Phosphophoryn Regulates the Gene Expression and Differentiation of NIH3T3, MC3T3-E1, and Human Mesenchymal Stem Cells via the Integrin/MAPK Signaling Pathway*

Extracellular matrix proteins (ECMs) serve as both a structural support for cells and a dynamic biochemical network that directs cellular activities. ECM proteins such as those of the SIBLING family (small integrin-binding ligand glycoprotein) could possess inherent growth factor activity. In this study, we demonstrate that exon 5 of dentin matrix protein 3 (phosphophoryn (PP)), a non-collagenous dentin ECM protein and SIBLING protein family member, up-regulates osteoblast marker genes in primary human adult mesenchymal stem cells (hMSCs), a mouse osteoblastic cell line (MC3T3-E1), and a mouse fibroblastic cell line (NIH3T3). Quantitative real-time PCR technology was used to quantify gene expression levels of bone markers such as Runx2, Osx (Osterix), bone/liver/kidney Alp (alkaline phosphatase), Ocn (osteocalcin), and Bsp (bone sialoprotein) in response to recombinant PP and stably transfected PP. PP up-regulated Runx2, Osx, and Ocn gene expression. PP increased OCN protein production in hMSCs and MC3T3-E1. ALP activity and calcium deposition was increased by PP in hMSC. Furthermore, an αvβ3 integrin-blocking antibody significantly inhibited recombinant PP-induced expression of Runx2 in hMSCs, suggesting that signaling by PP is mediated through the integrin pathway. PP was also shown to activate p38, ERK1/2, and JNK, three components of the MAPK pathway. These data demonstrate a novel signaling function for PP in cell differentiation beyond the hypothesized role of PP in biomineralization.

The tissue microenvironment consists of soluble and immobilized growth factors, ECM 1 molecules, and cells that orches-trate the tissue-specific cell growth, differentiation, and survival that are crucial for tissue development, morphogenesis, and remodeling (1)(2)(3)(4). A particular group of ECM proteins, the SIBLING protein family consisting of bone sialoprotein (BSP), osteopontin (OPN), dentin matrix protein 1 (DMP-1), and dentin matrix protein 3 (DMP-3), also called dentin sialophosphoprotein (DSPP) (5), may have important signaling roles. Although not extremely similar in protein sequence, members of the SIBLING family share common traits such as RGD integrin-binding domains and casein kinase II phosphorylation sites (5). As it was shown more extensively in systems other than bone or dentin, the ECM plays an important role in growth factor function (1,4,6,7). This cooperative/synergistic process can involve a convergence of intracellular signaling pathways triggered by ECM proteins and growth factors and becomes important in tissue regeneration.
Reports in the literature show evidence of type I collagen involvement in the ECM-directed differentiation of bone cells (8,9). Type I collagen regulates BMP-2 activity via an interaction with ␣ 1 ␤ 2 integrins (8,10). BMP-7 and type I collagen demonstrated a synergistic activation of Ocn and Bsp gene expression via the MAPK pathway for accelerated osteogenesis (11). Type I collagen is the major protein constituent of the bone and dentin ECM, comprising ϳ90% of the total protein matrix (9,12,13). The remaining 5-10% is composed of noncollagenous proteins (NCPs), which have proposed functions in the formation of mineralized tissues (12, 14 -16). Bone and dentin formation share many properties, among which are several common ECM proteins (13,15). OCN, MEPE, osteonectin (ON), OPN, BSP, DMP-1 and DMP-3 are all acidic NCPs that possess Ca 2ϩ and hydroxyapatite (HA) binding properties (16 -20). Initiation, nucleation, and inhibition of HA crystal growth is rigorously controlled by the NCPs (13,(21)(22)(23)(24). However, little is known regarding the role of these ECM proteins in signaling related to tissue morphogenesis. OPN (25), MEPE (26), and DMP-1 (15) have been shown to regulate gene expression; thus there is building interest in the NCPs, specifically the SIBLINGS, and how they might also contribute to intracellular activities and synergize with growth factors (5).
