Basic fibroblast growth factor regulates phosphate/pyrophosphate regulatory genes in stem cells isolated from human exfoliated deciduous teeth

Background Basic fibroblast growth factor (bFGF) regulates maintenance of stemness and modulation of osteo/odontogenic differentiation and mineralization in stem cells from human exfoliated deciduous teeth (SHEDs). Mineralization in the bones and teeth is in part controlled by pericellular levels of inorganic phosphate (Pi), a component of hydroxyapatite, and inorganic pyrophosphate (PPi), an inhibitor of mineralization. The progressive ankylosis protein (gene ANKH; protein ANKH) and ectonucleotide pyrophosphatase phosphodiesterase 1 (ENPP1/ENPP1) increase PPi and inhibit mineralization, while tissue-nonspecific alkaline phosphatase (ALPL; TNAP) is a critical pro-mineralization enzyme that hydrolyzes PPi. We hypothesized that regulation by bFGF of mineralization in SHEDs occurs by modulation of Pi/PPi-associated genes. Methods Cells were isolated from human exfoliated deciduous teeth and characterized for mesenchymal stem cell characteristics. Cells were treated with bFGF, and the osteogenic differentiation ability was determined. The mRNA expression was evaluated using real-time polymerase chain reaction. The mineralization was examined using alizarin red S staining. Results Cells isolated from primary teeth expressed mesenchymal stem cell markers, CD44, CD90, and CD105, and were able to differentiate into osteo/odontogenic and adipogenic lineages. Addition of 10 ng/ml bFGF to SHEDs during in vitro osteo/odontogenic differentiation decreased ALPL mRNA expression and ALP enzyme activity, increased ANKH mRNA, and decreased both Pi/PPi ratio and mineral deposition. Effects of bFGF on ALPL and ANKH expression were detected within 24 h. Addition of 20 mM fibroblast growth factor receptor (FGFR) inhibitor SU5402 revealed the necessity of FGFR-mediated signaling, and inclusion of 1 μg/ml cyclohexamide (CHX) implicated the necessity of protein synthesis for effects on ALPL and ANKH. Addition of exogenous 10 μm PPi inhibited mineralization and increased ANKH, collagen type 1a1 (COL1A1), and osteopontin (SPP1) mRNA, while addition of exogenous Pi increased mineralization and osterix (OSX), ANKH, SPP1, and dentin matrix protein 1 (DMP1) mRNA. The effects of PPi and Pi on mineralization could be replicated by short-term 3- and 7-day treatments, suggesting signaling effects in addition to physicochemical regulation of mineral deposition. Conclusion This study reveals for the first time the effects of bFGF on Pi/PPi regulators in SHEDs and implicates these factors in how bFGF directs osteo/odontogenic differentiation and mineralization by these cells. Electronic supplementary material The online version of this article (10.1186/s13287-018-1093-9) contains supplementary material, which is available to authorized users.

Results: Cells isolated from primary teeth expressed mesenchymal stem cell markers, CD44, CD90, and CD105, and were able to differentiate into osteo/odontogenic and adipogenic lineages. Addition of 10 ng/ml bFGF to SHEDs during in vitro osteo/odontogenic differentiation decreased ALPL mRNA expression and ALP enzyme activity, increased ANKH mRNA, and decreased both P i /PP i ratio and mineral deposition. Effects of bFGF on ALPL and ANKH expression were detected within 24 h. Addition of 20 mM fibroblast growth factor receptor (FGFR) inhibitor SU5402 revealed the necessity of FGFRmediated signaling, and inclusion of 1 μg/ml cyclohexamide (CHX) implicated the necessity of protein synthesis for effects on ALPL and ANKH. Addition of exogenous 10 μm PP i inhibited mineralization and increased ANKH, collagen type 1a1 (COL1A1), and osteopontin (SPP1) mRNA, while addition of exogenous P i increased mineralization and osterix (OSX), ANKH, SPP1, and dentin matrix protein 1 (DMP1) mRNA. The effects of PP i and P i on mineralization could be replicated by short-term 3-and 7-day treatments, suggesting signaling effects in addition to physicochemical regulation of mineral deposition.
(Continued on next page)

