The authors state that they have no conflicts of interest.
Osteoblasts differentiate from mesodermal progenitors and play a pivotal role in bone formation and mineralization. Several transcription factors including runt-related transcription factor 2 (RUNX2), Osterix (OSX), and activating transcription factor4 (ATF4) are known to be crucial for the process, whereas the upstream signal transduction controlling the osteoblast differentiation sequence is largely unknown. Here, we explored the role of c-jun N-terminal kinase (JNK) in osteoblast differentiation using in vitro differentiation models of primary osteoblasts and MC3T3-E1 cells with ascorbic acid/β-glycerophosphate treatment. Terminal osteoblast differentiation, represented by matrix mineralization, was significantly inhibited by the inactivation of JNK with its specific inhibitor and exogenous overexpression of MKP-M (MAP kinase phosphatase isolated from macrophages), which preferentially inactivates JNK. Conversely, enhanced mineral deposition was observed by inducible overexpression of p54JNK2, whereas it was not observed by the overexpression of p46JNK1 or p46JNK2, indicating a distinct enhancing role of p54JNK2 in osteoblast differentiation. Inactivation of JNK significantly inhibited late-stage molecular events of osteoblast differentiation, including gene expression of osteocalcin (Ocn) and bone sialoprotein (Bsp). In contrast, earlier differentiation events including alkaline phosphatase (ALP) activation and osteopontin (Opn) expression were not inhibited by JNK inactivation. Although the expression levels of two transcription factor genes, Runx2 and Osx, were not significantly affected by JNK inactivation, induction of Atf4 mRNA during osteoblast differentiation was significantly inhibited. Taken together, these data indicate that JNK activity is specifically required for the late-stage differentiation events of osteoblasts.
Osteoblasts, as well as osteocytes, which are a terminally differentiated form of osteoblasts, are responsible for bone formation by producing bone matrix proteins, which subsequently induce tissue mineralization. Osteoblasts differentiate and mature from their progenitors in response to various regulatory factors including bone morphogenetic proteins (BMPs), IGF-1, fibroblast growth factor 2 (FGF-2), PTH, TNF-α, Wnts, and extracellular matrix signals. Matrix mineral deposition occurs at the terminal stage of osteoblast differentiation and is associated with maximal expression of osteocalcin (Ocn). When primary osteoblasts or clonal osteoblastic cell lines, such as MC3T3-E1, are cultured with ascorbic acid (AA) and β-glycerophosphate (βGP) in vitro, the cells undergo differentiation and eventually produce calcified nodules resembling woven bone. This in vitro model has been successfully used in various studies to explore the molecular mechanisms of osteoblast differentiation.
Although several transcription factors, such as RUNX2 (runt-related transcription factor 2), OSX (Osterix), and ATF (activating transcription factor) 4, have been shown to be crucial to bone development, the molecular details of intracellular signal transduction controlling stage-specific osteoblast differentiation remain enigmatic. It is well recognized that the mitogen activated protein kinase (MAPK) superfamily, including p44/p42 extracellular signal-regulated kinases (ERKs), p38 MAPKs, and p54/p46 c-jun N-terminal kinases (JNKs), integrates signals from a diverse range of extracellular stimuli, and plays important roles in cellular functions such as proliferation, differentiation, and cell death in a variety of cell types. Activation of MAPKs requires dual phosphorylation of Tyr and Thr residues in the activation loop of the molecules. Phosphorylation of the two residues is catalyzed by a family of dual specificity kinases termed MAPK kinases (MKKs), whereas their dephosphorylation is catalyzed by dual specificity phosphatases termed MAPK phosphatases (MKPs). Among MKP family members, some show highly selective substrate specificity, whereas others efficiently inactivate all three classes of MAPKs. We previously cloned an MKP termed MKP-M (also referred to as DUSP16 or MKP-7), which preferentially inactivates JNKs., Despite intensive study, the physiological roles of MAPKs in osteogenic differentiation remain mostly unclear. Previous studies on the role of ERK in osteoblast differentiation resulted in different conclusions., The role of p38 MAPK in osteoblast differentiation is also disputable. For example, kinase activity of p38 MAPK was reported to be stimulatory or inhibitory in BMP-2–induced osteoblast differentiation.
