Thus, these findings suggest that it is feasible that the osteoprotective VK action mediates its transcriptional control of the VK target genes via PXR/SXR. In fact, several PXR/SXR genes recently have been shown to transcriptionally respond to VK (8).
To test this idea, we screened VK target genes in an osteoblastic cell line (MC3T3-E1) treated with MK4 with two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (2D SDS-PAGE). A prime osteoblastogenic factor, Msx2, was identified, and a PXRE was located in its gene promoter. MK4 interacted with the PXRE via PXR/RXRα binding in vivo and in vitro. Osteoblast genesis in MC3T3-E1 cells was induced by MK4, but knockdown of Msx2 by RNA interference (RNAi) abrogated the MK4 effect. The present study suggests that Msx2 is a target gene for VK-activated PXR/SXR. It implies that the osteoprotective VK action takes place, at least in part, on a genomic level by stimulating osteoblast differentiation through Msx2 gene induction.
MATERIALS AND METHODS
Plasmids.
The full-length cDNA for the mouse PXR (15) was subcloned into a pcDNA3 expression vector (Invitrogen) tagged with the hemagglutinin epitope at the N terminus. The mouse Msx2 promoters, the sequences of which were derived from the Ensembl genome browser (http://www.ensembl.org/index.html), were subcloned into a pGL3-Basic vector (Promega).
Animals.
Estrogen receptor α knockout (ERαKO) mice were kindly provided by P. Chambon (19). Genotyping for ERαKO mice was routinely performed on DNA isolated from tail snips by a PCR procedure (19).
Cell culture, transient transfection, and luciferase assay.
MC3T3-E1 and ST2 cells were cultured with alpha minimum essential medium (α-MEM) (GIBCO) containing 10% fetal bovine serum (FBS) (GIBCO) at 37°C and 5% CO2 (31). HEK 293T cells were cultured with Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS (GIBCO) at 37°C and 5% CO2. For transfection, cells were plated in the corresponding medium supplemented with 10% charcoal-stripped FBS in 12-well plates 1 day before transfection. Transfection was performed with Lipofectamine (Invitrogen) with Plus reagent (Invitrogen) or Polyfect (QIAGEN) as directed by the manufacturer's protocol. After 3 h, 17β-estradiol (E2) (10 nM) and MK4 (10 μM) were added to α-MEM containing 2% FBS, and the cells were incubated continuously at 37°C for 24 h. As a reference to normalize transfection efficiency, 2.5 ng/well of pRL-CMV plasmid (Promega) was cotransfected in all experiments. Luciferase activity was determined using the luciferase assay system (Promega) (5).
Osteoblast primary culture.
Calvaria of newborn mice were digested for 80 min at 37°C in phosphate-buffered saline (PBS) containing 0.1% collagenase A (Roche). Cells were plated with α-MEM containing 10% FBS. Media were changed every 3 days (32).
ALP staining.
For alkaline phosphatase (ALP) staining, primary osteoblasts, MC3T3-E1 and ST2 cells, were washed twice with PBS, fixed in 3.7% formaldehyde, and stained with a mixture of 0.1 mg/ml naphthol AS-MX phosphate (Sigma), 0.6 mg/ml fast-blue BB salt (Sigma), 2 mM MgCl2, 5 μl/ml N,N-dimethylforamide (Wako), and 100 mM Tris-HCl (pH 8.8) buffer at 37°C for 5 to 10 min. When the cells turned blue, the cells were washed twice with PBS (32).
ALP assay.
The ALP assay was performed using the LabAssay ALP kit (Wako) according to the manufacturer's instructions. In brief, primary osteoblasts, MC3T3-E1 and ST2 cells, were washed twice with ice-cold PBS and then were solubilized with lysis buffer (20 mM Tris-HCl [pH 7.9], 1% NP-40, 1 mM EDTA, 150 mM NaCl, 2.5 mM MgCl2, 5% glycerol) containing protease inhibitors, followed by determination of the ALP activity in lysates. The protein concentration was determined with a Bio-Rad protein assay kit.
qRT-PCR.