Background
Basic fibroblast growth factor (bFGF; also known as FGF2) is a member of the fibroblast growth factor family of related signaling molecules crucial for development, maintenance, and repair of tissues. While it regulates various cell processes including proliferation, migration, and differentiation, bFGF also controls maintenance of stemness in human stem cells [1,2]. In this regard, bFGF preserves the undifferentiated state of human pluripotent stem cells [3]. Downregulation of OCT3/4 is observed in human embryonic stem cells treated with fibroblast growth factor receptor (FGFR) inhibitor, suggesting a role for bFGF in regulation of genes governing pluripotency [4]. Hence, bFGF is employed as one of the key components in growth medium for human embryonic stem cells and human induced pluripotent stem cells.
Stem cells isolated from human exfoliated deciduous teeth (SHEDs) were reported in 2003 [5]. The appeal of these postnatal mesenchymal stem cells (MSCs) is that they can be obtained non-invasively (isolated after natural exfoliation of primary teeth) and show the multipotent ability to differentiate into osteogenic, adipogenic, and neurogenic lineages [6]. Although they are considered to be MSCs based on known developmental processes, SHEDs exhibit novel characteristics compared with bone marrow-derived MSCs, possibly due in part to their origin from cranial neural crest-derived ectomesenchyme and/or different patterns of signaling that occur in the craniofacial-dental region compared to the postcranial skeleton [5].
Functions of bFGF have been established in regulating some aspects of SHED biology. As it does for embryonic stem cells, addition of bFGF to culture medium maintains stemness in SHEDs in long-term culture [7]. In this context, bFGF enhances colony-forming unit (CFU) ability and the expression of pluripotent markers (e.g., OCT4, REX1, and NANOG) [7], but does not influence cell proliferation [7,8]. bFGF plays a potent role in regulating the differentiation of SHEDs into an odontoblast-like phenotype. Previous reports demonstrate that bFGF attenuates tissuenonspecific alkaline phosphatase (gene: ALPL; protein: TNAP) expression and mineral deposition by SHEDs and dental pulp stem cells (DPSCs), in vitro [1,9]. bFGF inhibits RUNX2 and BGLAP mRNA expression as well as mineral deposition by SHEDs in vitro via the regulation of ERK and Wnt signaling [8]. In addition, bFGF treatment of SHEDs leads to the reduction of ectopic bone formation in vivo [8].
Further, inhibition of endogeneous bFGF function using a chemical inhibitor of fibroblast growth factor receptor (FGFR) leads to increased mineralization upon osteogenic induction [6]. Correspondingly, SHEDs transfected with shRNA against bFGF exhibit higher mineral deposition than controls [6]. All of this accumulated evidence strongly supports a negative influence of bFGF on differentiation of SHEDs into mature, mineralizing odontoblast-like cells.
Inorganic phosphate (P i ) and pyrophosphate (PP i ) play crucial roles in physiological and pathological extracellular matrix (ECM) mineralization. P i is a primary component of hydroxyapatite crystals that are deposited in the biomineralization of the bone and teeth, while PP i is a potent inhibitor of crystal precipitation and growth [10]. In addition to their physicochemical roles in mineralization, both P i and PP i have been reported to have signaling effects on cells, though mechanisms remain incompletely understood. The addition of P i promoted mineralization in rat osteoblasts [11]. Conversely, PP i supplementation resulted in reduction of mineral deposition in vitro with reduced cell proliferation and collagen synthesis in murine cementoblasts [12]. Local pericellular P i and PP i concentrations are regulated by TNAP, ectonucleotide pyrophosphatase/phosphodiesterase-1 (ENPP1;ENPP1), and the progressive ankylosis protein (ANKH; ANKH in humans and Ank/ANK in mice) [10]. TNAP cleaves extracellular PP i into P i , facilitating ECM mineralization. ENPP1 increases extracellular PP i by cleaving nucleotide triphosphates, while ANKH regulates transport of intracellular PP i to the extracellular space. Functions of ANKH and ENPP1 result in increased extracellular PP i and subsequently inhibit mineralization. Studies on human loss-of-function mutations and genetically engineered mouse models have demonstrated that TNAP, ENPP1, and ANKH/ANK have profound effects on skeletal and dental mineralization [12][13][14][15][16][17][18].
To date, little is known about how bFGF affects expression of P i /PP i regulators. Studies using MC3T3.E1 murine pre-osteoblasts showed that bFGF upregulates Enpp1 and Ank mRNA expression, while Alpl mRNA levels are downregulated [19]. Another report demonstrated that bFGF inhibits ALPL expression in SHEDs [9]. However, an influence of bFGF on the other key P i /PP i regulatory genes in SHEDs has not yet been investigated, making it unclear how these genes contribute to bFGF regulation of osteo/odontoblast differentiation and mineralization. In the present study, we aimed to investigate in SHEDs the effect of bFGF on P i and PP i regulatory genes and roles of P i and PP i on mineralization of SHEDs.

Cell isolation and culture
The study was approved by Human Research Ethics Committee, Faculty of Dentistry, Chulalongkorn University (Approval number 2015-007). The procedure was performed according to the Declaration of Helsinki. Informed consent was obtained from parents. Primary teeth with no pathological lesions scheduled for extraction according to the clinical treatment plan were collected and stored in culture medium. Dental pulp tissues were obtained, and an explantation protocol was applied for cell isolation, using 35-mm tissue culture plate [2,6]. The migrated cells were subcultured when cell confluence was achieved. Cells were cultured in Dulbecco's modified Eagle medium (DMEM, Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA), 2 mM L-glutamine (Gibco, USA), 100 U/mL penicillin (Gibco, USA), 100 μg/mL streptomycin (Gibco, USA), and 5 μg/mL amphotericin B (Gibco, USA). Cells were maintained in 100% humidity at 37°C and 5% carbon dioxide.