DMP-1, a SIBLING protein and NCP localized in dentin and bone ECM, was shown to stimulate mouse embryonic mesenchymal stem cells toward an osteoblastic lineage (15). Overexpression of DMP-1 in C3H101/2 and MC3T3-E1 cells resulted in up-regulation of Runx2, Ocn, Bsp, Alp, On, and Opn and increased mineralized nodule formation (15). DMP-1 is a highly acidic, phosphorylated protein consisting mainly of glutamic acid, serine, and aspartic acid residues (27). In situ hybridization experiments have shown that DMP-1 is expressed by hypertrophic chondrocytes, osteoblasts, and odontoblasts (28). Another dentin ECM protein, phosphophoryn (PP), a cleavage product of DSPP (16), has been implicated as a regulator of mineral crystal formation (16,18,19,29). DSPP is localized to chromosome 4, linking mutations in the gene to dentinogenesis imperfecta type II (15,16). Although initially thought to be tooth-specific, the Dspp message is also localized in mouse calvaria and rat tibia, although at a much lower levels (30). PP is the most abundant NCP in dentin ECM, comprising ϳ50% of the ECM protein sector (16). Like other proteins in the bone/ dentin microenvironment, PP is highly phosphorylated and anionic in character (18). PP is exceedingly rich in aspartic acid and serine residues, and ϳ85-90% of the serine residues are phosphorylated in the endoplasmic reticulum (31)(32)(33)(34). The majority of the protein sequence consists of (DSS) n repeats, where n could be as high as 24 (35,36). Odontoblasts secrete PP along the mineralization front (21,37,38) and, typical of other NCPs, the physiochemical properties of PP dictate high affinity for Ca 2ϩ , which implicates a role in the nucleation or modulation of HA crystal formation (13,22,24). An RGD domain is present at the N-terminal domain of PP (13,19), suggesting an auxiliary function in ECM-cell communication and in the initiation of intracellular signaling pathways. PP, like DMP-1, is a SIB-LING protein family member and could have an important regulatory role in osteogenic signaling and growth factor activity. Consequently, we hypothesized that PP is a signaling molecule involved in the formation of mineralized tissues. To test this hypothesis, we quantified expression of Runx2, Osx, Ocn, and Bsp and measured phenotypic outcome by OCN protein release, ALP activity, and mineral deposition. In addition, PP might have alternative roles based on cell type and post-translational modifications. Therefore, we analyzed the outcome of PP treatment using a recombinant form of PP (non-phosphorylated) in three independent cell types, hMSC, MC3T3-E1, and NIH3T3. We also created a stably transfected NIH3T3 cell line that secretes PP. Our rationale was to assess the signaling role of PP that could be phosphorylated by the stably transfected fibroblasts. Here, we report that PP, an NCP and SIB-LING protein, drives cell differentiation through integrin signaling and activation of the MAPK pathway.

EXPERIMENTAL PROCEDURES
Materials-Vectors for the production of recombinant PP (pGEX) and stable transfection (pShooter-ER) were obtained from Amersham Biosciences and Invitrogen, respectively. BL21 cells were obtained from Invitrogen. Luria-Bertani medium, ampicillin and isopropyl-1-thio-␤-Dgalactopyranoside were obtained from Sigma. Thrombin, glutathione-Sepharose 4B, and p-aminobenzamidine-Sepharose 4 Fast Flow were obtained from Amersham Biosciences. The FuGENE transfection reagent was obtained from Roche Applied Science. Human adult mesenchymal stem cells (hMSCs) were obtained from BioWhittaker, Inc. (Walkersville, MD). Human mesenchymal stem cell medium, mesenchymal cell growth supplement, L-glutamine, penicillin, and streptomycin were obtained from BioWhittaker, Inc. and added to the medium according to the manufacturer's specifications to prepare complete mesenchymal stem cell medium. MC3T3-E1 (clone 4) and NIH3T3 cells were obtained from the American Type Culture Collection (Manassas, VA). Dulbecco's modified Eagle's medium, fetal bovine serum, penicillin/streptomycin, and trypsin-EDTA were obtained from Invitrogen. 1, 25-(OH) 2 vitamin D 3 was obtained from Biomol (Plymouth Meeting, MA). The antibody to ␣ v ␤ 3 integrin (anti-␣ v ␤ 3 ) was obtained from Chemicon (Temecula, CA). An OCN ELISA kit was obtained from Zymed Laboratories Inc.. PBS, ALP assay kits, alizarin red S, and cetylpyridinium chloride (CPC) were obtained from Sigma Diagnostics, Inc. Total protein assay kits were obtained from Bio-Rad. RNeasy Kit and DNase I were obtained from Qiagen (Valencia, CA). The RiboGreen kit was obtained from Molecular Probes (Eugene, OR). All quantitative real time PCR reagents, primers, and probes were purchased from Applied Biosystems (Foster City, CA). Protease inhibitors were purchased from Pierce Biotechnology. Anti-phospho-p38, anti-phospho-ERK1/2 and anti-phospho-JNK were purchased from Cell Signaling Inc. (Beverly, MA). Western Lightning chemiluminescence reagents were purchased from PerkinElmer Life Sciences.