JNK was originally identified by its ability to specifically phosphorylate the transcription factor c-jun in its N-terminal transactivation domain. JNK consists of three isoforms (JNK1, 2, and 3) deriving from distinct genes. Among them, JNK1 and JNK2 are ubiquitously expressed, whereas JNK3 is mainly expressed in brain, testis, and heart. Alternative splicing at the C terminus yields 46- and 54-kDa polypeptides for each JNK isoform, but the biological significance of the shorter and longer JNK polypeptides remains largely unclear. The 46-kDa form is predominant for JNK1, whereas the 54-kDa form is predominant for JNK2 and JNK3. Additionally, alternative exon use in the middle region yields α and β types for JNK1 and JNK2. JNK1 and JNK2 possess structural similarities and many overlapping biological functions. However, recent evidence has shown some functional differences between the two kinases. For example, JNK1, but not JNK2, promotes c-jun–dependent fibroblast proliferation. It has also been reported that JNK1 is distinctly involved in TNF-α–induced apoptosis. However, possible functional differences among JNK isoforms in osteogenic differentiation remain mostly unknown.
In this study, we explored the role of JNK in osteoblasts and found that the terminal osteoblastic differentiation was significantly inhibited by JNK inactivation. Conversely, enhanced matrix mineralization was observed by inducible overexpression of p54JNK2 in an isoform-specific manner. More specifically, JNK inhibition significantly suppressed late stage molecular events of osteoblastic differentiation, such as gene expression of Ocn and bone sialoprotein (Bsp). In contrast, earlier differentiation events including alkaline phosphatase (ALP) activation and osteopontin (Opn) expression were not inhibited by JNK inactivation. As for transcriptional factors, induction of Atf4 expression during osteoblastic differentiation was significantly inhibited, whereas expression levels of Runx2 and Osx were not significantly affected by JNK inactivation. Thus, these data indicate that JNK activity is essential for the late-stage differentiation of osteoblasts.
MATERIALS AND METHODS
Reagents and antibodies
SP600125, a specific JNK inhibitor, and U0126, a specific inhibitor of ERK activation pathway, were purchased from BIOMOL International (Plymouth Meeting, PA, USA) and CALBIOCHEM (San Diego, CA, USA), respectively. Recombinant human BMP-2 was obtained from PeproTech (Rocky Hill, NJ, USA). Antibodies specifically recognizing phosphorylated forms of JNKs, ERKs, p38 kinases, and Smad1/5 were purchased from Cell Signaling Technology (Danvers, MA, USA). Antibodies against ERK1/2, JNK1/2, JNK2, and p38 kinases, were also from Cell Signaling Technology. A specific antibody against JNK1 was obtained from R&D Systems (Minneapolis, MN, USA). Antibodies against β-actin and GAPDH were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The anti-Flag M2 antibody was from Sigma-Aldrich (St Louis, MO, USA). The anti- MKP-M polyclonal antibody was previously described.
MC3T3-E1 cells were obtained from RIKEN Cell Bank (Tsukuba, Japan) and maintained in Eagle's αMEM (Sigma-Aldrich) containing 10% FBS, 50 units/ml penicillin, and 50 mg/ml streptomycin. Primary osteoblasts were isolated from newborn mouse calvariae. Briefly, calvariae of newborn C57BL/6 mice were excised under aseptic condition, rinsed twice with ice-cold αMEM, and incubated in enzyme solution (αMEM containing 0.25% collagenase I and 0.125% trypsin) with agitation. After consecutive enzyme treatments (6 × 20 min), the fourth, fifth, and sixth supernatants were centrifuged. The pellets were resuspended in αMEM containing 10% FBS for cell culture. Differentiation of MC3T3-E1 and primary osteoblasts were induced by the addition of 50 μg/ml AA and 5 mM β-GP in the culture medium.
Alizarin red S staining
Culture dishes were washed with Ca2+-free PBS for three times and fixed in 1% formaldehyde/PBS for 20 min at 4°C. After five washes with distilled water, the cells were stained in 1% alizarin red S solution for 5 min to visualize matrix calcium deposition. The remaining dye was washed out by several washes with distilled water, and the stained cells were photographed.
Cell lysate containing 2 μg total protein was added to the assay buffer (1 mM MgCl2, 50 mM Tris-HCl/pH 9.2) containing 2 mM p-nitrophenol phosphate. After 10-min incubation at 37°C, the reaction was terminated by the addition of 0.45 M NaOH. The absorbance of p-nitrophenol liberated in the reactive solution was read at 420 nm. The relative ALP activity was defined as nanomoles of p-nitrophenol phosphate hydrolyzed per minute per 1 μg of total protein.