Flow cytometry
Single cell suspensions were obtained by detaching cells with trypsin/EDTA solution. Cells were centrifuged, and the supernatant culture medium was discarded. Cells were rinsed with 1% FBS in PBS and further immunostained with primary antibodies conjugated to fluorescent dye, including anti-human CD44 (BD Bioscience Pharmingen, USA), PerCP-CyTM5.5-conjugated antihuman CD90 (BD Bioscience Pharmingen, USA), PE-conjugated anti-human CD105 (BD Bioscience Pharmingen, USA), and PerCP-conjugated anti-CD45 (BD Bioscience Pharmingen, USA). Cells were analyzed using a FACSCalibur flow cytometer using CellQuest software for operation and gating (BD Bioscience, USA).

Osteo/odontogenic induction
Cells were seeded in 24-well plates at a density of 50,000 cells per well. After 24 h, culture medium was replaced with osteogenic induction medium consisting of growth medium supplemented with 252.39 μM L(+)-ascorbic acid sodium salt, 10 mM β-glycerophosphate, and 100 nM dexamethasone. The medium was changed every 48 h. Mineral deposition was analyzed using alizarin red S staining, as described below. Osteogenic marker gene expression was determined using quantitative real-time polymerase chain reaction, as described below. To test involvement of FGF signaling, in some experiments, SHEDs were pretreated with an FGFR inhibitor (SU5402) 30 min prior to bFGF exposure and cells were maintained in osteogenic induction medium supplemented with bFGF and SU5402 for the length of the experiment.

Adipogenic induction
Cells were seeded in 24-well plates at a density of 12,500 cells per well. Adipogenic induction medium consisted of growth medium supplemented with 0.1 mg/mL insulin, 1 mM dexamethasone, 1 mM isobutylmethylxanthine (IBMX), and 0.2 mM indomethacin. The medium was changed every 72 h. Oil Red O staining was performed to evaluate intracellular lipid accumulation according to a previously published protocol [2].

Alkaline phosphatase activity (ALP) assay
Cells were lysed in an alkaline lysis buffer. The samples were then incubated with p-nitrophenol phosphate solution (2 mg/mL, Invitrogen, USA). After 15 min, the reaction was stopped by the addition of 50 mM NaOH. The absorbance was measured at 410 nm. Total protein was evaluated using a bicinchoninic acid (BCA) assay (Thermo Scientific, USA). ALP activity was calculated and normalized to the amount of total protein.

Alizarin red S staining
Alizarin red S staining was employed to determine calcium deposition in culture [20]. Cells were fixed with cold methanol for 10 min and washed with deionized water. Subsequently, cells were incubated with 1% alizarin red S solution for 3 min at room temperature under gentle agitation. Cells were then washed with deionized water to remove excess staining. Precipitated dye was solubilized in 10% w/v cetylpyridinium chloride solution, and the absorbance was measured at 570 nm.

Quantitative polymerase chain reaction (QPCR)
RNA isolation was performed using RiboEx total RNA isolation solution (GeneAll, Seoul, South Korea). Complimentary DNA (cDNA) was synthesized from RNA using reverse transcriptase ImPromII kit (Promega, Madison, WI, USA). QPCR was performed using Fas-tStart Essential DNA Green Master kit (Roche Diagnostic, USA) on the MiniOpticon real-time PCR system (Bio-Rad, USA). Target gene expression values were normalized to 18S expression values and further normalized to controls. Oligonucleotide sequences are listed in Additional file 1: Table S1.

Phosphate and pyrophosphate assay
After osteo/odontogenic differentiation, the culture medium was collected and analyzed using the EnzChek® Phosphate assay kit (Molecular PROBES, Oregon, USA) for quantitative analysis of P i and the EnzChek® Pyrophosphate assay kit (Molecular PROBES, Oregon, USA) for detecting free PP i following the manufacturer's protocol. Absorbance was measured at 360 nm.

Statistical analyses
Each experiment was performed using cells from at least four different donors. The Mann-Whitney U test was employed for comparisons of two groups. For comparison of three or more groups, the Kruskal-Wallis test was performed, followed by pairwise comparison with type I error correction. All statistical analyses were performed using Prism 7 (GraphPad Software, CA, USA). Statistical significance was indicated by p < 0.05.