Generation of Recombinant PP and Transfected Cell Line-Isolated mouse genomic PP was used as a template to amplify exon 5 by PCR. The primers used were designed with SalI and XbaI at the 5Ј-ends of the gene-specific sequence (bold letters). In addition we inserted five random bases 5Ј to the restriction site to allow SalI and XbaI digestions. The primers used were 5Ј-CTAATGTCGACATGGAGAGTGGCAGCC-GTGGAGA-3Ј (forward) and 5Ј-GCATTCTAGATTAAAGCACCCGC-CATTCAAATCG 3Ј (reverse).
The thermocycling conditions were three cycles of 94°C for 70 s (denaturation), 52°C for 70 s (annealing), and 72°C for 2 min (extension) followed by 30 cycles of 94°C for 70 s (denaturation), 62°C for 70 s (annealing), and 72°C for 2 min (extension). The obtained PCR fragment was inserted into the pGEX-4T-3 vector and transformed into the bacterial host BL21. Cells were cultured in Luria-Bertani medium with ampicillin for 4 h at 30°C. Protein expression was induced by 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. The bacterial lysate was cleared by centrifugation and applied directly to glutathione-Sepharose 4B. After washing with PBS, the glutathione S-transferase-bound protein was eluted with thrombin. Thrombin was removed from eluates with p-aminobenzamidine immobilized on a Sepharose 4 Fast Flow matrix. The purified protein was electrophoresed on a polyacrylamide gel to verify the molecular mass ( Fig. 1), and the amino acid composition matched the cloned sequence. Recombinant PP (rPP) was stored at Ϫ80°C until use. The PP PCR product was also inserted into the pShooter-ER vector and transfected into NIH3T3 using the FuGENE 6 Transfection reagent. Stably transfected cells were selected using G418. Expression of the PP gene was verified by reverse transcription PCR, and PP protein secretion was verified by dot blot analysis (data not shown). The anti-PP was a generous gift from Dr. Arthur Veis at Northwestern University (39). PP produced by stably transfected NIH3T3 is denoted tPP.
Cell Culture-hMSC, MC3T3-E1, and NIH3T3 were plated in 35-mm culture wells and grown in basal media to ϳ70% confluence. The basal medium for stem cells was complete mesenchymal stem cell medium, and for MC3T3-E1 and NIH3T3 it was Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin/ streptomycin. Cells were then treated with either rhBMP-2 (MC3T3-E1 and NIH3T3, 50 ng/ml; hMSC, 100 ng/ml) or 50 ng/ml rPP in basal media. The 50 ng/ml rPP concentration was determined based on a pilot experiment using a dose curve of 50, 100, and 250 ng/ml rPP for MC3T3 and NIH3T3 cells. Cells were cultured for 2, 4, and 8 days prior to RNA extraction. Media was renewed every second day. Where noted, hMSCs were also cultured in the presence of 100 nM dexamethasone (dex) as the negative control. We chose these time points to analyze gene expression based on recommendations from the literature (40,41).
RNA Extraction and Quantification-Total RNA was extracted using the RNeasy Kit with DNase I treatment according to the manufacturer's protocol. RNA content was determined using the RiboGreen RNA quantification kit. Conventional RNA quantification using 260/280 absorbance readings proved to be too imprecise to match the specificity of qPCR. Total RNA content was photometrically analyzed with a Tecan Spectrafluor plate reader with excitation at 485 nm and emission at 595 nm. RNA concentrations were calculated based on a standard curve of control ribosomal RNA.
Quantitative Real Time PCR-Cells were harvested from the culture treatments at the time points described above. After extraction and quantification of RNA, quantitative real time PCR (qPCR) analysis was carried out using Taqman® one-step RT-PCR Master Mix. 10 -30 ng of total RNA were added per 50-l reaction with sequence-specific primers (200 nM) and Taqman® probes (200 nM). Sequences for all target gene primers and probes are shown in Table I. 18S primers and probes were designed by and purchased from Applied Biosystems. qPCR assays were carried out in triplicate on an ABI Prism 7000 sequence detection system. Thermocycling conditions were 48°C for 30 min (reverse transcription) and 95°C for 10 min (initial denaturation) followed by 40 cycles at 95°C for 15 s (denaturation) and 60°C for 45 s (annealing and extension). The threshold was set above the non-template control background and within the linear phase of target gene amplification to calculate the cycle number at which the transcript was detected (denoted C T ).