For quantitative real-time reverse transcription-PCR (qRT-PCR), 1 μg of total RNA from each sample was reverse transcribed into first-strand cDNA with random hexamers using Superscript III reverse transcriptase (Invitrogen). Primer sets for all genes were purchased from Takara Bio Inc. (Tokyo, Japan). Real-time RT-PCR was performed using SYBR premix EX Taq (Takara) with the thermal cycler Dice RealTime System TP800 (Takara) according to the manufacturer's instructions. Experimental samples were matched to a standard curve generated by amplifying serially diluted products using the same PCR protocol. To correct for variability in RNA recovery and the efficiency of reverse transcription, glyceraldehyde-3-phosphate dehydrogenase cDNA was amplified and quantified in each cDNA preparation. Normalization and calculation steps were performed as reported previously (34).
2D SDS-PAGE analysis.
For 2D SDS-PAGE analysis, MC3T3-E1 cells were washed with ice-cold PBS, collected by centrifugation at 2,000 × g, resuspended in 50 μl lysis buffer (20 mM Tris-HCl [pH 7.9], 1% NP-40, 1 mM EDTA, 150 mM NaCl, 2.5 mM MgCl2, 5% glycerol) containing protease inhibitors, incubated on ice for 30 min, and then centrifuged for 30 min at 12,000 × g. After centrifugation, the supernatants were cleaned with a 2D-Clean-Up kit (Amersham). These samples were separated with the 2D SDS-PAGE system (Bio-Rad). The 2D SDS-PAGE gels were visualized with the SilverQuest silver staining kit (Invitrogen).
ABCD assay.
For the avidin-biotin complex DNA (ABCD) assay, sense and antisense oligonucleotide DNAs that were biotinylated at the 3′ terminus were incubated at 100°C for DNA annealing and then were cooled slowly at room temperature. To prepare the beads-DNA complex, biotin-conjugated double-stranded oligonucleotide DNA was mixed with a 50% slurry of avidin beads (Tetralink tetrameric avidin resin; Promega). Cells were lysed with lysis buffer (10 mM Tris-Cl [pH 7.8], 1 mM EDTA, 150 mM NaCl, 0.1% NP-40) containing protease inhibitors for 1 h at 4°C. Whole-cell lysates were clarified by centrifugation, mixed with the 50% slurry of avidin beads, and pelleted at 3,000 × g. Supernatants were added to the beads-DNA complex and rotated for 1 h at 4°C to mix. The beads were collected by centrifugation at 3,000 × g and were washed successively in the lysis buffer. Proteins were resolved with SDS-PAGE, and Western blotting was performed with the corresponding antibody (12).
ChIP analysis.
Chromatin immunoprecipitation (ChIP) analysis was performed using the ChIP assay kit (Upstate Biotechnology Inc.) (13, 38) according to the manufacturer's instructions. MC3T3-E1 cells were cultured in the presence or absence of MK4 and E2. Soluble chromatin prepared from 1 × 106 cells was immunoprecipitated with antibodies against the indicated proteins.
Generation of adenovirus.
Recombinant adenoviruses carrying RNAi of Msx2 and PXR were constructed using the RNAi-Ready pSIREN-Shuttle vector (BD Biosciences Clontech) first and then were moved to the Adeno-X vector (Clontech) by being spliced into the I-CeuI and PI-SceI site. The parental virus genomes in HEK 293 cells (ATCC) were constructed according to the manufacturer's protocol (20). MC3T3-E1 cells then were infected by incubation with the recombinant adenovirus.
RESULTS
Vitamin K2 promotes osteoblast differentiation.
We explored the osteoprotective effects of K2 on a molecular level to better understand how K2 clinically prevents osteoporosis. We first tested the effects of MK4, a K2. The effect of MK4 was measured by ALP assay of primary cultured mice osteoblastic cells, derived from calvaria, to detect osteoblastic differentiation. As shown in the lower panels of Fig. Fig.11 A and B (lanes 1 to 4), ALP activity was induced in the primary cultured osteoblastic cells treated with MK4 for 24 h. Likewise, E2 was stimulatory, and the actions of MK4 and E2 were additive. To clarify the stages of differentiation at which MK4 and E2 act, we tested ST2 cells and MC3T3-E1 cells. ST2 cells are bipotential cells derived from mouse bone marrow mesenchymal stem cells that differentiate into either osteoblasts or adipocytes under the proper conditions (32). MC3T3-E1 cells are derived from mouse calvaria and are preosteoblastic cells that can be induced to differentiate into mature osteoblasts (30). As shown in the middle panels of Fig. 1A and B (lanes 5 to 8), ALP induction following treatment with MK4 and/or E2 for 24 h was observed in MC3T3-E1 cells but not in ST2 cells, as shown in the bottom panels of Fig. 1A and B (lanes 9 to 12). The additive action of MK4 and E2 in MC3T3-E1 cells was more pronounced following a longer culture period (Fig. 1C and D). Our results suggest that MK4 and E2 act from the stage at which preosteoblasts differentiate into mature osteoblasts but not at the earlier stage of the cell fate decision. To confirm this, we measured the expression of osteoblast genesis marker genes in the MC3T3-E1 cells treated with MK4 and E2 by qRT-PCR. Early osteoblast genesis marker genes (encoding type I collagen α and ALP) were induced by both MK4 and E2 treatments, while mRNA levels of late osteoblast genesis marker genes (encoding osteopontin and osteocalcin) were unaltered following 24 h of treatment (Fig. (Fig.1E1E).