SHED cell isolation and characterization
Cells were harvested from pulp chambers of the extracted primary teeth. Expression of mesenchymal and hematopoietic stem cell surface markers was evaluated using flow cytometry. Consistent with expectations for SHEDs, the isolated cells expressed MSC markers, CD44, CD90, and CD105 ( Fig. 1a-c), but lacked expression of hematopoietic stem cell marker, CD45 (Fig. 1d). The percentage of cells expressing CD44, CD90, and CD105 was over 95%, 99%, and 92%, respectively (Fig. 1e). In contrast, only 1% were CD45-expressing cells. Significant upregulation of ALPL and bone γ-carboxyglutamate protein (BGLAP) mRNA expression was observed after osteo/odontogenic induction for 7 days (Fig. 1f ), and after 14 days, a marked increase of mineral deposition was evident (Fig. 1g). Upregulation of adipogenic markers, lipoprotein lipase (LPL) and peroxisome proliferator-activated receptor γ (PPARγ), was observed after adipogenic induction for 8 days (Fig. 1h), and intracellular lipid accumulation was subsequently observed after 16 days (Fig. 1i). This combined evidence of MSC markers and multipotent differentiation ability

bFGF inhibits in vitro mineralization and alters osteo/ odontogenic gene expression in SHEDs
During osteo/odontogenic induction over 14 days, bFGF mRNA levels were reduced in a time-dependent manner until they were decreased by about 50% (p < 0.05) (Fig. 2a). Addition of 10 ng/mL bFGF to odontogenic induction medium led to the significant attenuation of mineral deposition at 14 days (p < 0.01) (Fig. 2b, c). This observation correlated with reduced ALP enzymatic activity in the bFGF-treated group after osteo/odontogenic induction for 7 days (Fig. 2d). Based on effects of bFGF on in vitro mineralization and ALP, we further investigated effects on genes associated with osteo/odontogenic differentiation and mineralization ( Fig. 2e-j). bFGF treatment reduced mRNA expression of transcription factor, Runt-related transcription factor 2 (RUNX2), on day 1 (40%; p < 0.05), but did not affect OSX expression (Fig. 2e, f ). bFGF decreased COL1A1 mRNA (by 40-50%; p < 0.05) on all days (Fig. 2g) and BGLAP (by 40%; p < 0.05) on day 3 (Fig. 2h), but did not affect expression of SPP1 or DMP1 (Fig. 2i, j).
bFGF alters expression of phosphate/pyrophosphate regulatory genes in SHEDs Experiments above confirmed bFGF regulates mineralization, ALP, and osteo/odontogenic genes in SHEDs. We next sought to determine if bFGF regulates expression of P i /PP i regulatory genes, including ALPL, ANKH, ENPP1, and the sodium-P i transporter, solute carrier family 20 member 1 (SLC20A1), during osteo/ odontogenic differentiation in SHEDs. Examination of Fig. 2 bFGF inhibits in vitro mineralization and alters osteogenic/odontogenic gene expression in SHEDs. a Expression of bFGF is reduced 50% (p < 0.05) in SHED cells during osteogenic induction over 14 days. Inclusion of 10 ng/mL bFGF in osteogenic induction medium significantly attenuates mineral deposition at 14 days (b, c) and reduces ALP enzymatic activity at 7 days (d). e-j Addition of bFGF reduces mRNA expression of RUNX2 (day 1), COL1A1 (days 1, 3, and 7), and BGLAP (day 3) (p < 0.05 for all), but does not significantly affect expression of OSX, SPP1, or DMP1. The Mann-Whitney U test was employed for comparisons of two groups. The Kruskal-Wallis test followed by pairwise comparison was used for comparison of three groups. Bars indicated the statistically significant difference (*p < 0.05, **p < 0.01). GM growth medium, OM osteogenic induction medium mRNA levels over 14 days of osteo/odontogenic differentiation demonstrated that ALPL was upregulated at day 7 (3-fold; p < 0.05) (Fig. 3a), while expression of ANKH tended towards decrease and ENPP1 and SLC20A1 tended towards increase, though these changes were not significant within the parameters of these experiments (Fig. 3b-d).
The influence of bFGF on P i /PP i regulatory gene expression was determined in osteogenic induction medium. bFGF treatment significantly inhibited ALPL mRNA expression on days 3 and 7 after (by 50% or more; p < 0.05) (Fig. 3e). Significant upregulation of ANKH mRNA was observed on all days (between 50 and 100% increased; p < 0.05) (Fig. 3f ). bFGF did not alter ENPP1 expression during osteo/odontogenic induction (Fig. 3g) and induced SLC20A1 mRNA (more than 100%; p < 0.05) on day 1 but not on later days (Fig. 3h).
SHEDs were pretreated with an FGFR inhibitor (SU5402) to confirm the involvement of bFGF signaling in modulation of ALPL and ANKH mRNA. SU5402 supplementation attenuated the influence of bFGF on both ALPL and ANKH mRNA expression, blocking the inhibitory effect of bFGF on ALPL (Fig. 3i) and preventing induction of ANKH upregulation (Fig. 3j). Inhibition of endogenous bFGF function by addition of SU5402 to osteogenic medium (in the absence of exogenous bFGF addition) resulted in a significant increase of ALPL expression at 7 days (3-fold; p < 0.050), but no significant regulation of ANKH mRNA levels (Fig. 3k).
In time course experiments, cells were supplemented with bFGF and maintained in growth medium with expression assayed at 6, 12, and 24 h. bFGF inhibition of ALPL mRNA began between 12 and 24 h (about 70% decrease; p < 0.05) (Fig. 4a). In contrast, the induction of ANKH mRNA by bFGF was observed at 6, 12, and 24 h (approximately 50-75% increase; p < 0.05) (Fig. 4b). Inclusion of FGFR inhibitor, SU5402, in growth medium for 24 h effectively abolished effects of bFGF on ALPL and ANKH expression (Fig. 4c, d). Inclusion of cycloheximide (CHX) in the media over 24 h also abolished effects of bFGF, strongly suggesting  1, 3, and 7), g does not significantly affect ENPP1 mRNA, and h increases SLC20A1 mRNA (day 1). Addition of 20 mM FGFR inhibitor (SU5402) inhibits bFGF-induced changes in i ALPL and j ANKH on day 7. k Addition of 20 mM SU5402 in the absence of exogenous bFGF increases ALPL (p < 0.05) but does not affect ANKH expression. The Mann-Whitney U test was employed for comparison of two groups. The Kruskal-Wallis test followed by pairwise comparison was used for comparison of three or more groups. Bars indicated the statistically significant difference. OM osteogenic induction medium (*p < 0.05) that protein translation is required as an intermediate step for the effects of bFGF on ALPL and ANKH (Fig. 4c, d).
Culture medium collected on days 1, 6, and 14 after osteo/odontogenic induction revealed bFGF significantly reduced P i /PP i ratio by day 14 (by about 80%; p < 0.05) (Fig. 4e), confirming a functional effect from reduced ALPL and increased ANKH expression. Addition of 0.05 U ALP rescued bFGF-attenuated mineral deposition (Fig. 4f).