Gene expression values were calculated based on the comparative ⌬⌬C T method (separate tubes) detailed in Applied Biosystems User Bulletin Number 2 (42). For each primers/probe set, validation experiments demonstrated that the efficiencies of target and reference gene amplification were approximately equal; the absolute value of the slope of log input amount versus C T was Ͻ0.1. Target genes were normalized to the reference housekeeping gene at 18S. Fold differences were calculated for each treatment group using normalized C T values for the negative control at the appropriate time point as the calibrator.
Ocn Gene Expression and Protein Release Analysis-hMSCs were cultured as before for 8 days in basal media or basal media plus 100 ng/ml rhBMP-2 or 50 ng/ml rPP and supplemented with 10 nM 1,25-(OH) 2 vitamin D 3 for the final 48 h of culture (41). MC3T3-E1 and NIH3T3 were cultured similarly but did not require vitamin D 3 for the induction of Ocn. Total RNA was extracted as described above and analyzed via qPCR for Ocn gene expression. For the OCN ELISA, cells were cultured in rhBMP-2-or rPP-containing medium for 8 days. For the final 48 h of culture, cells were cultured in medium without serum added. Conditioned medium was collected and stored at Ϫ80°C until use. The OCN ELISA was performed according to the manufacturer's instructions. OCN concentration (nanograms per milliliter) was calculated from a standard curve and normalized to the total protein of the cell lysate as determined by the Bio-Rad protein assay.
ALP Activity-Cells were cultured in the above mentioned media supplemented with 10 mM ␤-glycerophosphate for all treatment groups. hMSCs were additionally supplemented with 100 nM dex. Cells were harvested by trypsinization and centrifugation after 7, 14, and 28 days in culture. MC3T3-E1 and NIH3T3 were cultured similarly without dex. Cell pellets were resuspended in 500 l of lysis buffer. Cell lysates were frozen at Ϫ80°C for at least 2 h prior to ALP activity assays. 5 l of thawed cell lysates were incubated with 200 l of ALP 10 reagent from the Sigma diagnostics kit for 30 min at 37°C. An initial absorbance reading (time 0) was taken at 405 nm prior to a 30-min incubation at 37°C and following it (time 30). ALP activity was calculated according the manufacturer's instructions. ALP activity was normalized to the total protein of the cell lysate.
Quantitative Alizarin Red Staining-hMSCs were cultured as before in 100 nM dex and 10 mM ␤-glycerophosphate-containing media for 28 days. Cells were fixed in 70% ice-cold ethanol for 1 h and rinsed with double distilled H 2 0. Cells were stained with 40 mM alizarin red S, pH 4.2, for 10 min with gentle agitation. Cells were rinsed five times with double distilled H 2 0 and then rinsed for 15 min with 1ϫ PBS and gentle agitation. Alizarin red was extracted from fixed cells by treatment with 500 l 10% CPC for 20 min with gentle agitation. Absorbance of extracted alizarin red in CPC solution was measured at 570 nm. Amount of alizarin red (in micrograms) was determined according to an alizarin red standard curve and normalized to the total protein of the cell lysate.
Integrin Blocking-hMSCs were seeded into 35-mm culture wells and allowed to reach ϳ70% confluence in basal media. Cells were washed twice with 1ϫ sterile PBS and treated with 15 or 25 g/ml anti-␣ v ␤ 3 diluted in serum-free basal media in a total volume of 500 l. Cells were incubated with rocking for 1 h at 37°C. Cells were then washed twice with 1ϫ sterile PBS and incubated in media containing 50 g/ml L-ascorbic acid phosphate and either 100 ng/ml rhBMP-2 or 50 ng/ml rPP for 48 h. Total RNA was extracted, and Runx2 gene expression was analyzed via qPCR.
MAP Kinase Activation-hMSC and NIH3T3 cells were cultured in triplicate as specified above, except that the cells were cultured over- night in serum-free media prior to rPP treatment. Conditions for treatment (time) were based on recommendations from the literature (43)(44)(45)(46). rPP (250 ng/ml) was added for 10, 20, 30, and 60 min, and the cells were lysed on ice in radioimmune precipitation assay buffer (150 mM NaCl, 1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mM Tris, pH 8) in the presence of protease inhibitors. The lysed cells were then pooled, and protein concentration was determined. The lysates were stored at Ϫ80°C at least 2 h prior to use. 30 g of total protein were loaded onto 10% SDS gels and subjected to SDS-PAGE. Gels were blotted onto polyvinylidene difluoride membranes and probed with anti-phosphorylated p38 (1:250), anti-phosphorylated ERK1/2 (1:250), and anti-phosphorylated JNK (1:250) by Western blotting. Bands were detected by chemiluminescence of horseradish peroxidase, and exposure to X-OMAT Kodak film was performed. We should note that 50 ng/ml rPP was the optimal dose for gene expression experiments, although treatment of cells with 250 ng/ml rPP yielded the optimum induction of MAPK components in comparison with the lower concentration of 50 and 100 ng/ml for rPP. For the gene expression experiments, 50 ng/ml rPP was probably sufficient because of the time of rPP-treatment and the high sensitivity of the qPCR assay. For the Western blotting technique, a shorter treatment time seemed to require a higher dose.
