Introduction

Osteoprotegerin (OPG), also known as osteoclastogenesis inhibitory factor (OCIF) or tumor necrosis factor receptor superfamily member 11b (TNFRSF11B), is a member of the tumor necrosis factor receptor (TNFR) family (Simonet et al. 1997; Yamaguchi et al. 1998). In contrast to most TNFR family members which are transmembrane receptors with functional cytoplasmic domains, OPG is a soluble secreted protein. It consists of approximately 401 amino acids and there is ∼85% identity between the mouse and human homologs. OPG has three major structural motifs: a four cysteine-rich TNF receptor domain (CRD) which is a characteristic motif among TNFR family members, a heparin-binding domain, and two death domain homologous regions (DDH), through which OPG is assembled as a homodimer (Yamaguchi et al. 1998).

There are two main known roles for OPG at the molecular level: inactivating the receptor activator of nuclear factor-кB ligand (RANKL) and the TNF-related apoptosis-inducing ligand (TRAIL) (Vitovski et al. 2007). As a decoy receptor, OPG binds to RANKL, whose functional receptor is the receptor activator of NF-кB (RANK), and inhibits the RANKL/RANK pathway. Nowadays, the RANKL/RANK/OPG system is believed to be one of the most important signaling axes regulating bone metabolism (Boyce and Xing 2008; Sattler et al. 2004; Theill et al. 2002). Among the systems, the importance of OPG has been demonstrated in OPG knockout mice which exhibited severe osteoporosis (Bucay et al. 1998; Mizuno et al. 1998). It has also been demonstrated that the OPG/RANKL ratio represents an important determinant of bone resorption in several disease models (Kearns et al. 2008). Additionally, it has been suggested that the RANKL/RANK/OPG system is associated with vascular calcification (Collin-Osdoby 2004; D’Amelio et al. 2009). By binding to RANK and subsequently increasing BMP4 production via the activation of the alternative NF-κB pathway, RANKL induced the deposition of a mineralized matrix produced by vascular smooth muscle cells (Panizo et al. 2009). OPG appears to inhibit this process. It has been shown that OPG knockout mice developed calcified lesions in the aorta and renal arteries (Bucay et al. 1998; Orita et al. 2007), which suggests that OPG plays a protective role through its prevention of vascular calcification. OPG also acts as a decoy receptor for TRAIL and blocks the initiation of apoptosis (Lane et al. 2012). It used to be believed that TRAIL selectively induces cancer cell apoptosis and has no influence on normal organs or tissues (Walczak et al. 1999). However, researches indicated that TRAIL may also induce apoptosis of normal human hepatocytes, cultured human microvascular endothelial cells, and kidney cells (Jo et al. 2000; Montanez-Barragan et al. 2014; Pritzker et al. 2004). As the natural inhibitor of TRAIL, OPG obviously protects the normal organs or tissues from TRAIL cytotoxicity. Indeed, abnormal OPG/TRAIL ratios have been shown to be associated with severe acute myocardial infarction and the early onset of diabetes mellitus (Secchiero et al. 2010; Vaccarezza et al. 2007). In another aspect, several tumor cells including prostate cancer cells, multiple myeloma cells, breast cancer cells, and colorectal cancer cells are also protected from TRAIL-induced apoptosis by OPG, which secreted by osteoblast-like cells and bone marrow stromal cells, as well as cancer cells themselves (Corey et al. 2005; De Toni et al. 2008; Neville-Webbe et al. 2004; Rachner et al. 2009; Shipman and Croucher 2003).

Nowadays, treatment with RANKL inhibitors presents a promising strategy for the treatment of various disease settings, e.g., osteoporosis, rheumatoid arthritis, cancer metastasis, and other metabolic bone diseases (Delmas 2008; Wada et al. 2006). As a natural RANKL inhibitor, OPG is also considered as a possible treatment for such diseases. A phase I clinical trial has shown that AMGN-0007, a derivative of OPG, can be an effective treatment for multiple myeloma and breast carcinoma-related bone metastases (Body et al. 2003). Not all parts of OPG are needed for drug development. Studies have revealed that the CRD domain (22–201) of OPG (OPG-CRD/OPGT) is responsible for the binding of RANKL (Luan et al. 2012; Nelson et al. 2012; Simonet et al. 1997). The effectiveness of OPGT as a neutralizing agent against RANKL has been verified with AMGN-0007, which consists of the 22–194 region of OPG and a C-terminal immunoglobulin Fc fragment (Body et al. 2003).