Northern and Western blot analyses
Total cellular RNA was extracted using TRIZOL reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Fifteen-microgram aliquots of total RNA were fractionated, transferred, and hybridized with an indicated [32P]labeled cDNA probe as previously described. The β-actin cDNA probe was previously described. For Runx2, Osx, Ocn, Opn, Bsp, and Atf4 mRNA gene expression, cDNA probes were prepared by RT-PCR from the total RNA isolated from MC3T3-E1 cells.
For Western blot analyses, total cellular lysate preparation and immunoblotting procedures were performed as previously described.
Tet-on inducible expression system
A Tet-on regulator plasmid, pEF-1α-Tet-on, was prepared by replacing the CMV promoter of a regulator plasmid, pTet-on (Clontech Laboratories, Mountain View, CA, USA), for the human elongation factor 1α (EF-1α) promoter. The pEF-1α-Tet-on was stably transfected into MC3T3-E1 cells by Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions, followed by selection with 0.5 mg/ml G418. The cDNAs of mouse JNK1α, JNK2α1, and JNK2α2 were amplified by RT-PCR from the total RNA isolated from MC3T3-E1 cells and cloned into pTRE2-Hyg vector (Clontech Laboratories), which contains the hygromycin-resistance gene. The cDNA of a C-terminally truncated form of mouse MKP-M (MKP-M A1T) with Flag epitope tag was amplified by PCR from the regular expression vector of Flag-tagged MKP-M A1T and cloned into pTRE2-Hyg vector. After verification by restriction mapping and sequencing, the constructed inducible expression plasmids, which were termed as pTRE2-Hyg-JNK1α1, pTRE2-Hyg-JNK2α1, pTRE2-Hyg-JNK2α2, and pTRE2-Hyg-Flag-MKP-M, were stably transfected into the MC3T3-E1 pEF-1α-Tet-on cell line by Lipofectamine 2000. As a negative control cell line, pTRE2-Hyg vector was stably transfected into the MC3T3-E1 pEF1α-Tet-on cell line. After selection with 0.15 mg/ml hygromycin B, the isolated resistant clones were tested for the inducible protein expression from the inserted cDNAs with 2 μg/ml doxycycline (DOX). Individual three to five cell lines with good inducible protein expression were analyzed for each construct.
RNA interference of Jnk2
MC3T3-E1 cells were transfected with small interfering RNA (siRNA) duplexes specific for murine Jnk2: r(GAUCAUUUCAGGUGAGCAA)dTdT and r(UUGCUCACCUGAAAUGAUC)dTdT obtained from Sigma-Aldrich or nontargeting control siRNA duplexes (Control siRNA-A; Santa Cruz Biotechnology Inc.) using Lipofectamine RNAiMAX (Invitrogen Life Technologies) according to the manufacturer's instructions.
Atf4 promoter assay
The 5′-upstream region of the mouse Atf4 gene containing a 1448-bp DNA sequence upstream of the putative transcriptional initiation site was cloned by PCR from C57BL/6 genomic DNA using two primers, GACTCGAGGCTAGGATAATTGGCTGTTAGA and GAAAGATCTGAATTTCCGGTGTCTTACAAG, and ligated into the luciferase reporter vector pGL4.19 (Promega, Madison, WI, USA). An 8-μg aliquot of the resultant plasmid (pGL4-Atf4) or the vector control (pGL4.19) was transfected into 70% confluent MC3T3-E1 cells in a 60-mm round culture plate along with 0.8 μg transfection control plasmid, pRL-TK, using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. At 48 h after transfection, relative luciferase activity was measured by Dual-Luciferase reporter assay kit (Promega) according to the manufacturer's instructions.
Matrix mineralization in osteoblast differentiation was inhibited by a JNK inhibitor
To explore the functional roles of JNK in osteogenic differentiation, we induced differentiation of MC3T3-E1 cells with AA/βGP in the absence or presence of various concentrations of SP600125, a specific JNK inhibitor. We first tested the specific inhibitory effects of SP600125 on JNKs in MC3T3-E1 cells (Fig. 1A). UV-induced p46/p54 JNK phosphorylation in MC3T3-E1 cells was inhibited by SP600125 at 2 μM or higher in a dose-dependent manner. In contrast, phosphorylation of p38 kinase, which was also induced by UV, was not inhibited by SP600125 at as much as 20 μM. Furthermore, SP600125 did not affect p42/p44 ERK phosphorylation induced by BMP-2 (data not shown). Thus, the inhibitory effect of SP600125 was specific to JNK among three MAPKs in MC3T3-E1 cells. When MC3T3-E1 cells were induced to differentiate by a combination of AA/βGP for 21 days, matrix mineralization visualized by alizarin red S staining was significantly inhibited by SP600125 in a dose-dependent manner (Fig. 1B).