Statistical Analyses-For qPCR assays the coefficient of variation (COV) was calculated from three assay replicates. For all treatment groups and target genes analyzed, the COV did not exceed 3%. All experiments were performed at least twice, and one representative experiment is reported as the mean of three treatment triplicates Ϯ S.E. One-way analysis of variance (ANOVA), followed by Fisher's least significant difference (LSD) multiple comparison post hoc test using SYSTAT 9 software (Richmond, CA), was performed to determine significance among treatment groups. A p value Ͻ0.05 was considered statistically significant.

PP-regulated Expression of Osteoblastic Transcription Factors-
We quantified the expression of several osteoblast gene markers in response to PP treatment in three cell types. The time points reported represent the highest change in gene expression for each target gene compared with control. For all experiments, the negative control consisted of a basal medium unless otherwise noted. rhBMP-2 was used as a positive control. rPP up-regulated Runx2 gene expression 2.5-fold over the basal medium control (p Ͻ 0.05) in both hMSC and MC3T3-E1 but not in NIH3T3 cells after 2 days in culture ( Fig. 2A). Traditionally, hMSCs are cultured in medium containing dex to guide the cells toward osteoblastic lineage (41). We did not initially include dex in our cultures to avoid masking any changes that PP or rhBMP-2 may have on gene expression (47,48). It was our observation that changes in Runx2 and Ocn gene expression were not detected by qPCR when dex was added in combination with PP or rhBMP-2 (data not shown). On the other hand, Osx is not expressed in hMSCs in basal media. A baseline expression of Osx is required to quantify a "fold over control" by qPCR. Therefore the addition of dex was necessary for the gene expression analysis of Osx in hMSCs only. For MC3T3-E1 and NIH3T3, dex was not needed because they both express basal levels of Osx. rhBMP-2 up-regulated Osx above control in all three cell types (19-fold in hMSC, 10-fold in MC3T3-E1, and 15-fold in NIH3T3; p Ͻ 0.05). Osx gene expression was up-regulated 8-fold over control by tPP in NIH3T3 after 4 days in culture and was not affected in either hMSC or MC3T3-E1 (Fig. 2B). It is of interest to note that PP up-regulated Runx2 in hMSC and MC3T3-E1 but not in NIH3T3 cells. Osx expression had the opposite expression pattern, i.e. PP up-regulated Osx expression in NIH3T3 but not in hMSC or MC3T3-E1 cells. These data concur with recent evidence that Osx and Runx could be activated independently (49).
Bsp gene expression was not affected by rPP in MC3T3-E1, whereas our positive control, rhBMP-2, increased Bsp gene expression 20-fold over control (data not shown). In NIH3T3 cells and hMSCs up to 21 days in culture we neither detected a baseline of Bsp gene expression in the control nor a change when rPP or rhBMP-2 were added (data not shown). Bsp is usually expressed at later stages of differentiation, peaking just before matrix mineralization (40). We repeated this experiment in the presence of dex in hMSC to determine whether dex would induce Bsp gene expression. We detected a low level of Bsp expression, but rhBMP-2 and rPP did not enhance further expression (data not shown).

PP-Induced Ocn Gene Expression and Protein Production-PP up-regulated
Ocn gene expression over control in all three cell types (4-fold in hMSCs, 6-fold in MC3T3-E1 cells, and 3-fold in NIH3T3 cells; p Ͻ 0.05) (Fig. 3A). PP increased OCN protein release ϳ10-fold above negative controls (p Ͻ 0.05) for both hMSC (40 versus 4 ng/ml) and MC3T3-E1 (10 versus 1 ng/ml) cells (Fig. 3B). There was no change in OCN protein release in NIH3T3 cells due to PP treatment. These data raise the following question. Is the role of PP during matrix mineralization due to the physical properties of PP and/or a capacity to signal the expression of other proteins that could be in involved in mineralization process?