Due to the therapeutic potential of OPG and its derivatives, it is valuable to investigate approaches for OPG or OPGT protein preparation. To date, the most common approaches employ eukaryotic expression systems, such as the CHO cell expression system (Body et al. 2003; Simonet et al. 1997), baculovirus-infected insect cells (Luan et al. 2012; Nelson et al. 2012), and the yeast expression system (He et al. 2000; LIU et al. 2009). Reports from Amgen Incorporation mentioned that Fc-OPG, an OPG derivative, was prepared from Escherichia coli (Bekker et al. 2001; Lacey et al. 2012; Morony et al. 1999). However, there has been no detailed description of their preparation procedure to our knowledge. In this study, we introduce a sortase A-aided E. coli expression system to produce the functional OPGT. This sortase A-aided E. coli expression system may contribute to the commercial applications of functional OPGT, due to its advantages of large-scale production and cost reduction, or it may contribute to scientific researches for its convenient use in protein preparations in further experiments related to multiple mutations or modifications of this protein.

Materials and methods

Construction of plasmids

The complementary DNA (cDNA) encoding SrtA was amplified using PCR from group A streptococci (GAS) M1 strain (90–226) (Fan et al. 2014). The DNA sequence encoding the LPETG peptide (LPETGGG) was introduced onto the 3′ end of the SrtA (accession number: NC002737) gene using the designed primer (Fan et al. 2014). The PCR products were cloned into the Nco I and Sal I site of the pET28a vector and the constructed vector was named pET28a-SrtA. The cDNA encoding full-length human OPG (accession number: U94332) was obtained using RT-PCR from the messenger RNA (mRNA) of human peripheral blood mononuclear cells. The full-length OPG and its truncation, a DNA sequence encoding OPGT (residues 22–201 of OPG, OPG-CRD), were then cloned into the Nco I and Xho I sites of the pET28a vector. The OPGT cDNA was also cloned into the BamH I and Sal I sites of the pGEX-6p-1 vector, and the Sal I and Xho I sites of the constructed pET28a-SrtA vector. The proteins produced from these constructed vectors were OPG/OPGT, GST-OPGT fusion protein, and SrtA-LPETG-OPGT fusion protein. The cDNA encoding the truncation of human TRAIL (residues 115–281 of TRAIL, soluble TRAIL, sTRAIL) was obtained using RT-PCR from the mRNA of human peripheral blood mononuclear cells and cloned into the BamH I and Sal I sites of the pGEX-6p-1 vector. The expressed plasmid for GST-RANKL (encoding residues 159–361 of human RANKL) and RANK (encoding residues 159–361 of human RANK) were prepared previously (Liu et al. 2015; Zhang et al. 2009). All constructed plasmids were confirmed by DNA sequencing.

Expression and purification of proteins

All recombinant proteins were expressed in the E. coli strain BL21(DE3). Bacteria with the constructed pGEX-6p-1 vectors were cultured in LB medium containing 100 μg/ml ampicillin at 37 °C under constant agitation (200 rpm). When an OD600 of 0.4 was attained, isopropyl β-d-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM for TRAIL expression or 0.1 mM for RANKL and OPGT expression. The cultures were then agitated at 150 rpm for an additional 5 h at 25 °C for TRAIL expression or overnight at 20 °C for RANKL and OPGT expression. The cells from 2 l of LB medium were harvested by centrifugation and lysed with a lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 1 mM EDTA, 1 mM DDT, 1 mM lysozyme, 0.5% v/v Triton X-100, pH 8.0) by an ultrasonic cell crusher (Scientz Biotech, Ningbo, China) at a condition of 300 W, 3 s on and 5 s off, total time 23 min. The GST fusion proteins in the supernatant of the lysate were purified using affinity chromatography with Glutathione-Sepharose Fast Flow 4B beads (5 ml of beads for supernatant from 2 l of bacteria culture) according to the manufacturer’s instructions (GE Healthcare). The GST tags were cleaved overnight at 4 °C using the PreScission protease (PSP, GE Healthcare, 20 μg for GST fusion protein prepared from 2 l of bacteria culture) and the proteins were further purified using size-exclusion chromatography (Superdex 200, GE Healthcare) in 0.1 M Tris pH 7.0 and 50 mM NaCl.

