Trps1 transcription factor represses phosphate-induced expression of SerpinB2 in osteogenic cells

Abstract

Serine protease inhibitor SerpinB2 is one of the most upregulated proteins following cellular stress. This multifunctional serpin has been attributed a number of pleiotropic activities, including roles in cell survival, proliferation, differentiation, immunity and extracellular matrix (ECM) remodeling. Studies of cancer cells demonstrated that expression of SerpinB2 is directly repressed by the Trps1 transcription factor, which is a regulator of skeletal and dental tissues mineralization. In our previous studies, we identified SerpinB2 as one of the novel genes highly upregulated by phosphate (P_i) at the initiation of the mineralization process, however SerpinB2 has never been implicated in formation nor homeostasis of mineralized tissues. The aim of this study was to establish, if SerpinB2 is involved in function of cells producing mineralized ECM and to determine the interplay between P_i signaling and Trps1 in the regulation of SerpinB2 expression specifically in cells producing mineralized ECM. Analyses of the SerpinB2 expression pattern in mouse skeletal and dental tissues detected high SerpinB2 protein levels specifically in cells producing mineralized ECM. qRT-PCR and Western blot analyses demonstrated that SerpinB2 expression is activated by elevated P_i specifically in osteogenic cells. However, the P_i-induced SerpinB2 expression was diminished by overexpression of Trps1. Decreased SerpinB2 levels were also detected in osteoblasts and odontoblasts of 2.3Col1a1-Trps1 transgenic mice. Chromatin immunoprecipitation assay (ChIP) revealed that the occupancy of Trps1 on regulatory elements in the SerpinB2 gene changes in response to P_i. In vitro functional assessment of the consequences of SerpinB2 deficiency in cells producing mineralized ECM detected impaired mineralization in SerpinB2-deficient cells in comparison with controls. In conclusion, high and specific expression of SerpinB2 in cells producing mineralized ECM, the impaired mineralization of SerpinB2-deficient cells and regulation of SerpinB2 expression by two molecules regulating formation of mineralized tissues suggest involvement of SerpinB2 in physiological mineralization.

Graphical Abstract

1. Introduction

SerpinB2/Plasminogen Activator Inhibitor-2 (PAI-2) is a serine protease inhibitor and a component of the plasminogen system [1–6]. The importance of the plasminogen system in bone development has been demonstrated using urokinase-type (uPA) and tissue-type plasminogen activator (tPA) double KO mice (tPA^−/−;uPA^−/− mice) [7]. These mice present the elongation of endochondral bones phenotype and increased bone mass. Histologically, accelerated mineralization of osteoid and accumulation of noncollagenous proteins in bone matrix was detected in tPA^−/−;uPA^−/− mice, which was attributed to defective osteoblast-driven proteolytic processing of extracellular matrix (ECM) components [7]. The involvement of the plasminogen system in bone mineralization was further supported by increased bone mineral density in plasminogen-deficient (Plg^−/−) mice [8]. Studies of Plasminogen Activator Inhibitor-1 (PAI-1) in transgenic mice also support a regulatory role of the plasminogen system in bone homeostasis, however the function of this protein in bone seems to be independent of the proteolytic activity [9]. Unlike other players of the plasminogen system, SerpinB2/PAI-2 has not been studied in the context of bone development and homeostasis, most likely due to the lack of apparent developmental abnormalities in SerpinB2 KO mice, although bone phenotype of these mice has not been assessed [10]. SerpinB2 has been widely studied in the context of immunity and cancer [11–15]. This multifunctional serpin has been implicated in a variety of biological processes, including immunity, proteostasis, cell proliferation and survival, and remodeling of ECM [1]. SerpinB2 is one of the most upregulated proteins following cellular stress [1, 16–19]. Interestingly, we identified SerpinB2 as one of the novel genes highly upregulated by inorganic phosphate (P_i) in committed osteogenic cells (17IIA11 cell line) [20]. P_i is essential for the physiological mineralization of skeletal and dental tissues, and has been implicated in pathological vascular calcification in chronic hyperphosphatemic conditions [21–24]. P_i accomplishes this function, in part, by acting as a signaling molecule, which stimulates osteogenic differentiation of progenitor cells and formation of mineralized ECM by committed osteogenic cells [25–28]. However, in many other cell types elevated extracellular P_i is cytotoxic and induces a stress response [29–31]. Hence, it is not clear whether the detected P_i-activated SerpinB2 expression is a part of the cellular stress response or SerpinB2 plays a role in physiological mineralization. An evidence indirectly supporting the latter one, was provided by a recent study of human cancer cell lines showing that SerpinB2 expression is repressed by the GATA-type transcription factor Trps1 [32]. Since under physiological conditions, Trps1 regulates endochondral bone formation and the mineralization process [33–35], this finding supports the potential physiological function for SerpinB2 in cells producing mineralized ECM. Interestingly, it has been shown that P_i-regulated genes, specifically the ones involved in mineralization, have promoters enriched in GATA-regulatory elements [36].

The aim of this study was to characterize the expression pattern of SerpinB2 in skeletal and dental tissues in vivo and to determine the consequences of SerpinB2 deficiency on mineralization in vitro. Additionally, we aimed to decipher molecular mechanisms regulating SerpinB2 expression in cells producing mineralized ECM by addressing the interplay between two molecules (P_i and Trps1) involved in the formation and homeostasis of mineralized tissues. We hypothesized that SerpinB2 plays a role in physiological mineralization. This hypothesis was tested by a series of in vitro and in vivo experiments in WT and transgenic mice.