Inorganic phosphate and pyrophosphate regulate mineralization by SHEDs
Effects of exogenous PP i on osteoblast and cementoblast mineralization in vitro have been reported [12,21]; however, to our knowledge, PP i has not been tested on stem cells, or more specifically on SHEDs. Addition of 10 μM PP i to osteogenic induction medium of SHEDs nearly completely inhibited mineral deposition by 14 days (Fig. 5a, b). At 7 days, inclusion of PP i increased COL1A1 (4-fold; p < 0.05) and ANKH (about 2-fold; p < 0.05), but did not significantly affect a number of other genes associated with mineralization and P i /PP i regulation, compared to untreated controls (Fig. 5c-l).
Effects of exogenous P i on osteoblast, cementoblast, and odontoblast mineralization and expression in vitro have been extensively studied [22][23][24][25], and experiments investigating effects of P i on various types of stem cells have been Fig. 4 Regulation of ALPL and ANKH expression by bFGF in SHEDs. Cells were maintained in normal growth medium in the presence or absence of 10 ng/mL bFGF for 6, 12, or 24 h. a Addition of bFGF significantly inhibits ALPL mRNA by 24 h. b The induction of ANKH mRNA by bFGF is observed at 6, 12, and 24 h. c, d For both ALPL and ANKH, inclusion of 20 mM FGFR inhibitor (SU5402) in growth medium abolishes effects of bFGF at 24 h. Addition of 1 μg/ml cyclohexamide (CHX) inhibits effects of bFGF on ALPL and ANKH expression, suggesting that protein translation is required. e In culture medium collected from SHEDs on days 1, 6, and 14 after osteogenic induction, bFGF significantly reduces P i /PP i ratio by day 14. f Addition of 0.05 U ALP rescues bFGF-attenuated ALP expression and mineral deposition in SHEDs. The Mann-Whitney U test was employed for comparisons of two groups. The Kruskal-Wallis test followed by pairwise comparison was used for comparisons of three or more groups. Bars indicate statistically significant differences (*p < 0.05, **p < 0.01). OM osteogenic induction medium reported [26,27]. However, to our knowledge, this relationship has not been explored in SHEDs or other dental stem cells. Addition of 5 mM P i to osteogenic medium significantly increased mineral deposition by SHEDs at both days 7 and 14 (about 3-and 2-fold, respectively; p < 0.05 for both) (Fig. 6a, b). At 7 days, inclusion of P i significantly induced OSX (10-fold), DMP1 (25-fold), SPP1 (50-fold), and ANKH (4-fold) mRNA (p < 0.05 for all); however, other genes were unaffected (Fig. 6c-l).
From the above experiments, it was not clear whether P i or PP i was acting early or late to influence cell mineralization in the 14-day experiment. Therefore, the effect of duration of P i or PP i treatment was investigated. Continuous administration of 5 mM P i or 10 μM PP i in osteogenic medium served as the positive control, no added P i or PP i served as negative control, and conditions A, B, and C incorporated shorterterm P i or PP i treatments of 1, 3, or 7 days, respectively, followed by completion of the experiment in osteogenic medium (Fig. 7a). Continuous treatment of SHEDs with P i or PP i resulted in significantly increased (nearly 3-fold; p < 0.05) or decreased (20-fold; p < 0.05) mineral deposition, respectively (Fig. 7b-e). Cells under condition A showed no significant difference in mineralization (Fig. 7b). Cells under condition B exhibited PP i -mediated inhibition of mineralization by 3 days (Fig. 7c). Lastly, cells under condition C showed increased mineralization from P i treatment (more than 2-fold; p < 0.05), while those receiving PP i treatment significantly reduced mineral deposition (Fig. 7d). These results within the first 7 days suggest that the supplementation of exogenous P i and PP i not only influences the precipitation and growth of mineral crystals, but may also affect intracellular mechanism(s).