Msx2 is a target gene for MK4 and E2 in MC3T3-E1 cells.
To elucidate the molecular pathway through which MK4 and E2 induce osteoblast genesis, we screened for MK4 target proteins in untreated and treated MC3T3-E1 cells with 2D SDS-PAGE (Fig. 2A and B). E2 was used for cotreatment, since it potentiated MK4 action in osteoblastic differentiation. As shown in Fig. Fig.2B,2B, expression levels of several candidate proteins were up-regulated or down-regulated by the treatment, and we tried to identify them by matrix-assisted laser desorption ionization-time-of-flight mass spectroscopy analysis. One of these was Msx2, a protein that was shown to be a critical factor for inducing mesenchymal stem cells to differentiate into preosteoblasts (26). We reasoned that Msx2 mediates MK4 action as an MK4 target gene. Indeed, induction of Msx2 gene expression by MK4 alone was visible, and the additive induction with E2 was at both the transcriptional level (qRT-PCR) and the translational level (Western blotting) (Fig. 2C and D).
Msx2 promoter confers responsiveness to MK4 and E2 through cognate NRs.
To test the idea that MK4 acts as a PXR ligand in the transcriptional control of the mouse Msx2 gene, we performed a promoter analysis of the Msx2 gene by using a luciferase reporter assay on MC3T3-E1 cells. The responsiveness to MK4 and E2 was tested with a luciferase gene reporter driven by the mouse Msx2 gene promoter region (−1 to −3000) (designated Msx2-3000). Either MK4 or E2 was sufficient to activate transcription alone, and synergy of MK4 with E2 was observed (Fig. (Fig.3A).3A). Knockdown of PXR and ERα by RNAi (Fig. 3B and C) abolished the MK4- and E2-induced transactivation of the Msx2 gene promoter (Fig. (Fig.3A,3A, lanes 5 to 16). Other known PXR ligands also were tested with this promoter. Though MK4 was effective in stimulating the transcription of the Msx2 gene promoter, none of the other tested ligands (pregnenolone 16α-carbonitrile, 17α-hydroxypregnenolone, mifepristone, and dexamethazone 21-acetate) stimulated transcription alone. Additionally, none demonstrated synergy with E2 (Fig. (Fig.3D).3D). We used a series of promoter deletion mutants to map an MK4-responsive element in the proximal promoter (−1 to −390) (data not shown) that contained a consensus PXR/RXR binding-related element (designated Msx2-PXRE) at around bp −230 (Fig. (Fig.3E).3E). Additionally, a consensus element (designated Msx2-ERE) related to the ER-responsive element (ERE) was found in the distal promoter region (−2672 to −2658) (see Fig. Fig.3E3E and and4A).4A). Short elements containing either Msx2-PXRE or Msx2-ERE conferred the expected response to the cognate NR ligand (data not shown). To verify the impact of the response elements, we introduced two point mutations into each of the elements (Fig. (Fig.3E).3E). As shown in Fig. Fig.3F,3F, Msx2-M1, which was mutated at Msx2-ERE, lost the E2 response; likewise, the MK4 response was abolished by the mutation (Msx2-M2) in Msx2-PXRE. As expected, no NR ligand response was seen in the mutant (Msx2-M3) that was mutated at both Msx2-PXRE and Msx2-ERE (Fig. (Fig.3F3F).
Direct DNA binding of PXR on the Msx2 promoter.