2. Materials and Methods

2.1. Mice

2.3Col1a1-Trps1 transgenic mice were generated previously [37]. Tissues from wild-type and 2.3Col1a1-Trps1 transgenic mice were harvested at postnatal day 7 and 14 (P7 and P14). Since the mice were pre-pubertal, we have not considered sex as a variable in this study. Animal studies were conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals, under a protocol approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

2.2. Cell lines and cell culture conditions

17IIA11 [38, 39], OCCM-30 [40, 41], MLO-A5 [42], MOVAS [43] and C3H10T1/2 cells [44] were used in this study. 17IIA11, OCCM-30, MOVAS and C3H10T1/2 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco; Thermo Fisher Scientific, Logan, UT; containing 154mM total Na concentration) supplemented with 5%–10% FBS (Thermo Fisher Scientific, Logan, UT), 100 units/ml penicillin and 100 μg/ml streptomycin (Cellgro, Manassas, VA) at 37 °C and 5% CO₂. MLO-A5 cells were maintained in alpha Modified Eagle Medium (aMEM, Gibco, Thermo Fisher Scientific, Logan, UT; containing 143mM total Na concentration) supplemented with 5% FBS, 5% FCS, 100 units/ml penicillin,100 μg/ml streptomycin and 1X L-Glutamine at 37°C and 5% CO₂. Two SerpinB2-deficient 17IIA11 cell lines (SB2-KD16 and SB2-KD17) were generated using lentivirus-mediated delivery of two different shRNAs-expressing constructs as described previously [20]. SB2-KD16 and SB2-KD17 cell lines were generated with TRCN0000080316 and TRCN0000080317 vector (mature antisense: TATGCGGAAGTTCTAGGATCT) (Dharmacon). Appropriate 17IIA11 control stable cell lines (generated with scrambled shRNA), Trps1-overexpressing (Trps1-OE) and Trps1-deficient (Trps1-KD) stable cell lines were described before [20]. The 17IIA11 shScr, SerpinB2-KD, Trps1-OE and Trps1-KD stable cell lines were maintained in growth medium supplemented with 10ug/ml of puromycin. For P_i responsive osteogenic gene expression, confluent cells after one day of seeding were stimulated with 5 mM Na-P_i for 12, 24 and 48 h. Mineralization was induced by supplementing growth medium with 5 mM Na-P_i buffer (pH=7.4) and 50 μg/mL ascorbic acid. Osteogenic medium was changed every 48h.

2.3. Cell proliferation assay (BrdU labeling)

Cells (1.0 × 10⁵) were plated on coverslips. After reaching confluency, cells were serum starved (used 0.5% FBS in the medium) overnight and immediately exposed to 5 mM P_i for 24h in standard growth medium (Supplementary Fig. 1A). After 24h, cells were incubated with BrdU labeling reagent (Invitrogen Cat#00–0103) for 6h, stained with AlexaFluor 488-conjugated anti-BrdU antibody (Life Technologies, Cat# B35130, 1:200 dilution), according to the manufacturer’s instructions and counterstained with DAPI. Cells were visualized under a fluorescent microscope (Nikon Eclipse 90i microscope) and BrdU positive cells from three randomly chosen fields from each independent slide were counted using NIS Elements software. Percentage of BrdU positive cells was then calculated, and the results are presented as the mean ± SD.

2.4. Histology

Femurs and mandibles from P7 and P14 wild-type and 2.3Col1a1-Trps1 mice were fixed in 10% buffered formalin for 24h and tissues were demineralized in 10% ethylenediaminetetraacetic acid (EDTA) solution (pH 7.4) prior to paraffin embedding. Serial 7 μm sagittal sections were collected and deparaffinized for hematoxylin and eosin (H&E) staining and immunohistochemistry.

2.5. Immunohistochemistry (IHC)

Sections used for analysis by IHC were subjected to antigen retrieval at 95°C in a citrate buffer (10mM, pH 6.0) for 10 minutes. Blocking was performed by incubating the sections with 10% BSA in 1XPBS/Triton X100 at room temperature for 1h. Sections were incubated with anti-SerpinB2 antibody (Abcam Cat# ab137588, 1:100 dilution) overnight at 4°C, followed by incubation with AlexaFluor 488-conjugated secondary antibody (Life Technologies Cat# A11008, 1:400 dilution) for 1h at room temperature. Nuclear counterstaining was performed using DAPI. Images were captured by a Nikon Eclipse 90i microscope.

2.6. Mineralization Assay

Mineralization was assessed using alizarin red (for calcium deposits) and von Kossa staining (for phosphate deposits), according to standard protocols in undifferentiated cells (day 0) and on days 5, 8, 12 and 15 of culture in osteogenic medium. Stained cells were imaged using Nikon TS100 microscope in a bright field. To quantify calcium deposits, alizarin red was extracted from stained cells and quantified by measuring the absorbance by spectrophotometry at 405 nm.

2.7. Fourier-transform infrared spectroscopy (FTIR)

FTIR spectroscopy was used to analyze mineralization in SerpinB2-deficient and 17IIA11 control cell lines on day 0, 5, and 8 of osteogenic differentiation. Four replicates from each group at each time point were collected. The cell sheets were scraped from the cell culture dishes, frozen in liquid N₂ and lyophilized. Five mg of lyophilized material from each sample were ground and mixed with 135 mg of potassium bromide (KBr), and pressed into a pellet at 1100 kg/cm² for 30 s using a hydraulic press. FTIR spectra of each sample was acquired at room temperature in the transmission mode, at a resolution of 4 cm⁻¹, and 64 scans per spectrum, using Bruker Vertex 70 FTIR spectrometer and Opus 6.5 software. Analyses were performed using Spectrum 10.02 software package (Perkin Elmer) and Origin Pro 2016 (OriginLab) to determine the mineral/protein ratio and the degree of mineral crystallinity for each sample. The details of the analyses are described in supplementary materials.