Discussion
For the first time, we report effects of bFGF on regulation of P i /PP i and mineralization in SHEDs. Addition of Addition of 10 μM PP i to osteogenic induction medium of SHED cells nearly completely inhibits mineral deposition by 14 days. c-l At 7 days, inclusion of PP i increases COL1A1 (p < 0.05) and ANKH (p < 0.05), but does not significantly affect other genes associated with mineralization and P i /PP i regulation, compared to untreated controls. The Mann-Whitney U test was employed for comparisons of two groups, and bars indicated the statistically significant difference (*p < 0.05). OM osteogenic induction medium bFGF to SHEDs during in vitro osteo/odontogenic differentiation decreased ALPL mRNA expression and ALP enzyme activity, increased ANKH mRNA, and decreased P i /PP i ratio and mineral deposition. Effects of bFGF on ALPL and ANKH mRNA expression were detected within 24 h and required FGFR signaling and protein synthesis. Addition of exogenous PP i inhibited mineralization and increased ANKH, COL1A1, and SPP1 mRNA, while addition of exogenous P i increased mineralization and increased OSX, ANKH, SPP1, and DMP1 mRNA. These effects of PP i and P i on mineralization could be replicated by short-term 3-and 7-day treatments, suggesting early signaling effects in addition to physical-chemical modulation of mineral deposition. This study reveals the effects of bFGF on P i /PP i regulators in SHEDs and implicates these factors in how bFGF directs osteo/odontogenic differentiation and mineralization by these cells.

Regulation by bFGF of osteo/odontogenic markers and mineralization in SHED cells
Previous studies demonstrated that bFGF inhibited mineralization in SHEDs and DPSCs [1,9]. Here, we confirmed this effect of bFGF on mineralization in SHEDs and extended understanding of the underlying mechanisms by examining the influence of bFGF on other osteo/odontogenic markers and P i /PP i regulators. Type 1 collagen, encoded by COL1A1 and COL1A2 genes, composes 90% of the extracellular matrix (ECM) of the bone, dentin, and cementum. bFGF significantly inhibited COL1A1 expression during osteo/odontogenic differentiation on days 1, 3, and 7. It has been shown that bFGF suppresses COL1 and COL3 while enhancing mRNA expression of matrix metalloproteinases MMP1 and MMP2 in human periodontal ligament (PDL) cells, suggesting a catabolic influence of bFGF on ECM regulation [28]. Similarly, collagen synthesis was reduced by c-l At 7 days, inclusion of P i increases OSX, DMP1, SPP1, and ANKH (p < 0.05 for all), but does not significantly affect other genes associated with mineralization and P i /PP i regulation, compared to untreated controls. The Mann-Whitney U test was employed for comparisons of two groups, and bars indicated the statistically significant difference (*p < 0.05). OM osteogenic induction medium bFGF treatment in rat hair dermal papilla cells [29]. Hence, the bFGF-induced reduction of COL1A1 expression could be beneficial in terms of scar formation during wound healing. On the contrary, in rabbit chondrocytes, bFGF did not influence COL1A1 mRNA levels but suppressed COL2 mRNA expression [30]. Tenocytes overexpressing bFGF increased COL1 and COL3 expression.
We demonstrated that bFGF inhibits RUNX2 and BGLAP mRNA expression on days 1 and 3, respectively. RUNX2 serves a master transcription factor regulating osteo/odontogenic differentiation and function, upstream of numerous key genes [31,32]. BGLAP encodes OCN, an extracellular matrix protein found in the bone, dentin, and cementum and posited to regulate mineralization as well as perform wider endocrine functions [33,34]. Previous publications implicated bFGF inhibition of osteo/odontogenic differentiation of SHEDs in part via reduction of RUNX2 and BGLAP mRNA [8]. Similarly, studies on stem cells from apical papilla (SCAP) showed that bFGF treatment during osteo/odontogenic induction decreased BGLAP mRNA levels at early but not late time points [35]. In PDL cells, bFGF inhibited mineralization when added to osteogenic medium, but with no apparent regulation of RUNX2 or BGLAP expression, suggesting alternative mechanisms [36]. However, bFGF induced RUNX2 expression in human periodontal ligament cells which are maintained in growth medium [36]. In mouse osteoblasts, bFGF attenuated RUNX2-induced Alpl expression [37]. The influence of bFGF on RUNX2 and OCN at specific time points during osteo/odontogenic induction of SHEDs and other cells requires further investigation.
These collected data concerning effects of bFGF on COL1A1, RUNX2, and BGLAP suggest cell-type-specific responses to bFGF. The reduction of all three of these osteo/odontogenic markers concomitant with decreased mineral deposition supports a link to attenuation of mineralization; however, additional studies are warranted to explore these relationships further.