To determine whether PXR bound to the proposed PXRE (Fig. (Fig.4A),4A), we examined the DNA binding ability of PXR and ERα on the Msx2 gene promoter with an ABCD assay. We tested whole-cell extracts of MC3T3-E1 cells transiently overexpressing PXR, RXRα, or ERα. PXR/RXRα bound to Msx2-PXRE, while ERα binding was not seen for Msx2-PXRE (Fig. (Fig.4B).4B). Conversely, ERα bound, as expected, to Msx2-ERE (Fig. 4A and B). Moreover, in this assay, p300 also associated with the receptors. To verify if the observed DNA binding of PXR and ERα in vitro reflected the physiological events in the Msx2 promoter, we performed a ChIP analysis on the endogenous promoter of the Msx2 gene. E2-induced recruitment of the ERα, but not PXR, was observed in the upstream region of Msx2-ERE. In contrast, the ligand-independent association of PXR took place in the region of the distal promoter containing Msx2-PXRE (Fig. (Fig.4C).4C). Consistent with the type of recruited receptors, the known NR coactivator p300 also was recruited to the promoter (Fig. (Fig.4C4C).
MK4-induced osteoblast genesis is mediated by PXR.
To confirm the physiological impact of PXR function on VK-induced osteoblast genesis, we knocked down either Msx2 or PXR in MC3T3-E1 cells by RNAi with adenovirus (Fig. 5A and B). Expression of endogenous Msx2 and PXR was significantly attenuated by RNAi (Fig. 5A and B). As expected, osteoblast genesis induced by MK4 was attenuated when cells were infected with Msx2 RNAi adenovirus (the middle panels of Fig. 5C and D [lanes 5 to 8]). Moreover, in primary osteoblasts derived from ERαKO mice, osteoblast genesis induced by E2 also was attenuated (Fig. (Fig.5E).5E). The induction of marker genes (encoding type I collagen α and ALP) was, as expected, also abrogated with Msx2 RNAi adenovirus (Fig. (Fig.5F,5F, lanes 5 to 8 and 17 to 20). Moreover, PXR RNAi also inhibited MK4-dependent osteoblast genesis, as shown in the bottom panels of Fig. 5C, D (lanes 9 to 12), and F (lanes 9 to 12 and 21 to 24).
DISCUSSION
This study demonstrates that Msx2 contains a PXRE in its gene promoter and that it is a direct target gene for MK4-bound PXR. Reflecting the similar effects of E2 and MK4, an ERE also was mapped in the Msx2 gene promoter. Although each of the response elements alone was not remarkable, the combined effects of MK4 and E2 were evident in transactivation. In osteoblastic differentiation induced by MK4, E2, or both, the knockdown of endogenous Msx2 abrogated the effects of the NR ligands. Thus, Msx2 likely mediates, at least in part, the effects of both MK4 and E2 on osteoblastic differentiation. Msx2 is one of the prime osteoblastogenic factors in an intact animal (26). Our observation that osteoblast genesis is induced by MK4 in cultured preosteoblastic cells suggests that MK4 exerts its osteoprotective action by upregulating Msx2 gene expression and increasing osteoblast genesis (Fig. (Fig.5G).5G). Since extracellular matrix Tsukushi genes also are VK target genes (8), the osteoprotective effects of VK in the human are mediated both by the classical pathway of protein γ-carboxylation and through genomic action via PXR/SXR.
PXR/SXR is activated by a number of endogenous and exogenous ligands (23). PXR/SXR plays a prominent role in the detoxification of chemicals (36). PXR/SXR is considered a global sensor for low-molecular-mass fat-soluble drugs that also is responsible for the timely degradation of these compounds (14). Thus, in contrast to steroid hormone receptors, PXR/SXR responds to a variety of ligands with unrelated structures. The three-dimensional structure of the PXR/SXR ligand binding domain (LBD) is known (35). The LBD is flexible, with a wider cave than those of steroid receptors (37). This permits PXR/SXR to capture a variety of ligands, including MK4. Shifting of the C-terminal transactivation helix H12 in the ERα LBD alters the conformation of the receptor, depending on whether an agonist or antagonist is bound. This plasticity enables ERα to respond to a variety of different ligands (27). It is possible that MK4 binding induces an H12 shift that is distinct from those induced by the other ligands, since the transactivation of PXR in the Msx2 promoters was induced by MK4 but not by the other known PXR/SXR ligands in the osteoblastic cell lines. Thus, the MK4-induced angle in the H12 shifting of PXR might be preferential to recruit coactivator/coactivator complexes to the Msx2 gene promoter in preosteoblasts. In fact, the MK4-induced association of PXR with the well-characterized coactivator (p300) was detectable in both an ABCD assay and a ChIP analysis of the Msx2 gene promoter. Our observations are a starting point for investigating coregulators (17, 24) associating with MK4-bound PXR in osteoblastic cells. The identification of PXR coregulators will enhance the understanding of the osteoprotective effects of VK at a molecular level.