2.8. Western blot

Whole protein extracts were prepared by cell lysis in RIPA buffer supplemented with cOmplete™, EDTA-free Protease Inhibitor Cocktail (Sigma). Protein extract (10μg) was subjected to electrophoresis in 4–12% Bis Tris gels (Invitrogen) and transferred to a PVDF membrane (Fisher). Antibodies against SerpinB2 (Abcam Cat # ab137588) and V5-tag (Invitrogen Cat# R960–25) were used at 1:1000 dilution, and α-tubulin (Sigma Cat # T5168) was used at 1:10,000. Horseradish peroxidase-conjugated secondary antibodies (R&D, GE Healthcare) were used at 1:5000. The blots were developed with the Pierce™ ECL Western Blotting Substrate (Fisher, Cat# 32106). The intensity of the bands was measured by ImageJ (http://imagej.nih.gov/ij) and normalized by α-tubulin.

2.9. Quantitative RT-PCR (qRT-PCR)

RNA was extracted using RNeasy mini kit (Qiagen). Total RNA (1 μg), after DNase I treatment (Life Technologies, Grand Island, NY, USA), was converted to cDNA with SuperScript III Reverse Transcriptase kit (Life Technologies, Grand Island, NY, USA). Gene expression analyses were performed using AB Biosystems 7500 Fast Real-Time PCR System and Fast SYBR Green reaction mix (Roche, Indianapolis, IN, USA). The relative expression of genes of interest were calculated from the average Ct values of three independent experiments and normalized against Gapdh. Transcript abundances relative to Gapdh were calculated by using the 2^−ΔC_T method. All primer sequences are listed in Supplementary Table 1.

2.10. Chromatin immunoprecipitation (ChIP)

A ChIP assay was performed as previously described [45], with some modifications. Briefly, cells were incubated for 10 min in 1% formaldehyde, followed by addition of 0.125 M glycine, washed with ice-cold PBS, pelleted, and lysed in SDS Lysis Buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl pH 7.5). Samples were sonicated to generate DNA fragments of 200 to 500 bp using a Q800R2 sonicator (QSONICA). Chromatin from 17IIA11 stable cell lines expressing V5-tagged Trps1 (Trps1-OE1 and Trps1-OE2 lines) and 17IIA11 control (17A Ctrl1 and 17A Ctrl2) cells maintained in standard growth medium and upon exposure to 5 mM P_i for 24h was analyzed. The V5 ChIP was carried out with V5 beads (Sigma A7345). Normal mouse IgG (Upstate Biotechnology, Lake Placid, NY) was used as control. V5-agarose and normal mouse agarose beads were previously washed once in 1X PBS and blocked with 1.5% fish skin gelatin (Sigma G7765) and 200 μg salmon sperm DNA in 1X PBS for 1h rotating at room temperature and washed with PBS. Diluted chromatin was taken up to 1 ml with ChIP dilution buffer (0.01% SDS, 1.1% Triton X100, 1.2 mM EDTA, 16.7 mM Tris/HCl pH8.1, 167 mM NaCl) and 1ml of the dilution was used for each chromatin immunoprecipitation. 1% of the chromatin used for one ChIP was preserved as an input sample and stored at −20°C. The lysate was pre-cleared by rotation at 4°C for 2h using 1ml of the dilution with protein G beads (Sigma Cat# 16–201). After centrifugation at 4°C for 1min at 700 × g, the supernatant was incubated with the corresponding antibody rotating overnight at 4°C on a rotating wheel. Washing was carried out as previously described [45]. Elution was done in HENG buffer (10 mM HEPES-KOH pH 9.0, 1.5 mM MgCl₂, 0.25 mM EDTA, 20% glycerol), 250 mM KCl, 0.3% NP40 and 0.5 mg/ml V5 peptide repeated 3 times, 20 min each [46]. DNA was purified and amplified by qPCR using the primers shown in Supplementary Table 1.

2.11. Statistical Analysis

Experiments were repeated 3 times or otherwise stated. Values are expressed as mean ± standard deviation (SD). Statistical significant differences were determined using the Student’s t-test. A p-value of <0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA).

3. Results

3.1. Phosphate increases SerpinB2 expression specifically in committed osteogenic cells

To determine whether the initially observed activation of SerpinB2 expression by P_i [20] is unique for the 17IIA11 committed osteogenic cell line or this is a uniform cellular response to P_i, we analyzed expression of SerpinB2 in multiple cell lines. For that, we selected cell lines representing different types of osteogenic cells capable of producing mineralized ECM: 17IIA11 preodontoblasts [38, 39], MLO-A5 osteoblasts [42] and OCCM-30 cementoblasts [40, 41]. For comparison, MOVAS vascular smooth muscle cells (a model of pathologic vascular calcification) [43], which undergo rapid mineralization under standard osteogenic conditions, and non-mineralizing C3H10T1/2 pluripotent mesenchymal cells [44] were analyzed. Considering that elevated P_i activates some genes within a few hours (early-response genes), while activation of other genes is delayed (late-response genes) [36], expression of SerpinB2 was analyzed in cells exposed to 5 mM P_i for 12h, 24h and 48h (Fig. 1). qRT-PCR analyses did not detect SerpinB2 expression in any of the analyzed cells maintained in standard growth condition (C_t values were similar to a “no template” control). Significantly upregulated expression of SerpinB2 was detected in 17IIA11, MLO-A5 and OCCM-30 osteogenic cells treated with P_i for 24h in comparison with cells grown without additional P_i (Fig. 1A–C). This effect of P_i on SerpinB2 expression was transient, and only 17IIA11 cells maintained significantly increased SerpinB2 expression at 48h. Contrary to osteogenic cells, no change of SerpinB2 expression was detected in MOVAS nor C3H10T1/2 cells exposed to elevated P_i (Fig. 1D, E). This cell type-specific activation of SerpinB2 expression in response to P_i, considered together with a previous study that did not detect upregulation of SerpinB2 in P_i-treated MC3T3-E1 osteogenic precursor cells [36], suggest that P_i stimulates SerpinB2 expression specifically in committed osteogenic cells.