Effect of bFGF on P i and PP i regulators in SHED cells
During osteo/odontogenic differentiation, SHEDs upregulated ALPL expression early, while ANKH, ENPP1, and SLC20A1 mRNA levels were not markedly changed during the differentiation period assayed here. Alpl was Fig. 7 Regulation of mineralization in SHEDs by continuous and short-term P i and PP i treatment. a Cells were treated with 5 mM P i or 10 μM PP i in osteogenic medium, with no P i or PP i serving as negative control, and continuous administration of P i or PP i serving as the positive control, whereas conditions A, B, and C incorporated shorter-term P i or PP i treatments of 1, 3, or 7 days, respectively, followed by completion of the experiment in osteogenic medium. b Continuous treatment of SHEDs with P i significantly increases (p < 0.05) mineral deposition while or continuous PP i significantly decreases (p < 0.05) mineral deposition. c Cells under condition A show no significant differences in mineralization from negative controls. d Cells under condition B exhibit PP i -mediated inhibition of mineralization by 3 days (p < 0.05). e Cells under condition C show increased mineralization from P i treatment (p < 0.05) and decreased mineralization from PP i treatment (p < 0.05). The Mann-Whitney U test was employed for comparisons of three or more groups, and bars indicated the statistically significant difference (*p < 0.05). OM osteogenic induction medium induced early in OCCM.30 murine cementoblasts, and Ank, Enpp1, Spp1, and Dmp1 mRNA increased at later times after mineral deposition [38]. In MC3T3.E1 murine pre-osteoblasts, bFGF induced Enpp1 and Ank but inhibited Alpl mRNA expression [19]. In MLO-Y4 murine osteocyte-like cells, bFGF upregulated Ank, Enpp1, Slc20a1, and Dmp1 [39]. Conversely, we showed here that bFGF suppressed ALPL but induced ANKH expression in SHEDs, while ENPP1 expression was not markedly changed by exposure to bFGF. These changes were blocked by addition of an FGFR inhibitor, confirming the requirement of FGF signaling. In addition, the effect of bFGF on ALPL and ANKH expression was attenuated by protein synthesis inhibitor, suggesting the involvement of intermediate factors in the regulation process. Ultimately, the consequence of bFGF-reduced ALPL and increased ANKH in SHEDs was borne out by significantly decreased P i /PP i ratio, an alteration that likely contributed to reduced mineralization in vitro and would be anticipated to have a potent effect to inhibit ECM mineralization.
Based on inhibition of mineralization and changes in P i /PP i regulatory genes in response to bFGF, we explored direct effects of exogenous PP i or P i on SHEDs during osteo/odontogenic differentiation. Addition of exogenous PP i inhibited mineralization and significantly upregulated COL1A1 and ANKH expression in SHEDs. In OCCM.30 cementoblasts, 10-100 μM PP i decreased mineral deposition, and removal of PP i from culture medium during early stages allowed mineral deposition to proceed as normal, suggesting a primarily physicochemical effect on mineral deposition by these cells [12]. In contrast, a dose of 2.5 μM PP i increased Alpl and Ank and decreased Enpp1 in MC3T3.E1 pre-osteoblasts, implicating direct signaling effects [21]. We documented that withdrawal of PP i from medium after 1 day allowed SHEDs to mineralize similar to untreated controls; however, PP i withdrawal after days 3 or 7 did not reverse the inhibitory effect of PP i on mineralization. Because SHEDs largely mineralize between 7 and 14 days in osteogenic medium, these results suggest that PP i not only inhibits mineral precipitation and growth, but may also regulate cells through additional signaling mechanisms, a hypothesis that requires further investigation. In Alpl knockout (Alpl −/− ) mice, a model for the hereditary error-in-metabolism, hypophosphatasia (HPP), loss of TNAP function increases PP i and prevents mineralization in the bones and teeth. Severely hypomineralized dentin in Alpl −/− mice was accompanied by loss of expression of Bglap and dentin sialophosphoprotein (Dspp) mRNA by odontoblasts [40], supporting a connection between PP i , mineralization, and altered expression in skeletal/dental cells in vivo.
DMP1 is an ECM protein closely related to OPN that is expressed by odontoblasts, osteoblasts, and osteocytes. P i induction of Dmp1/DMP1 was previously documented in OCCM.30 murine cementoblasts and human PDL fibroblasts [12,52]. DMP1 is critical for dentin formation and periodontal development. Dmp knockout (Dmp1 −/− ) mice exhibit a dentinogenesis imperfecta-like phenotype including thin and hypomineralized dentin and an enlarged pulp cavity in the molars [53]. Additionally, loss-of-function of DMP1 is responsible for autosomal recessive hypophosphatemic rickets, an endocrine disorder marked by low circulating P i and severe hypomineralization of the bones and teeth [54].
OSX is a transcription factor directing osteoblast and odontoblast differentiation [55,56]. The upregulation of OSX in SHEDs by P i observed in the present study paralleled P i induced Osx expression in rat kidney fibroblast and human vascular smooth muscle cells [57,58]. In those cells, P i induced upregulation of Osx occurred via Akt1 and ERK signaling [57].
Addition of P i to SHEDs increased expression of ANKH mRNA. Induction of Ank was also achieved by addition of 5 mM P i to OCCM.30 murine cementoblast cells, hypothesized to be a response to pro-mineralization conditions and a sort of negative feedback to compensate the overabundance of extracellular P i and to balance P i /PP i ratio, and regulate the pace of mineralization [22]. Loss of function of Ank/ANKH leads to decreased PP i and increased and sometimes ectopic mineralization, and Ank mutant or knockout mice feature skeletal differences and massively increased acellular cementum [12,14,17,59]. These potent effects of ANK/ANKH on mineralized tissue development in vivo justify further investigation of the interactions of bFGF, ANKH, PP i , and mineralization in SHEDs.
Influence of local P i and PP i regulation on tooth development and regeneration P i and PP i participate in tooth development and regeneration. Their roles were demonstrated by the investigation in knockout and mutated mouse models. Alpl mutation, resulting in increased local PP i concentrations, causes delayed tooth eruption, aplasia or hypoplasia of cementum, and defective insertion of Sharpey's fibers [10,60]. Conversely, mice lacking Ank or Enpp1 function exhibited hypercementosis with otherwise normal dental and periodontal structure [10,61]. These data strongly support the significant role of local P i /PP i ratio via regulation by TNAP, ANK/ANKH, and ENPP1, in directing tooth development and structural formation.
Besides the role of P i /PP i in tooth development, modulation of this ratio has also been proposed as an alternative approach for regenerative treatment. TNAP immobilized fibrin scaffolds enhanced P i concentration and osteogenic differentiation in vitro as well as increased calvarial bone regeneration in vivo [62]. Local delivery of polyphosphate led to the increase alveolar bone regeneration around rat molars [63]. Studies of Ank knockout mice demonstrated that decreased PP i promoted more rapid and greater cementum regeneration in a periodontal defect model compared to wild-type mice [64], implying a role for P i /PP i regulation in periodontal tissue regeneration. However, more work must indeed be done to understand the mechanism(s) and to develop treatment strategies in translational and clinical settings.