Bacteria with the constructed pET28 vectors were cultured in LB medium containing 100 μg/ml kanamycin, at 37 °C under constant agitation (200 rpm). For the expression of all proteins (RANK, OPG, OPGT, and SrtA-OPGT), IPTG was added to a final concentration of 1 mM for an additional 5 h at 37 °C when OD600 of 0.6 was attained. The cells were harvested by centrifugation and the pellet was re-suspended in 100-ml ice-cold washing buffer (50 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% v/v Triton X-100, pH 8.0), and then disrupted and homogenized by sonication for 10 min (3 s on, 5 s off) using an Ultrasonic Cell Crusher and washed extensively with washing buffer. After repeated sonication and centrifugation, the inclusion bodies were dissolved at room temperature in 6 M guanidine hydrochloride, 50 mM Tris pH 8.5, 1 mM EDTA, 150 mM NaCl, and 10 mM DTT to produce a final protein concentration of 30 mg/ml (wet weight of the inclusion bodies). The refolding of recombinant proteins was performed by further dilution with 20 mM Na2HPO4 pH 7.3, 1 M l-arginine, 20% glycerol, 10 mM reduced glutathione, and 1 mM oxidized glutathione to a concentration of 10 mg/ml, followed by dialysis against 20 mM Na2HPO4 pH 7.3, 0.5 M l-arginine, and 10% glycerol for 12 h at 4 °C. Additional dialyses against 20 mM Na2HPO4 pH 7.3, 0.2 M l-arginine and 5% glycerol for 12 h at 4 °C were followed by two dialysis steps against 20 mM Na2HPO4 pH 7.3 for 12 h at 4 °C. After centrifugation at 20,000g for 10 min, the supernatant was collected and the refolded OPGT proteins, which with a designed His-tag on the C-terminus of OPGT, were further purified by the Ni-NTA affinity chromatography according to the manufacturer’s instructions (GE Healthcare). Briefly, the supernatant diluted in 100 ml of the purification buffer (50 mM NaH2PO4, pH 8.0, 0.5 M NaCl ) was mixed with the 5 ml of Ni-NTA affinity resins for 1 h at 4 °C under constant agitation (30 rpm). The resins were then collected by low-speed centrifugation (800g) for 5 min and washed with 20 ml of the purification buffer with 20 mM imidazole. After repeated centrifugation and washing of resins for three times, the binding proteins were then eluted by 20 ml of the purification buffer with 200 mM imidazole. After concentration by ultrafiltration, the eluted proteins were further purified using size-exclusion chromatography (Superdex 200, GE Healthcare) at the flow rate of 0.5 ml/min in 0.1 M Tris pH 7.0 and 50 mM NaCl.

MALDI-TOF mass spectrometer

MALDI-TOF mass spectrometer was used to characterize the sample prepared by the size-exclusion chromatography. In our study, MALDI analysis was performed on a Bruker Autoflex time-of-flight mass spectrometer. The instrument was equipped with a delayed ion-extraction device and a pulsed nitrogen laser operated at 337 nm, and its available accelerating potential was in the range of 20 kV. The MALDI uses a ground-steel sample target with 384 spots. The range of laser energy was adjusted to slightly above the threshold in order to obtain good resolution and good signal-to-noise ratio. Sinapinic acid (SA) was used as the matrix for MALDI analysis. We obtained the mass spectra reported in the positive-ion linear mode with delayed extraction for 50 ns and external mass calibration by using two points that bracketed the mass range of interest. Each mass spectrum was typically summed with 50 laser shots.

Surface plasmon resonance

The binding affinities of OPGT to RANKL and TRAIL were determined by surface plasmon resonance (SPR) using BIAcore 3000 at 25 °C. The RANKL or TRAIL was immobilized to ∼2000 response units on one channel of a CM5 sensor chip in 10 mM sodium acetate, pH 5.5 using the standard amine coupling method (Kim et al. 2003). Recombinant TNFRSF9 was immobilized on another channel as a reference. Different concentrations of OPGT (0, 3.125, 6.25, 12.5, 25, 50, and 100 nM) in 10 mM HEPES, 150 mM NaCl, 0.005% Tween 20, pH 7.4 were injected into both channels and the signals were recorded as sensorgrams. Sensorgrams were fitted into the 1:1 binding model using BIA evaluation software 4.1 (Biacore, GE Healthcare), and the equilibrium-dissociation constants (KD) were calculated.

Osteoclast differentiation assay

The osteoclast differentiation assay was performed according to methods we reported previously (Liu et al. 2010). Briefly, the murine monocytic cell line RAW264.7 was cultured in 24-well plates (1 × 104 cells/well) and maintained in α-MEM containing 10% (v/v) FBS. Then, 50 ng/ml of RANKL was added to the cell cultures to induce the differentiation of RAW264.7 cells into osteoclasts. Varying concentrations of OPGT were added to inhibit the RANKL-induced differentiation effect. After 4 days, the cells were fixed and stained using the Acid Phosphatase, Leukocyte (TRAP) Kit (Sigma-Aldrich, 387A) according to the manufacturer’s instructions. The TRAP-positive, multinucleated (>3 cells per well) cells were counted under a light microscope using methods described previously (Polek et al. 2003). Aliquots of cells were used to quantify TRAP activity using methods described previously (Zhao et al. 2013). Briefly, the cells were washed once with PBS and lysed in 100 μl of cold lysis buffer (90 mM citrate buffer, pH 4.8, 0.1% Triton X-100 containing 80 mM sodium tartrate) for 10 min. After lysis, 100 μl of substrate solution (20 mM p-nitrophenyl phosphate in the above lysis buffer) was added and incubated for an additional 5 min. The reaction was terminated by adding 40 μl of 0.5 M NaOH. The optical density was read at a wavelength of 405 nm. Each experiment was repeated four times for statistical analysis.