Fig. 1.

Phosphate increases SerpinB2 expression specifically in committed osteogenic cells. qRT-PCR results showing SerpinB2 expression in cells maintained in standard growth medium (Ctrl) in comparison with cells exposed to 5 mM P_i (P_i) for 12, 24 and 48h. (A) 17IIA11, (B) MLO-A5, (C) OCCM-30, (D) MOVAS, and (E) C3H10T1/2 cell lines (non-mineralizing control). qRT-PCR data are presented as the mean relative transcript abundance of SerpinB2 mRNA normalized to Gapdh ±S.D. from three independent RNA preparations per cell line; *p<0.05, and **p<0.01.

3.2. Inverse correlation between expression of Trps1 and SerpinB2

The Trps1 transcription factor has been shown to directly repress SerpinB2 expression in cancer cell lines [32]. Therefore, after establishing that P_i signaling molecule is a positive regulator of SerpinB2 in osteogenic cells, we analyzed the effect of Trps1 on SerpinB2 expression in these cells. For that, we used previously generated Trps1-deficient (Trps1-KD), Trps1-overexpressing (Trps1-OE) and appropriate control 17IIA11 stable cell lines [20]. Analyses of SerpinB2 mRNA levels by qRT-PCR demonstrated that P_i-stimulated SerpinB2 expression was dampened by Trps1 overexpression (Fig. 2A). This effect of Trps1 overexpression was also apparent on the protein level, as demonstrated by Western blot analyses showing reduced levels of SerpinB2 protein in P_i-stimulated Trps1-OE 17IIA11 stable cell lines in comparison with P_i-stimulated 17IIA11 control cells (Fig. 2B). These results suggest that increased levels of Trps1 prevent upregulation of SerpinB2 expression by P_i.

Fig. 2.

Inverse correlation between expression of Trps1 and SerpinB2 in 17IIA11 (17A) osteogenic cell line. (A) qRT-PCR results showing SerpinB2 expression in three 17IIA11 control cell lines (17ACtrl1, 17ACtrl2, 17ACtrl3) and two 17IIA11 cell lines overexpressing Trps1 (Trps1-OE1, Trps1-OE2) maintained in standard growth medium and upon exposure to 5 mM P_i (P_i) for 24 and 48h. qRT-PCR data are presented as the mean relative transcript abundance of SerpinB2 mRNA normalized to Gapdh ±S.D. Asterisks denote statistical significant difference: *p<0.05. (B) Representative Western blot images and a graph (showing densitometry-based quantification of the Western blot results by ImageJ) of SerpinB2 protein (47kDa) in 17IIA11 control cell lines (17ACtrl1, 17ACtrl2, 17ACtrl3) and 17IIA11 cell lines overexpressing Trps1 (Trps1-OE1, Trps1-OE2) maintained in standard growth medium and upon exposure to 5 mM P_i for 24h. α-tubulin was used as the protein loading control. (C) qRT-PCR results showing SerpinB2 expression in three 17IIA11 control cell lines (17A WT, shScr1, shScr2) and two 17IIA11 cell lines deficient in Trps1 (Trps1-KD1, Trps1-KD2) maintained in standard growth medium. Individual data points of 3 independent experiments per cell line, mean ± SD are shown. Asterisks denote statistical significant difference: ***p<0.001, ****p<0.0001. (D) Representative Western blot images and graph (showing densitometry-based quantification of the Western blot results by ImageJ) of SerpinB2 protein (47kDa) in 17IIA11 control cell lines (17A WT, shScr1, shScr2) and two 17IIA11 cell lines deficient in Trps1 (Trps1-KD1, Trps1-KD2) maintained in standard growth medium. α-tubulin was used as a protein loading control.

On the contrary, 17IIA11 cells deficient in Trps1 demonstrated increased basal expression of SerpinB2 and higher levels of SerpinB2 protein in comparison with control 17IIA11 cell lines (Fig. 2C, D). This suggests that Trps1 downregulates SerpinB2 and loss of this repressor activity is sufficient to increase SerpinB2 expression without additional extracellular stimuli. Of note, because Trps1-deficient cells do not respond to elevated extracellular P_i (data not shown), we did not analyze the effect of P_i on SerpinB2 expression in Trps1-KD cells.

3.3. SerpinB2 deficiency does not affect P_i-stimulated proliferation of osteogenic cells

To understand the role of SerpinB2 in osteogenic cells, we analyzed the consequences of SerpinB2 deficiency on the physiology of 17IIA11 cells. For that, we generated two SerpinB2-deficient (SB2-KD16 and SB2-KD17) 17IIA11 stable cell lines using two different shRNAs. Efficient depletion of SerpinB2 was confirmed on the protein level (Supplementary Fig. S2).

Considering that P_i and Trps1 are regulators of cell proliferation, in addition to being regulators of the mineralization process [25, 47–54], and that SerpinB2 has also been implicated in the regulation of proliferation in vitro [55–57]; first, we analyzed the effect of SerpinB2 deficiency on cell proliferation. Results of the BrdU incorporation assay demonstrated that there is no difference in proliferation of control and SerpinB2-KD 17IIA11 cell lines grown in standard conditions and that P_i increases proliferation of both control and SerpinB2-deficient cells (Fig. 3 and Supplementary Fig. S1). These results suggest that SerpinB2 deficiency does not affect P_i-stimulated cell proliferation.