We next examined the effect of U0126, a specific inhibitor of ERK activation pathway, on MC3T3-E1 differentiation. BMP-2–induced ERK phosphorylation in MC3T3-E1 cells was inhibited by U0126 at 2.5 μM or higher in a dose-dependent manner (Fig. 1C). In contrast, phosphorylation of Smad1/5, which is a direct downstream target of BMP-2, was not inhibited by U0126 at as much as 20 μM. Furthermore, UV-induced phosphorylation of JNKs and p38 kinase was not affected by 20 μM U0126 (data not shown), indicating the inhibitory effect of U0126 was specific to the ERK activation pathway in MC3T3-E1 cells. When MC3T3-E1 cells were induced to differentiate, matrix mineralization was significantly promoted by U0126 at 2.5 μM or higher (Fig. 1D).
We also analyzed the effects of SP600125 and U0126 on AA/βGP-induced matrix mineralization of primary osteoblasts prepared from newborn mouse calvariae. Similar to the results of MC3T3-E1 cells, SP600125 treatment significantly delayed the matrix mineralization by primary osteoblasts in a dose-dependent manner (Fig. 1E). In contrast, unlike MC3T3-E1 cells, primary osteoblasts showed significantly delayed matrix mineralization in the presence of U0126 (Fig. 1E). Neither cell density nor viability of MC3T3-E1 and primary osteoblasts was affected by SP600125 or U0126 at concentrations up to 10 μM (data not shown).
Matrix mineralization by MC3T3-E1 cells was inhibited by the expression of a JNK-specific phosphatase
We prepared MC3T3-E1 cell line, which is induced to express a truncated and stable form of MKP-M (MKP-M A1T), a JNK-specific phosphatase, on stimulation by DOX, using a Tet-on inducible expression system (Fig. 2A). In this inducible cell line, UV-treated JNK phosphorylation is strongly inhibited, whereas p38 phosphorylation is intact, in the presence of 2 μg/ml DOX (Fig. 2B), indicating the inhibitory effect of the exogenous MKP-M was specific to JNK. Neither cell density nor viability was affected by DOX treatment in either the MKP-M-transfected or vector control cell line (data not shown). When cells were induced to differentiate, matrix mineralization was delayed by the addition of DOX in the MKP-M A1T-transfected cell line but not in the vector control cell line (Fig. 2C). This result, in combination with the above result using a specific JNK inhibitor, strongly suggested that JNK activity plays an important role in the differentiation process of osteoblasts.
JNK inhibition specifically suppressed the late-stage molecular events of osteoblast differentiation
Osteoblast differentiation is characterized by the appearance of a series of mRNAs preferably or specifically expressed in the osteoblast/osteocyte cell lineage. To examine the role of JNK in the expression of these mRNA markers, we performed Northern blot analyses on MC3T3-E1 cells induced to differentiate in the presence or absence of SP600125. Induction of Ocn and Bsp mRNAs, which represents late-stage osteoblast differentiation, was significantly inhibited by the JNK inhibitor (Fig. 3A). In contrast, the amount of Opn mRNA, which is expressed in the earlier stage osteoblast differentiation, was not affected (Fig. 3A). We also examined the effects of U0126, an ERK pathway inhibitor, on the three mRNA markers of osteoblast differentiation. It was found that the addition of U0126 moderately promoted the onset of Opn mRNA induction, whereas it did not significantly change the course of Ocn or Bsp mRNA induction (Fig. 3A).
We performed similar analyses on primary osteoblasts derived from newborn mouse calvariae. Being similar to MC3T3-E1 cells, induction of Ocn and Bsp mRNAs was significantly inhibited by the JNK inhibitor (Fig. 3B). In contrast, the level of Opn mRNA was significantly elevated by the JNK inhibitor on days 18 and 24 of differentiation (Fig. 3B).
The catalytic activity of ALP is known to be increased in osteoblasts in the early phase of differentiation. We thus measured ALP activity of MC3T3-E1 and primary osteoblasts during differentiation with or without the presence of SP600125. We found that the increase of ALP activity during osteoblast differentiation was not inhibited by the JNK inhibitor in either MC3T3-E1 cells or primary osteoblasts (Fig. 3C).