Acknowledgments
We thank P. Chambon for kindly providing ERαKO mice; S. Takezawa, F. Ohtake, M. S. Kim, S. Fujiyama, Y. Mezaki, R. Fujiki, M. Kouzu-Fujita, T. Matsumoto, and Y. Imai for technical assistance; T. Yoshizawa and H. Kawashima (Niigata University) for the kind gift of Max2 expression vector; Eisai Co., Ltd., for the gift of MK4; T. Matsumoto, D. Inoue, and R. Okazaki for helpful discussions; and H. Higuchi and K. Hiraga for manuscript preparation.
This work was supported in part by a Grant-in-Aid for Basic Research Activities for Innovative Bioscience (BRAIN).
REFERENCES
Articles from Molecular and Cellular Biology are provided here courtesy of Taylor & Francis
Full text links
Read article at publisher's site: https://doi.org/10.1128/mcb.00813-07
Read article for free, from open access legal sources, via Unpaywall: https://europepmc.org/articles/pmc2169156?pdf=render
Citations & impact
Impact metrics
Citations of article over time
Alternative metrics
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1128/mcb.00813-07
Article citations
Current knowledge of bone-derived factor osteocalcin: its role in the management and treatment of diabetes mellitus, osteoporosis, osteopetrosis and inflammatory joint diseases.
J Mol Med (Berl), 102(4):435-452, 16 Feb 2024
Cited by: 0 articles | PMID: 38363329 | PMCID: PMC10963459
Review Free full text in Europe PMC
Vitamin K and Hallmarks of Ageing: Focus on Diet and Gut Microbiome.
Nutrients, 15(12):2727, 12 Jun 2023
Cited by: 0 articles | PMID: 37375631 | PMCID: PMC10301624
Review Free full text in Europe PMC
The role of pregnane X receptor (PXR) in substance metabolism.
Front Endocrinol (Lausanne), 13:959902, 16 Aug 2022
Cited by: 7 articles | PMID: 36111293 | PMCID: PMC9469194
Review Free full text in Europe PMC
The Interface between Cell Signaling Pathways and Pregnane X Receptor.
Cells, 10(11):3262, 22 Nov 2021
Cited by: 4 articles | PMID: 34831484 | PMCID: PMC8617909
Review Free full text in Europe PMC
High Pregnane X Receptor (PXR) Expression Is Correlated with Poor Prognosis in Invasive Breast Carcinoma.
Diagnostics (Basel), 11(11):1946, 20 Oct 2021
Cited by: 2 articles | PMID: 34829293 | PMCID: PMC8624096
Go to all (56) article citations
Data
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Augmentation of estrogen receptor-mediated transcription by steroid and xenobiotic receptor.
Endocrine, 33(3):305-316, 01 Jun 2008
Cited by: 6 articles | PMID: 19011999
Vitamin K2 regulation of bone homeostasis is mediated by the steroid and xenobiotic receptor SXR.
J Biol Chem, 278(45):43919-43927, 14 Aug 2003
Cited by: 212 articles | PMID: 12920130
The pregnane X receptor regulates gene expression in a ligand- and promoter-selective fashion.
Mol Endocrinol, 19(5):1170-1180, 13 Jan 2005
Cited by: 77 articles | PMID: 15650019
Steroid and xenobiotic receptor mediates a novel vitamin K2 signaling pathway in osteoblastic cells.
J Bone Miner Metab, 26(1):9-12, 10 Jan 2008
Cited by: 19 articles | PMID: 18095058
Review
[Vitamin K function mediated by activation of steroid and xenobiotic receptor].
Clin Calcium, 19(12):1770-1778, 01 Dec 2009
Cited by: 1 article | PMID: 19949268
Review