Fig. 3.

SerpinB2 deficiency does not affect P_i-stimulated proliferation of osteogenic cells. Proliferation of control (17IIA11, shScr1, shScr2) and SerpinB2-deficient (SB2-KD16 and SB2-KD17) 17IIA11 cell lines was compared using BrdU labeling and quantification of BrdU-positive cells. (A) Representative IF images showing BrdU-positive control (shScr2) and SerpinB2-deficient (SB2-KD16) 17IIA11 cell lines maintained in standard growth medium (Ctrl) and stimulated with 5 mM P_i (P_i) for 24h. (B) Quantification of cell proliferation, expressed as the percentage of BrdU-positive cells. Individual data points of 3 experiments per cell line, mean ± SD are shown. Asterisks denote statistical significant difference: *p<0.05, **p<0.01.

SerpinB2 has been reported to protect cells from stress, in part by limiting Caspase-3 activation [16, 58], while P_i can induce stress response in non-osteogenic cells [29–31]. Therefore, we addressed a potential role of SerpinB2 in protecting osteogenic cells from P_i-induced stress. Western blot analyses (Supplementary Fig. S3) of apoptotic markers, cleaved Caspase-3 and cleaved PARP, did not detect significant levels of these proteins in cells treated with P_i in comparison with cells treated with DTT (dithiothreitol, a positive control for apoptosis). Importantly, levels of cleaved Caspase-3 and cleaved PARP in P_i-treated SerpinB2-deficient cells were as low as in P_i-treated control cells (Supplementary Fig. S3). These results suggest that SerpinB2 deficiency does not trigger osteogenic cell sensitivity to P_i-induced stress.

3.4. SerpinB2 deficiency results in delayed and decreased mineralization of 17IIA11 cells

Next, we analyzed the effect of SerpinB2 deficiency on mineralization of 17IIA11 cells. Mineralization of control and SerpinB2-deficient cells was compared at day 5 and day 8 of culture in osteogenic medium, which represent ongoing mineralization and the mineralization plateau, respectively [20, 38]. At day 5, mineralization of control 17IIA11 cells was well advanced as demonstrated by alizarin red and von Kossa staining, while very few calcium and phosphate deposits were detected in SerpinB2-deficient cells (Fig. 4A–C and Supplementary Fig. 4). The decreased mineralization, although to a lesser degree, was also detected at day 8, 12 and 15 in SerpinB2-deficient cells (Fig. 4A–C and Supplementary Fig. 4). FTIR spectroscopic analyses revealed differences in the mineral/protein ratios, which are used to determine the degree of mineralization in tissues and cell cultures, between control 17IIA11 and SerpinB2-deficient cells (Fig. 4D,E and Supplementary Fig. 5). Specifically, the ν₄/Amide I ratio in the control at day 5 was 0.287 while in the SerpinB2-KD it was 0.253. A greater difference was observed on day 8 with a ν₄/Amide I ratio of 0.310 in the control and 0.234 in the SerpinB2-KD. These differences were highly significant only at day 8 (Fig. 4E). We have also assessed the crystallinity of the mineral in the control and SerpinB2-KD 17IIA11 cells. The mineral crystallinity in the cultures of SerpinB2-deficient cells was significantly lower than in the controls on both time points (Fig 4F). These results indicate that deficiency of SerpinB2 in committed osteogenic cells negatively affects mineralization, as evidenced by the decrease in mineral deposition and crystallinity, providing further evidence that SerpinB2 supports the mineralization process.

Fig. 4.

SerpinB2 deficiency results in delayed and decreased mineralization in vitro. (A) Representative images of alizarin red staining at mineralization day 0, 5 and 8 of 17IIA11 control (17A WT, shScr1, shScr2) and SerpinB2-deficient (SB2-KD16 and SB2-KD17) cell lines. (B) Quantification of alizarin red is presented as the mean ± SD from three independent experiments per cell line. Asterisks denote statistically significant difference: *p<0.05; **p<0.01; ****p<0.0001. (C) Representative images of von Kossa staining at mineralization day 0, 5 and 8 of 17IIA11 control and SerpinB2-deficient cell lines. (D) Representative FTIR spectra (in the 1800–400 cm⁻¹ region) of ECM from 17IIA11 control and SerpinB2-deficient cells at mineralization day 5 and 8. The solid arrow shows Amide I peak in the 1800–1500 cm⁻¹ region and the arrowhead shows ν₄ peak in the 700–500 cm⁻¹ region. (E) Mineral content, expressed as mineral (ν₄ of phosphate) to organic (protein Amide I) ratios and (F) mineral crystallinity of 17IIA11 control and SerpinB2-deficient cells at mineralization day 5 and 8. Data represents mean ± SD of 4 individual experiments. Asterisks denote statistically significant difference: *p<0.05, **p<0.01, ***p<0.001.

3.5. SerpinB2 is highly expressed in skeletal and dental cells producing mineralized ECM

To determine if the observed effect of SerpinB2 deficiency on mineralization is physiologically relevant, we analyzed expression of SerpinB2 in bone and developing teeth. Considering that SerpinB2 protein exists in two forms, intracellular and extracellular [2, 59–62], we used immunohistochemistry (IHC) to be able to determine the localization of SerpinB2 protein.