Expression of Atf4 was inhibited by JNK inhibition, whereas expression of Runx2 or Osx was not affected
The osteoblast differentiation sequence is controlled by several transcription factors including RUNX2, OSX, and ATF4. We thus analyzed the effects of JNK inhibition on the gene expression of these transcription factors. MC3T3-E1 cells used in our assays express Runx2 mRNA without any treatment, which remained constant during AA/βGP-induced differentiation (Fig. 4A). Expression levels of Runx2 were not affected by the presence of SP600125 (Fig. 4A). Similarly, the induction time course of Osx mRNA remained unaltered with SP600125 treatment (Fig. 4A). In contrast, induction levels of Atf4 mRNA were significantly inhibited by JNK inhibition (Fig. 4A).
We also performed similar analyses on primary osteoblasts derived from newborn mouse calvaria. The transcriptional induction of Runx2 and Osx was not affected by SP600125 (Fig. 4B). In contrast, being similar to MC3T3-E1 cells, induction Atf4 mRNA was significantly inhibited by the JNK inhibitor in primary osteoblasts (Fig. 4B).
We further tested the role of JNK in transcriptional regulation of Atf4 in osteoblasts. Atf4 mRNA expression was immediately induced by anisomycin, a potent JNK activator, in MC3T3-E1 cells (Fig. 4C). Atf4 was also induced by UV, which potently activates JNKs (data not shown). We cloned a mouse Atf4 gene promoter and performed a luciferase reporter assay. The promoter activity of the reporter plasmid containing 1448 bp upstream DNA of the putative transcriptional initiation site (pGL4-Atf4) was significantly inhibited by JNK inhibition (Fig. 4D).
Generation of MC3T3-E1 cell lines inducibly expressing JNK isoforms
Two types of JNKs, JNK1 and JNK2, are ubiquitously expressed in various cell types including osteoblasts. Both JNK1 and JNK2 consist of four isoforms caused by differential splicing and exon use. To determine which JNK isoform is predominantly expressed in osteoblasts, we prepared mRNA from MC3T3-E1cells and performed RT-PCR using four distinct pairs of primers specific for JNK1-1 (p46JNK1), JNK1-2 (p54JNK1), JNK2-1 (p46JNK2), and JNK2-2 (p54JNK2) (Fig. 5A). We failed to obtain detectable RT-PCR product of JNK1-2 cDNA, indicating the mRNA level of JNK1-2 is very low in MC3T3-E1 cells. As for JNK1-1, JNK2-1, and JNK2-2, we sequenced 10 different cDNA clones, amplified by RT-PCR, for each isoform. All the 30 cDNA clones showed the α type sequences, indicating that the α type alternative exon use is predominant for both JNK1 and JNK2 in MC3T3-E1 cells.
To identify the isoform-specific roles of JNKs, we generated MC3T3-E1 cell lines, each of which is induced to overexpress JNK1α1, JNK2α1, or JNK2α2 using the Tet-on inducible expression system. The induced protein expression was detected by Western blotting using anti-JNK1 (for JNK1α1) and anti-JNK2 (for JNK2α2) antibodies. Because the anti-JNK2 antibody recognizes the C-terminal region of JNK2α2, not detecting JNK2α1, the induced protein expression of JNK2α1 was confirmed by the increased intensity of the p46 band using a pan anti-JNK1/2 antibody, which detects JNK2 better than JNK1. In the presence of DOX (2 μg/ml), each JNK isoform was induced within 24 h and was at least stable for 4 days (Fig. 5B). Furthermore, longer DOX treatment experiments showed that the induced exogenous JNK protein expression levels remained constantly high for as long as 24 days in the JNK1α1, JNK2α2 (Fig. 5C), and JNK2α1 (data not shown) cell lines.
p54JNK2 distinctly promotes osteoblastic differentiation of MC3T3-E1 cells
We induced osteogenic differentiation of three independent MC3T3-E1cell clones for each JNK isoform (JNK1α1, JNK2α2, or JNK2α1) by AA/βGP with or without DOX. Typical results are shown in Fig. 6A. Terminal osteoblastic differentiation evidenced by matrix mineralization was significantly promoted by the DOX-induced overexpression of JNK2α2 (Fig. 6A). In contrast, osteogenic differentiation was not affected by the DOX-induced overexpression of either JNK1α1 or JNK2α1 (Fig. 6A).
We subsequently isolated RNA from MC3T3-E1 Tet-on JNK2α2 cell line under differentiation with or without DOX and tested for the expression of Runx2, Osx, Atf4, and Ocn (Fig. 6B). DOX-induced overexpression of JNK2α2 promoted Ocn expression, which was consistent with the enhanced matrix mineralization in the presence of DOX (Fig. 6A). As for transcription factors, neither Runx2 nor Osx expression was affected, whereas Atf4 expression was significantly promoted by DOX treatment (Fig. 6B).