Phosphophoryn Increased ALP Activity in hMSCs-To further examine the function of the PP in osteogenic lineage progression we examined ALP activity, which is a common phenotypic marker for osteogenesis. ALP activity was measured over a time course of 7-28 days. Fig. 4 shows the highest increase in ALP activity in hMSC at day 14. In contrast, rPP-treated groups did not increase ALP activity in MC3T3-E1 and NIH3T3 cells or in the stably transfected NIH3T3 cells over the time course examined. The positive control, rhBMP-2, increased ALP activity over basal medium control for all three cell types.
Phosphophoryn Increased Calcium Deposition in hMSCs-When added in combination with 100 nM dex and 10 mM ␤-glycerophosphate in hMSCs, both rhBMP-2 and rPP demonstrated increased alizarin red staining after 28 days in culture (Fig. 5A). Without dex, no alizarin red is detected. Upon quantification of alizarin red stain with 10% CPC, dex alone, dex with rhBMP-2, and dex with rPP exhibited increased alizarin red staining over cultures that did not contain dex (Fig. 5B). Furthermore, rhBMP-2 and rPP demonstrated increased alizarin red staining in cultures supplemented with dex compared with hMSC cultured with dex alone. These data reinforce the above-posed question as to the role of PP in signaling and/or mineral deposition. We therefore investigated the signaling mechanism of PP via integrin/RGD interactions.
PP Regulates Gene Expression via the ␣ v ␤ 3 Integrin-PP has an RGD domain; therefore, we hypothesized that PP may be functioning in osteoblastic gene expression via binding to the ␣ v ␤ 3 integrin on the cell surface. As demonstrated above, rPP stimulated Runx2 gene expression in hMSCs. We therefore decided to test our integrin interaction hypothesis on hMSCs using a blocking antibody to the ␣ v ␤ 3 integrin. Upon the addition of the integrin blocking antibody, Runx2 gene expression due to rPP was decreased by ϳ60% compared with an uninhib-ited control (Fig. 6). Runx2 gene expression was not inhibited by anti-␣ v ␤ 3 in the rhBMP-2-treated hMSCs, as BMP-2 acts via it own specific receptors (types I and II). We suggest that PP functions to up-regulate bone-specific gene markers via binding to the ␣ v ␤ 3 integrin and the triggering of intracellular signaling pathways.
PP Regulates Gene Expression via the MAPK Pathways-To further study the signaling pathway following PP interaction with the ␣ v ␤ 3 integrin receptor, we investigated the involvement of the MAPK pathway and, more specifically, the activation of p38, ERK1/2, and JNK. As shown in Fig. 7, the p38 activation was apparent within 10 min of exposure to rPP in both NIH3T3 and hMSC cells. The ERK1/2 pathway seems to be active only in NIH3T3 cells, whereas the hMSCs were not activated compared with the control cells. When JNK was assessed for its activation, it was evident that there was a positive response at 10 min. These data clearly show that rPP is signaling via the MAPK pathway as demonstrated by the phosphorylation of p38, ERK1/2, and JNK. We believe that the gene activation shown in this paper by quantitative PCR is due to the activation of the MAPK pathway and the translocation of its components to the nucleus where they activate transcription of target genes.
FIG. 5. Alizarin red stain of hMSC. Cells were cultured as before for 28 days with 10 mM ␤-glycerophosphate and in the presence (ϩ) or absence (Ϫ) of 100 nM dex. A, alizarin red stain for calcium. B, quantification of alizarin red stain via extraction with 10% CPC in 10 mM phosphate buffer. Bars equal mean Ϯ S.E.; n ϭ 3. *, significant from without dex (Ϫ dex) control; †, significant from with dex (ϩ dex) control; p Ͻ 0.05.
FIG. 6. ␣ v ␤ 3 integrin blocking. hMSCs were pre-treated with 10 g/ml anti-␣ v ␤ 3 for 1 h and then supplemented with 50 g/ml L-ascorbic acid phosphate with 100 ng/ml rhBMP-2 or 50 ng/ml rPP and cultured for 48 h. qPCR analysis for Runx2 was performed. Gene expression is calculated as a percentage of unblocked control. Bars equal mean Ϯ S.E., n ϭ 3. *, significant from control; p Ͻ 0.05.

DISCUSSION
Previous studies have suggested that PP functions in mineralization because of its calcium binding properties and its highly acidic and anionic character (18,21,22,24,37,38). However, to date, PP has not been investigated as a signaling molecule that might regulate differentiative gene expression. Although well defined in several cell systems (1,4,6,7), ECMdirected cell differentiation in the bone microenvironment has not been well documented and is limited to type I collagen (8,11). We demonstrate here that PP, as a non-collagenous ECM protein and a SIBLING family member, is another component of the complex cascade of signaling events that leads to the formation of highly organized mineralized tissues.