Analyses of femurs and mandibles of P7 mice detected high SerpinB2 protein levels in hypertrophic chondrocytes and osteoblasts of primary spongiosa (Fig. 5A). SerpinB2 protein was also detected in resting and proliferating chondrocytes, although at much lower levels than in hypertrophic chondrocytes and osteoblasts. Of note, SerpinB2 was predominantly localized in the cytoplasm of chondrocytes and osteoblasts, but not in the cartilage nor bone ECM at P7. Analyses of alveolar bone of P7 mice confirmed high levels of SerpinB2 protein in cytoplasm of osteoblasts but not in ECM.

Fig. 5.

SerpinB2 is highly and specifically expressed in cells producing mineralized ECM and hypertrophic chondrocytes. (A) Representative H&E (upper row) and corresponding IHC images (lower row) of distal femur of P7 WT mice showing high expression of SerpinB2 in hypertrophic chondrocytes (HC), osteoblasts (Ob) of secondary ossification center (SOC), and osteoblasts (Ob) of primary spongiosa. (B) Representative H&E and corresponding IHC images of alveolar bone and mandibular incisor of P7 WT mice showing high expression of SerpinB2 in osteoblasts (Ob) of alveolar bone and odontoblasts (Od) in the incisor. (C) Representative H&E and corresponding IHC images of mandibular incisor of P14 WT mice showing SerpinB2 protein in the apical part of odontoblasts (Od) and ameloblasts (Am), dentino-enamel junction (DEJ), predentin (pD) and developing enamel (En).

In the developing mouse incisor, high levels of SerpinB2 protein were detected in odontoblasts, but not in dental pulp cells (Fig. 5A). More detailed analyses of SerpinB2 at the later stage of tooth development when enamel is formed, revealed that ameloblasts are also highly positive for SerpinB2 in incisors of P14 mice (Fig. 5B). In both odontoblasts and ameloblasts, levels of SerpinB2 increased as the cells progressed in their differentiation with no or very little SerpinB2 detected in preodontoblasts and preameloblasts (Fig. 5B). Unlike in bone, extracellular SerpinB2 was detected in predentin and forming enamel ECM (Fig. 5B). In summary, we determined that SerpinB2 is highly and specifically expressed by cells producing mineralized ECM in both skeletal and dental tissues. This and the presence of SerpinB2 in predentin and enamel suggests its involvement in physiological mineralization.

3.6. Trps1 represses SerpinB2 expression in osteogenic cells

Interestingly, the SerpinB2 expression pattern in mineralizing tissues is opposite to the expression of the Trps1 transcription factor, which is highly expressed before mineralization occurs, but its expression declines after the initiation of mineralization [35, 37]. Therefore, our next step was to determine if SerpinB2 expression can be repressed by Trps1 in cells producing mineralizing ECM in vivo. We analyzed skeletal and dental tissues of 2.3Col1a1-Trps1 transgenic mice [37], which maintain high expression of Trps1 in mature osteoblasts and odontoblasts during mineralization of bone and dentin, respectively. IHC analyses of femurs and mandibles at P7 revealed that overexpression of Trps1 in osteoblasts results in decreased levels of SerpinB2 in osteoblasts lining trabecular bone (Fig. 6A), but no apparent difference was observed between osteoblasts in the alveolar bone of 2.3Col1a1-Trps1 and WT mice (Fig. 6B). In P14 incisors, lower levels of SerpinB2 were apparent in 2.3Col1a1-Trps1 odontoblasts in comparison with odontoblasts of WT mice (Fig. 6C). There was no difference in levels of SerpinB2 in WT and 2.3Col1a1-Trps1 ameloblasts. These results indicate that Trps1 represses SerpinB2 expression in vivo, in a cell-autonomous manner.

Fig. 6.

Transgenic overexpression of Trps1 results in cell type-specific downregulation of SerpinB2. (A) Representative H&E and IHC images of distal femur trabecular bone of P7 WT and 2.3Col1a1-Trps1 transgenic mice (Tg). Arrowheads indicate SerpinB2 in osteoblasts (Ob). Lower levels of SerpinB2 are apparent in osteoblasts in the trabecular bone of 2.3Col1a1-Trps1 in comparison with WT mice. (B) Representative IHC images of alveolar bone of P14 WT and 2.3Col1a1-Trps1 transgenic mice. Arrowheads indicate SerpinB2 in osteoblasts (Ob); arrows point at SerpinB2 in endothelial cells (End). There is no apparent difference in SerpinB2 levels in osteoblasts lining the alveolar bone nor in endothelial cells of WT and 2.3Col1a1-Trps1 mice. (C) Representative H&E and IHC images of mandibular incisor of P14 WT and 2.3Col1a1-Trps1 transgenic mice. Arrowheads indicate SerpinB2 in odontoblasts (Od), arrows point at SerpinB2 in ameloblasts (Am). 2.3Col1a1-Trps1 odontoblasts show lower levels of SerpinB2 than WT odontoblasts. There is no apparent difference in SerpinB2 levels in ameloblasts of WT and 2.3Col1a1-Trps1 mice.