Endogenous protein expression profiles of JNK isoforms during osteoblastic differentiation of MC3T3-E1 cells
We analyzed endogenous protein expression levels of JNK isoforms during osteoblastic differentiation of MC3T3-E1 cells. Total cell lysates were isolated from MC3T3-E1 cells differentiating in the presence of AA/βGP and analyzed by anti-JNK1 (recognizing p46JNK1 and p54JNK1), anti-JNK2 antibody (recognizing p54JNK2), or pan anti-JNK1/2 antibody (recognizing p46JNK2 and p54JNK2 better than p46JNK1and p54JNK1) (Fig. 7A). As differentiation progressed, the protein expression level of p46JNK1 remained constant (Fig. 7B). A low level protein expression of p54JNK1 was detected, which slightly increased during differentiation (Fig. 7B). In contrast, protein expression level of p54JNK2 was significantly increased on days 18 and 24 (Figs. 7A and 7C). Comparing the JNK1/2 and JNK1 blots, we presumed the expression level of p46JNK2 decreased in the late stage of differentiation (Figs. 7A and 7B). Finally, using a phospho-specific antibody, JNK phosphorylation was detected in the late stage of differentiation mainly on the p54 form of JNK (Fig. 7D).
We analyzed the effect of Jnk2 knockdown on osteoblast differentiation. We confirmed that the JNK2 siRNA effectively inhibited JNK2 protein expression at least for 12 days, whereas it did not affect JNK1 expression in MC3T3-E1 cells (Fig. 7B). The JNK2 siRNA was transfected on day 9 of AA/βGP treatment to inhibit the JNK2 expression in the late differentiation period. As a result, the JNK2 siRNA effectively inhibited matrix mineralization on day 21 (Fig. 7C).
In this study, using the osteoblastic cell line, MC3T3-E1, and primary osteoblasts, we showed that the matrix mineralization by osteoblasts was significantly downregulated by JNK inhibition in two independent experimental systems using a JNK-specific inhibitor (Fig. 1) and the overexpression of MKP-M, a JNK-specific phosphatase (Fig. 2). These results strongly suggest a critical role of JNK in the process of osteoblast differentiation. Bone matrix maturation is largely regulated by the gene expression of a series of bone matrix proteins in osteoblasts, which strictly depends on osteoblast differentiation stages. In this study, specific inhibition of JNK significantly inhibited the gene expression of Ocn and Bsp, whereas it did not inhibit the earlier differentiation events such as increases in ALP activity and gene expression of Opn (Fig. 3). Thus, JNK activity is essential for the progression through the late stage of osteoblast differentiation, whereas JNK activity is sufficient in the early stage of osteoblast differentiation. Consistently, JNK phosphorylation was detectable only in the late osteoblastic differentiation stage of MC3T3-E1 cells (Fig. 7).
Bone tissue-specific OCN remains one of the few proteins that exhibits strictly osteoblast-restricted expression. OCN is induced with the onset of matrix mineralization at the terminal stage of osteoblast differentiation. In mice, there are two OCN-encoding genes: OG1 and OG2. They encode the same protein and display the same expression pattern, and their proximal promoter regions are 93% identical over the first 1.3 kb. Molecular regulation of OG2 gene transcription has been extensively studied. The OG2 gene promoter is modularly organized and exhibits both positive and negative regulatory elements., The essential 647-bp fragment of the OG2 promoter contains two kinds of osteoblast-specific cis-acting elements termed OSE1 and OSE2. OSE2 has been shown to be the binding site for a transcription factor, RUNX2. Two OSE2s, OSE2a and OSE2b, which share the core sequence of AACCACA, are located at −131/−137 and −602/−608 in the OG2 promoter, respectively. Both OSE2a and OSE2b are conserved in the other Ocn gene, OG1, indicating they are functionally essential. OSE1, which is located at −131/−137 of the OG2 promoter, was previously shown to be of an equal importance to OSE2 for the stimulation of the Ocn gene expression. DNA-protein interaction analysis showed that OSE1 is a binding site for another transcription factor, ATF4. Consistently, Ocn gene expression is severely decreased in the skeletons of Atf4-deficient mice, indicating that ATF4 is a positive transcriptional regulator of Ocn.