We reason that PP functions in the formation of mineralized tissues in two principal roles (Fig. 8). First, PP binds to integrin receptors on the cell surface via its RGD domain, activating the MAPK signaling pathways that culminate in a mature osteoblast characterized by expression of early and late differentiative marker genes such as Runx2, Osx, and Ocn. Second, PP is localized within the ECM of bone and dentin and functions as a regulator of calcium phosphate deposition and crystal growth. The preference for one function over another is unknown but may be related to phosphorylation state. We propose that the mineralization process could probably be the following: 1) the result of the physiochemical properties of PP independent of cells as PP binds to HA by its highly acidic and anionic residues; 2) due to the signaling role of PP by regulating other bone/dentin genes; or 3) a combination of these two roles. We suspect that PP has both an autocrine and a paracrine effect on local cells. Progenitor (hMSC) and differentiated cells (NIH3T3 and MC3T3-E1) alike may be responsive to PP in the microenvironment. As such, PP could induce differentiation in progenitors and provide maintenance support for the differentiative pathway of mature osteoblasts/odontoblasts and even fibroblasts.
Signaling via integrins/MAPK activates many of the same pathways as growth factors, converging for an enhanced effect (1,6 -8,11). Non-collagenous ECM proteins could provide key signals that potentiate growth factor activity (1) for an enhanced therapeutic outcome. The literature reports that ECMdirected signals, transduced by integrins, play indispensable roles in the regulation of tissue-specific gene expression of primary osteoblasts (3, 50 -52). As a regulator of bone/dentin gene expression via the MAPK pathway, PP might activate these intracellular signaling pathways in concert with or independent of growth factors, resulting in optimized growth factor activity. At this time, the timing of signaling by PP versus BMP-2 is unknown. Our data demonstrate that PP and BMP-2 stimulate many of the same target genes, suggesting dual roles that may synergize inside the cell as an amplified signal.
Runx2 gene expression was induced by rPP in hMSC at a level equal to that induced by rhBMP-2. Runx2 was also upregulated by rPP in MC3T3-E1 but not in NIH3T3. Runx2 has been shown to function in an early commitment to the osteogenic lineage as a transcriptional regulator of other osteoblast genes (53)(54)(55). It is reasonable that Runx2 was only stimulated in cells that could be considered multipotential (hMSC) and osteoblast-like (MC3T3-E1). Perhaps the NIH3T3 cells, which are fibroblastic, lack other factors that are necessary for inducible Runx2. Up-regulation of Runx2 was moderate (ϳ2-3-fold) for either rhBMP-2 or rPP across all three cell lines. It has been documented that enhancement of downstream RUNX2 targets is regulated by "activated" RUNX2 (i.e. phosphorylated RUNX2) and not due to increases in Runx2 message per se (56,57). Runx2 is an important regulator of osteogenic and chondrogenic differentiation (54), and induction by PP suggests a role in progression of mesenchymal stem cells and osteoprecursors toward a more mature cell.
Osx was not induced by rPP in either hMSC or MC3T3-E1 cells. However, in PP-transfected NIH3T3 cells Osx was upregulated compared with the basal control. Furthermore, rh-BMP-2 induced Osx gene expression more potently than did tPP in NIH3T3. Interestingly, although Runx2 was not activated by rPP in NIH3T3, Osx was induced. Conversely, in FIG. 7. Activation of the MAPK pathway. hMSC and NIH3T3 were cultured with rPP for 10, 20, 30, and 60 min. Cell lysates were harvested and subjected to SDS-PAGE and probed for phosphorylated p38 (phos-p38), phosphorylated ERK1/2 (phos-Erk), and phosphorylated JNK (phos-Jnk) by Western blotting. Cont., control. hMSCs Runx2 gene expression was increased, but not Osx by rPP. rhBMP-2 induced Osx in all three cell lines with an exceptional induction in hMSC. To date, there is no conclusive evidence that Osx is a direct target of Runx2 (i.e. Osx may be activated via RUNX2-independent pathway(s)) (49). Up-regulation of Runx2 and Osx, the transcriptional units of osteogenesis, indicates progression of the osteogenic lineage and increases the likelihood for activation of downstream phenotypic changes. The connection between Osx and Runx2 remains unclear at this time. It is interesting that Osx was inducible by PP in a fibroblastic cell line but not in mesenchymal stem cells or osteoblasts. This finding suggests a uniqueness of Osx that could be related to the differentiative state of the cell. Moreover, the varied response shown in the activation of Runx2 and Osx across cell types becomes unified in the activation of downstream markers such as Ocn.