3.7. Trps1 binding to the SerpinB2 regulatory elements is regulated by P_i

Next, we used ChIP assay to determine if the observed downregulation of SerpinB2 in osteoblasts and odontoblasts overexpressing Trps1 is due to direct interaction of the Trps1 transcription factor with the SerpinB2 gene. To identify potential Trps1 binding sites that may serve as gene expression regulatory elements, we screened 2kb fragment upstream of the SerpinB2 transcriptional start site and the entire 4kb of the SerpinB2 first intron using TRANSFAC (TRANScription FACtor database). We identified 8 candidate GATA consensus sequences (BS1–8; Fig. 7A, Supplementary Table S1). Due to the lack of efficient anti-Trps1 antibodies, ChIP experiments were done in 17IIA11 stable cell lines expressing V5-tagged Trps1 (Trps1-OE1 and Trps1-OE2 lines) and 17IIA11 control (17A Ctrl1 and 17A Ctrl2) cell lines generated with an empty vector [20, 37]. As a positive control for the Trps1 binding, we used a previously identified fragment of the Dspp promoter [37]. The fold increase of the Dspp fragment in protein-DNA complexes immunoprecipitated with V5 antibodies from Trps1-OE cells over 17IIA11 control cells was used as a reference to evaluate the enrichment in DNA fragments from the SerpinB2 gene. A fragment of the Gapdh gene, which does not contain GATA consensus elements, served as a “non-binding” control. Immunoprecipitation of protein-DNA complexes using V5 antibodies, followed by qPCR, revealed enrichment of binding sites BS5 and BS7 in the DNA immunoprecipitated from cell lines expressing V5-Trps1 in comparison with the control cell lines (Fig. 7B, left panel). Interestingly, the Trps1 binding to BS5 site was detected only in Trps1-OE cells maintained in standard growth medium, but not in cells exposed to 5 mM P_i (P_i) for 24h (Fig. 7B). Also, the Trps1 binding to the BS7 site was decreased by P_i, however this effect was less prominent and was detected only in the Trps1-OE2 cell line. In contrast, Trps1 binding to BS6 was increased by elevated P_i levels in comparison with the standard growth conditions. ChIP experiments with control antibodies (normal mouse IgG) showed no differences for any of the analyzed DNA fragments (Supplementary Fig. S7A). In summary, we determined that Trps1 directly binds to GATA consensus sequences in the first intron of the SerpinB2 gene, and this binding is affected by the extracellular P_i levels.

Fig. 7.

Phosphate affects Trps1 binding to the SerpinB2 gene. (A) Approximate location of 8 putative Trps1 binding sites (BS; red bars) in the mouse SerpinB2 promoter and intron 1 identified using TRANSFAC database. Black bars indicate position of SerpinB2 exon 1 and 2. (B) Results of the ChIP assays expressed as fold enrichment of Trps1 binding sites in DNA-protein complexes immunoprecipitated with V5 antibody from Trps1-OE cells in comparison with control 17IIA11 cells. Left panel: results of ChIP assays performed in cells maintained in standard growth medium; right panel: results of ChIP assays performed in cells exposed to 5 mM P_i for 24h.

4. Discussion

SerpinB2/PAI-2 is an intriguing but enigmatic protein, which on one side, has been attributed a plethora of molecular and cellular functions, while, on the other side, its major physiological function remains to be unveiled. Although SerpinB2/PAI-2 has been often described as a component of the plasminogen system [1–6], functional assays, analyses of SerpinB2^−/− mice and clinical data implicate SerpinB2 in the inflammation, immune response and cancer [10–14, 63, 64]. Results of our studies expand the array of SerpinB2 functions by showing that SerpinB2 is also directly involved in the physiological mineralization process occurring in skeletal and dental tissues by regulating this function of osteogenic cells in a cell-autonomous manner. First, we determined that the SerpinB2 is highly and specifically expressed in skeletal and dental tissues (Fig. 5). We showed that upregulation of SerpinB2 expression coincides with chondrocytes maturation and the initiation of mineralization in vivo. Consistently with this expression pattern, in vitro experiments uncovered that SerpinB2 expression is activated by elevated P_i (Fig. 1), which is one of the major stimulators of mineralization. Importantly, this effect of P_i on expression of SerpinB2 is specific to osteogenic cells. We have also shown that SerpinB2 expression is directly regulated by another mineralization-related molecule, the Trps1 transcription factor. However, in contrast to the activation by P_i, SerpinB2 expression is repressed by Trps1 (Fig. 2 and 6). We have also uncovered that P_i changes the occupancy of Trps1 on cis-elements in the SerpinB2 gene, providing mechanistic insights into P_i-regulated gene expression (Fig.7). Finally, the involvement of SerpinB2 in the mineralization process was supported by analyses of consequences of SerpinB2 deficiency on the function of committed osteogenic cells, which demonstrated impaired mineralization of the SerpinB2-deficient cells (Fig. 4).

The combination of in vitro and in vivo approaches used in this study indicates that the expression of SerpinB2 in osteogenic cells contributes to proper physiological mineralization, hence SerpinB2 regulates this process in a cell-autonomous manner. First, the functional assays comparing formation of mineralized ECM by control and SerpinB2-deficient 17IIA11 committed osteogenic cells revealed delayed initiation of the deposition of mineral in the ECM, and reduced extent of the ECM mineralization (Fig. 4, Suppl. Fig. 4). Second, the IHC detection of high levels of SerpinB2 in all cell types producing mineralized ECM suggests that SerpinB2 role in mineralization is not limited to one cell type, but it is common to all skeletal and dental tissues that produce mineralized ECM (Fig. 5).

We detected SerpinB2 in both intracellular and ECM compartments of mineralizing tissues (Fig. 5, 6 and Suppl. Fig. 8). Previous reports have shown that intracellular and extracellular SerpinB2 isoforms execute different functions [13]. The intracellular 47kDa SerpinB2 isoform has a cytoprotective role and it is involved in the protein folding [19, 58, 60], while the extracellular 60kDa isoform participates in ECM remodeling [3, 12]. High levels of SerpinB2 protein in predentin (a counterpart of bone osteoid) and unmineralized enamel matrix (Fig. 5B, Suppl. Fig. 8) suggest that this serine protease inhibitor is involved in remodeling of ECM components to allow for proper deposition of hydroxyapatite. For example, serine proteinase KLK-4 is strongly expressed by ameloblasts during the maturation stage of amelogenesis [65], when the enamel organic matrix is completely degraded and replaced by the mineral. Although, no studies of SerpinB2 KO enamel have been conducted to date, one can surmise that the role of SerpinB2 in the enamel matrix is to control the rate of proteolysis by KLK-4, which is critical for the proper enamel maturation.