BSP is a bone matrix glycoprotein and works as a nucleator of hydroxyapatite crystals through its glutamic acid-rich clusters. BSP is considered as a mid- to late osteoblast differentiation marker that is upregulated at the onset of bone matrix mineralization. Two RUNX2 binding sites (R1 and R2) are located at −83/−89 and −184/−190 in the mouse Bsp gene promoter, and similar sequences are found in the human and rat Bsp promoters. Site-specific mutagenesis showed that both sites act as osteoblast-specific transcriptional enhancers and together account for nearly two thirds of the total promoter activity. ATF4 is presumed to be another transcriptional regulator of Bsp, because Bsp gene expression is severely decreased in the skeletons of Atf4-deficient mice. Although the direct involvement of Atf4 in the Bsp promoter has not been documented, a functional cAMP response element (CRE) is located at −68/−75 of the mouse, rat, and human Bsp promoter and may act as an ATF4 binding site.
Thus, RUNX2 is essentially involved in the transcription of both Ocn and Bsp. In addition to these two genes, RUNX2 upregulates the transcription of many other bone matrix proteins including collagen Ia, OPN, fibronectin, matrix metallopeptidase 13 (MMP13), and osteoprotegerin, as a master regulator of osteoblast differentiation, and mice deficient in Runx2 lack osteoblasts., However, for the following reasons, we do not presume that the promoting effects of JNK on osteogenic differentiation are through direct induction or activation of RUNX2. First, specific inhibition of JNK activity by SP600125 did not affect Runx2 mRNA expression during differentiation of either MC3T3-E1 or primary osteoblasts. Second, gene expression of Opn, which is also under the control of RUNX2 activity, was not downregulated by JNK inhibition.
OSX, a zinc finger-containing transcription factor, is also required for osteoblast differentiation. Osx-deficient mice lack osteoblasts, similar to the phenotypes in Runx2 null mice. These data showed that JNK inhibition did not alter Osx mRNA expression during osteogenic differentiation of either MC3T3-E1 or primary osteoblasts, indicating it is not likely that JNK regulates osteoblast differentiation through OSX. Consistent with this finding, it has recently been reported that forced expression of Osx in mesenchymal stem cells (MSCs) stimulated the expression of Opn and Alp, but not Ocn, suggesting that OSX does not directly mediate terminal differentiation of osteoblasts.
ATF4, also known as CREB2 (cAMP-response element-binding protein 2), is a basic leucine-zipper transcriptional factor that belongs to ATF/CREB protein family. Similar to RUNX2 and OSX, an essential role of ATF4 in bone development in vivo was shown by a study using Atf4-deficient mice, which showed a marked reduction in bone mineralization and trabecular development throughout life. The promoting effects of ATF4 on bone development is not through the modulation of RUNX2 or OSX expression level, because Atf4-deficient mice showed normal levels of these two transcriptional factors. In contrast to Runx2 and Osx, the increase of Atf4 mRNA during osteogenic differentiation was significantly inhibited by JNK inhibition in both MC3T3-E1 and primary osteoblasts (Fig. 4), indicating that JNK activity may promote osteoblast differentiation by increasing the amount of ATF4. The reported phenotypes of Atf4-deficient mice are consistent with this hypothesis, because the expression of OCN and BSP was markedly reduced in Atf4-deficient osteoblasts. More notably, Atf4-deficient osteoblasts showed normal expression of genes characteristic of earlier stages of differentiation including Runx2 and Osx, showing that ATF4 is specifically required for the late-stage osteoblast differentiation.
The molecular involvement of ATF4 in the Ocn promoter activation has recently been shown in detail. Atf4 activates Ocn expression only in the presence of RUNX2, indicating that cooperative interaction between RUNX2 and ATF4 is essential. A more recent report showed that general transcription factor IIAα (TFIIAα) interacts with both RUNX2 and ATF4 on the Ocn promoter and activates ATF4. This molecular model is consistent with our current data and may explain why JNK-dependent Atf4 expression is essential for the Ocn expression during osteoblast differentiation.