An increase in OCN protein release was detected in both hMSCs and MC3T3-E1 cells. We were unable to detect an increase in OCN protein release due to rPP or tPP in NIH3T3 cells. We speculate that NIH3T3 cells, like fibroblasts, do not have the capacity to produce the OCN protein or secrete it, although they do express the Ocn gene inducible by either rhBMP-2 or tPP. However, secretion of the OCN protein suggests that hMSCs and MC3T3-E1 cells have differentiated into more mature osteoblast-like cells. PP increased ALP activity only in hMSCs. We also analyzed Alp gene expression in the presence of dex, and the outcome was a slight increase in Alp gene expression due to rPP, validating our ALP activity assay result (data not shown). PP also enhanced the final outcome of osteoblastic differentiation, i.e. matrix mineralization. In the presence of dex, rPP-treated hMSCs deposited significantly more calcium as evidenced by alizarin red stain.
Although neither an increase in ALP activity nor an enhancement of mineralization occurred in MC3T3-E1 cells due to rPP, it is possible the rPP plays different roles at various stages of maturation, as was evidenced by the varied responses in gene expression across cell types. Furthermore, whereas tPP induced expression of Osx and Ocn in NIH3T3 cells, the recombinant form of the protein had no effect on any gene examined in NIH3T3 cells. We speculate that NIH3T3 cells that are genetically re-programmed to produce PP also phosphorylate the protein. The dose, phosphorylation state, or other native post-translational modifications of PP may have important implications in its function; NIH3T3 cells treated with rPP did not form a mineralized matrix, whereas PP-transfected NIH3T3 have shown extensive mineral deposition in vitro (data not shown). There was no change in matrix mineralization due to rPP in MC3T3-E1 (data not shown). The matrix mineralization of hMSCs could be explained by their inherent capacity to differentiate into the osteogenic lineage when provided the proper signals. These data raise many interesting questions such as the following two. Does the phosphorylation state of the protein have a role in the signaling function of PP? Will this novel role as a signaling protein affect matrix mineralization? The difference of response of rPP and tPP in the NIH3T3 cells suggests that the degree of phosphorylation of PP is likely a key regulator of the function of PP. The extracellular maintenance of the phosphorylation state of PP is unknown and could provide an additional level of control over its dual functions as a matrix mineralization regulator and a signaling molecule. Matrix mineralization by PP could be a cell-independent event based on its characteristics of acidity, anionic character, and HA binding properties in addition to degree of phosphorylation. However, the role of PP in gene expression of bone-and dentinspecific genes depends on cell surface receptors such as integrins and the activated pathways intracellularly.
Because ␣ v ␤ 3 integrins are also crucial for cell attachment and migration, the role of PP as a signal transducer via the MAPK pathway may be critical in the proper formation of mineralized tissues and possibly in abnormal pathologies such as cancer. Recently, it was shown that DMP-1 is expressed in lung cancer (58). The authors postulate that SIBLING protein family members may function in metastasis to bone, as ␣ v ␤ 3 integrins are over-expressed in tumor cells at metastatic sites (59). Therefore, metastatic tumor cells bind to SIBLING proteins in bone ECM via their ␣ v ␤ 3 integrin receptors. Two other SIBLING proteins, BSP and OPN, may also contribute to tumor cell survival by recruiting complement factor H to the cell surface, protecting them from lysis by the alternate complement pathway (60). Our data agree with this concept that SIBLING proteins elicit their actions by interactions with integrin receptors. Clearly, SIBLING protein family members are more than static ECM proteins; they play dominant and crucial roles in both normal and pathological tissue morphogenesis.
In conclusion, we have demonstrated that PP not only stimulates differentiative gene expression in a variety of cell types including mesenchymal stem cells, pre-osteoblasts, and nonosseous fibroblasts but also enhances matrix mineralization, the end-stage of osteoblast differentiation. The evidence demonstrated here suggests that cellular responses to PP differ with cell type and origin and may be based on the method of PP delivery, the dosage, and the post-translational modifications of PP. We suggest that the non-collagenous ECM proteins such as the SIBLING proteins of the bone/dentin microenvironment have important and specific functions related to cell fate, mineralization, and tissue morphogenesis. Future studies will be focused on understanding the degree to which PP functions in these two distinct yet related roles. Specifically, the phosphorylation state of PP will be investigated as it pertains to the function of PP in either or both of its roles as a regulator of dentinogenesis/osteogenesis. Elucidating the intricate signaling pathways of ECM proteins that orchestrate the development of mineralized tissues will support the design of novel tissue-regenerative therapies. In turn, PPs may possess the ability to harness growth factor activity and, together, enhance their signaling capacities.