Benefit of bFGF in future clinical applications
The present study aimed to investigate regulatory mechanisms whereby bFGF affected SHED differentiation and mineralization. We concluded that continuous bFGF treatment of SHEDs inhibited mineral deposition, in part via regulation of genes controlling P i /PP i metabolism. However, several reports demonstrated that bFGF pretreatment before osteogenic induction promoted osteogenic potential of various human stem/progenitor cells [65]. Our findings parallel those of Fakhry and colleagues, which indicated that continuous treatment of calvarial osteoblasts with bFGF inhibited osteogenic gene expression and mineralization, while short-term treatment of bFGF stimulated osteogenic differentiation [66]. Moreover, previous publications reported the beneficial utilization of bFGF in hard tissue formation through various approaches. Local application of bFGF enhanced dental implant stability and osseointegration in animal models [67,68]. In addition, bFGF has been utilized in alveolar bone and periodontal tissue regeneration, i.e., periodontal defect treatment and alveolar ridge preservation [69,70]. The anabolic effect of bFGF in vivo could result from the brief exposure of local cells to the released proteins, prompting osteogenic differentiation. In addition, bFGF regulates angiogenesis, cell migration, and cell proliferation [71,72]. Hence, these processes could help to facilitate the healing and formation of hard tissue in vivo.
Another potential use of bFGF in clinical regenerative therapy would be to maintain stemness of stem and progenitor populations in vitro. It has been shown that mesenchymal stem cells isolated from dental tissues lose their stem cell properties and decline in proliferation during long-term culture in vitro [7]. Our previous reports revealed that continuous bFGF treatment in vitro promoted maintenance of stemness in dental tissue-derived mesenchymal stem cells [1,2,7]. The bFGF supplementation in long-term culture of SHEDs enhanced colony-forming unit ability and stem cell marker gene expression [7]. Hence, bFGF could be useful to amplify cell number while maintaining stem cell properties in vitro for future regenerative application of SHEDs.

Conclusions
This study reveals the effects of bFGF on osteogenic differentiation ability of SHEDs. In addition, it has been demonstrated that bFGF influences P i /PP i regulators in SHEDs and implicates these factors in how bFGF directs osteo/odontogenic differentiation and mineralization by these cells.

Additional file
Additional file 1: Table S1