These data showed that Atf4 expression in osteoblast differentiation is at least partly dependent on JNK activity. Transcriptional regulation of Atf4 has not been extensively studied. Atf4 expression was markedly reduced in cells occupying skeletal elements of E16 Runx2-deficient mice, indicating that Atf4 expression in osteoblast differentiation is dependent on the presence of RUNX2. However, factors other than RUNX2 are also likely to be involved in the regulation of Atf4 expression. A recent report showed that stimulation of osteoblasts by PTH induced Atf4 mRNA in a protein kinase A-dependent manner.Atf4 expression is detected in cell types other than osteoblasts. For example, treatment of breast cancer cells with heregulin, a ligand for epidermal growth factor receptor 3/4, induced Atf4 mRNA. Noticeably, the heregulin-induced Atf4 expression was significantly inhibited by dominant negative mutants of MEKK, an upstream kinase of JNK and p38 activation pathways. Transcriptional upregulation of Atf4 by anisomycin (Fig. 4C) and the promoter assay result (Fig. 4D) in this study support the idea that JNK is directly involved in the Atf4 gene promoter activity in osteoblasts. According to the nucleotide sequence, the 5′ upstream region of the mouse Atf4 gene contains putative AP-1 binding sites (−1088/−1081 and −658/−651) that conform well to the consensus sequence (unpublished data). Because AP-1 is a well-established target of JNK signals, it is reasonable to presume that these putative AP-1 binding sites may be functional in osteoblast differentiation.
Osteoblasts, like most other cell types, express both JNK1 and JNK2. Despite structural similarities and many overlapping biological functions, recent evidence has shown functional differences between JNK1 and JNK2 in some cell types. Some of the functional differences are presumably caused by the different substrate-binding affinities of JNK1 and JNK2. For example, JNK1 phosphorylates insulin receptor substrate-1 (IRS-1) to be inactivated more preferentially than JNK2 does,, which explains why JNK1 is an important mediator of insulin resistance associated with obesity., On the other hand, JNK2 efficiently phosphorylates tau, a microtubule-associated protein, which is correlated with its ability to inhibit microtubule assembly. Furthermore, an experiment using jnk2−/− cells indicated a role for JNK2, rather than JNK1, in the inactivation of TIF-IA, an RNA polymerase I-specific transcription factor and the subsequent rRNA downregulation. It should be noted, however, that some of the functional differences in these studies may derive from differences between p46 and p54, rather than JNK1 and JNK2, because p46JNK1 and p54JNK2 are the two predominant forms.
Among the three JNK isoforms expressed well in MC3T3-E1 cells, inducible expression of JNK2α2 (p54JNK2) significantly promoted osteoblastic differentiation, whereas that of JNK1α1 (p46JNK1) or JNK2α1 (p46JNK2) was ineffective (Fig. 6). These results indicate that the C-terminal extension region of p54JNK2 may be essential for the differentiation-promoting activity of JNK2. The molecular significance of the C-terminal extension region of JNK is currently unclear. However, because the C-terminal extension region is conserved in JNK1/2/3, it is reasonable to presume that it is important by influencing JNK functions such as substrate specificity or enzyme activity. Interestingly, the p54/p46 ratio of JNK2 was increased during osteoblastic differentiation of MC3T3-E1 cells (Fig. 7A), indicating that the increased cellular content of p54JNK2 may contribute to the differentiation of this cell line. Consistently, matrix mineralization of MC3T3-E1 cells was inhibited by siRNA-mediated Jnk2 knockdown (Figs. 7B and 7C). We currently do not know the molecular mechanisms for the change of p54/p46 ratio. It may be caused by the biased alternative splicing of Jnk2 mRNA. It is also possible that the protein or mRNA stability of JNK2 isoforms varies during osteoblastic differentiation. Further study is obviously needed to elucidate these possibilities.
We additionally observed that matrix mineralization by MCT3-E1 cells was enhanced by U0126, a MEK1/2 inhibitor, which suggests that ERK pathway activation negatively regulates osteoblastic differentiation of this cell line (Fig. 1D). However, in contrast to MC3T3-E1 cells, U0126 clearly inhibited matrix mineralization of primary osteoblasts from mouse calvaria (Fig. 1E), indicating the regulatory role of ERKs in osteogenic differentiation may be cell line specific. The role of ERKs in osteoblastic differentiation has previously been studied with inconsistent results. In recent reports, matrix mineralization of MC3T3-E1 cells and pre-osteocytic MLO-A5 cells was increased by an MEK1/2 inhibitor, whereas selective expression of constitutively active ERK1 by a transgenic approach accelerated in vitro differentiation of calvaria cells, as well as in vivo bone development. Although the mechanisms for the discrepancy between these results including ours are currently unknown and need further study, these data support the hypothesis that the role of ERK signaling in osteoblast differentiation varies depending on the cell types or differentiation stages of osteoblasts.
In summary, this study showed a distinct role of JNK in late stage of osteoblastic differentiation presumably through the induction of Atf4 expression. Because the proper control of osteoblastic differentiation is essential for the maintenance of bone mass through life, this study may provide insights into a new list of therapeutic target molecules of bone metabolic disorders.
This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant 19390474).