It is likely that the role of SerpinB2 in cells producing mineralized ECM is not limited to ECM remodeling, as high levels of SerpinB2 were also detected within a cell body of hypertrophic chondrocytes, osteoblasts lining trabecular and alveolar bone, odontoblasts and ameloblasts (Fig. 5 and 6). Studies in neurons have demonstrated that intracellular SerpinB2 interacts with misfolded proteins and components of the ubiquitin-proteasome system [19]. Since production of ECM in skeletal and dental tissues involves high protein synthesis and upregulation of secretory machinery [66], the intracellular SerpinB2 detected in cells producing mineralized ECM may participate in proper folding and secretion of organic components of ECM. However, SerpinB2 deficiency does not impair survival of osteogenic cells stimulated for production of mineralized ECM in our in vitro analyses (Supplementary Fig. S3). These results, together with the lack of deleterious skeletal and dental defects in SerpinB2^−/− mice [10], suggest that this protein is not essential for proper formation of mineralized tissues. However, it is important to note that these tissues have not been analyzed in SerpinB2^−/− mice. Hence, it is possible that these mice exhibit less apparent phenotype or they develop exacerbated phenotype under pathological conditions or have an impairment of the tissue repair process. This hypothesis is supported by clinical studies implicating SERPINB2 in periodontal disease [67, 68], as the formation of unsound mineralized tissues of the dento-alveolar complex (bone and cementum) is a strong contributing factor to this chronic inflammatory disease.

Development and homeostasis of mineralized tissues, including the dento-alveolar complex, is highly dependent on P_i [23, 27, 69–72]. This signaling molecule stimulates osteogenic cells, which, in response, form mineralized ECM. This is accomplished, in part, by changes (activation or repression) in expression of many genes involved in formation of mineralized ECM [27]. Results of our studies suggest that upregulation of SerpinB2 by P_i is an integral component of the mineralization program activated in osteogenic cells in response to this signaling molecule. However, the mechanism, by which P_i regulates the transcription is largely unknown, and the mechanism of P_i-induced SerpinB2 expression in osteogenic cells remains to be elucidated. Thus far, only AP-1 transcription factors have been suggested to mediate transcriptional response to P_i [36]. The same study revealed that promoters of many mineralization-related genes that are upregulated by P_i are enriched in GATA consensus sequences. Here, we identified Trps1 as a transcription factor that might be directly involved in regulation of gene expression in response to P_i in osteogenic cells. Using ChIP we detected direct Trps1 binding to the GATA consensus elements in the SerpinB2 first intron. Interestingly, the direct interaction of Trps1 with SerpinB2 regulatory elements was changed by elevated extracellular P_i (Fig. 7). This indicates that activation of P_i signaling leads to changes of transcription factors occupancy on their target promoters. Our results also uncovered that Trps1 is an active repressor of the SerpinB2 gene in osteogenic cells, as demonstrated by increased expression of SerpinB2 in Trps1-deficient 17IIA11 cells and decreased expression of SerpinB2 in Trps1-overexpressing osteogenic cells in vitro and in vivo (Fig. 2 and 6). These results together with the results of the ChIP assay (Fig. 7) suggest that Trps1 represses SerpinB2 through direct binding to its regulatory elements.

In our previous studies, we have detected decreased levels of P_i-stimulated proteins Dmp1 and Phex in cells overexpressing Trps1, although the mechanism has not been studied [20, 73]. Thus far, only a few mineralization-related genes have been demonstrated to be directly regulated by Trps1: Runx2, Bglap (osteocalcin) and Dspp [37, 74, 75]- none of these have been reported as a P_i-regulated gene. However, the high abundance of GATA elements in many mineralization-related genes regulated by P_i [36], strongly suggest that Trps1 is involved in the transcriptional regulation of gene expression in response to P_i. Of note, the FGF pathway, which has been implicated in P_i signaling [76, 77], has not been connected to Trps1 yet. Thus, the mechanisms of the Trps1-P_i signaling interaction remains to be elucidated.

In summary, this study identified the serine protease inhibitor SerpinB2/PAI-2 as a novel protein supporting mineralization. Furthermore, we have shown that P_i activates SerpinB2 expression in osteogenic cells, however the overexpression of Trps1 inhibits this effect (Fig. 8). To the best of our knowledge, SerpinB2 is the first example of P_i-stimulated gene that is also regulated by Trps1. Data presented here suggest a potential mechanism of transcriptional regulation of osteogenic cells physiology in response to P_i, whereby the presence of high levels of Trps1 prevents mineralization-related genes (those with GATA regulatory elements) from activation by P_i. Our future studies will investigate the interplay between P_i signaling and Trps1 transcription factor in the regulation of expression of mineralization-related genes as a potential tool for regulation of the mineralization process in pathological conditions.

Fig. 8.

A model of regulation of SerpinB2 expression by mineralization-regulating factors: (A) P_i signaling molecule activates SerpinB2 expression in osteogenic cells through yet unknown transcriptional mechanism. P_i-induced expression of SerpinB2 supports mineralization. (B) Trps1 transcription factor represses SerpinB2 expression through direct interaction with its regulatory elements. Diminished SerpinB2 expression in cells forming mineralized ECM impairs mineralization.

Highlights:

• SerpinB2 is highly expressed by cells producing mineralized ECM.

• SerpinB2 expression is activated by phosphate, while the Trps1 transcription factor represses SerpinB2 in committed osteogenic cells.

• SerpinB2 deficiency impairs mineralization in vitro.