Apoptosis assays

TRAIL-mediated apoptosis was assessed using sub-G1 DNA content analysis with a slight modification (Allen et al. 2012). Briefly, colo205 cells were exposed to 100 ng/ml of sTRAIL and varying concentrations of OPGT for 3 h. Cells were harvested, washed, re-suspended with phosphate-buffered saline (PBS pH 7.4) and fixed in 80% ethanol on ice. After being washed with PBS and re-suspended with PBS containing 0.2% Triton X-100 and 0.5 mg/ml of RNase A, cells were stained with 25 μg/ml propidium iodide and then subjected to analytic flow cytometry using a FACSort (BD Biosciences, Mountain View, CA).

Results

Expression of OPG using pET28a and pGEX-6p-1 expression system

For the preparation of functional OPG, we first expressed the full-length OPG, which fused with a C-terminal His-tag for affinity purification by NTA chromatography, using the E. coli strain BL21(DE3) containing the constructed pET28a-OPG vector (Fig. 1a). The results showed that although the RANK protein which was used as a positive control for the expression system was quite well expressed, there was no obvious expression of OPG protein (Fig. 1c). We were also unable to obtain any soluble OPG protein after purification from cell lysate using NTA affinity chromatography and the inclusion body amount was insufficient for next-step purification (data not shown). Because the CRD domain (22–201) of OPG (OPG-CRD, OPGT) was reported to be sufficient in binding and neutralizing RANKL (Luan et al. 2012; Nelson et al. 2012; Simonet et al. 1997), we then tried to prepare OPGT (OPG-CRD) instead of full-length OPG using the E. coli expression system. First, the OPGT sequence was inserted into the pET-28a vector (Fig. 1b). The results showed that there was still no obvious expression of the OPGT protein (Fig. 1c), and still no soluble OPGT protein and almost no inclusion body obtained. The OPGT sequence was then inserted into the pGEX-6P vector, in which a GST sequence was linked to the upstream region of the OPGT sequence (Fig. 1d). The fusion protein, GST-OPGT, was successfully expressed in BL21(DE3) by the induction of IPTG (Fig. 1e). However, we were still unable to obtain soluble OPGT protein after the GST agarose purification and the following PSP proteolytic cleavage (lane of purified GST-OPGT, Fig. 1f). SDS-PAGE analysis revealed that the expressed GST-OPGT mainly existed as inclusion bodies in the precipitate (lane of precipitation of lysate, Fig. 1f).

Fig. 1
figure 1

Expression of OPG or OPGT using the pET-28a and pGEX-6p vector. a, b Schematic gene structure of constructed pET-28a-OPG and pET-28a-OPGT. c Expression of OPG and OPGT using the constructed pET-28a vector. Protein expression was induced with IPTG and cells were loaded onto 12% SDS-PAGE gel for analysis. The pET-28a-RANK was used as a positive control. d Schematic gene structure of the constructed pGEX-6p-OPGT. e Expression of GST-OPGT using the constructed pGEX-6p-OPGT vector. Protein expression was induced with IPTG at 20 °C and cells were loaded onto 12% SDS-PAGE gel for analysis. f SDS-PAGE analysis of the purified GST-OPGT fusion protein. After ultrasonication, centrifugation, and purification by affinity chromatography, the supernatant and precipitation of the lysate and the purified GST-OPGT fusion protein were analyzed by SDS-PAGE

Full size image

Expression and purification of OPGT using SrtA system

Although an abundant amount of GST-OPGT was found in the precipitate, due to the difficulty in refolding GST fusion protein, we abandoned the attempt to denature and refold the GST-OPGT protein. However, the largely increased production of OPGT by fusion with a GST moiety raised the possibility that the addition of N-terminal polypeptides may increase the expression of OPG proteins greatly. We then constructed a pET28a-SrtA-OPGT plasmid to prepare OPGT protein (Fig. 2a). Sortase A (SrtA) has been widely used as an in vitro tool to cleave or post-translationally modify engineered proteins. It selectively recognizes a LPXTG motif and cleaves in between the threonine and glycine (Navarre and Schneewind 1994). When a target protein is linked to its c-terminus by a LPXTG motif, it can cleave the linker region and release the target protein. In the constructed pET28a-SrtA-OPGT vector, a SrtA gene sequence and a DNA sequence of the LPETG linker (LPETGGG) were linked to upstream of the OPGT gene sequence. After IPTG-induced expression and subsequent cell lysis, the expressed protein was found as an inclusion body in the insoluble portion of the lysate. The result from the SDS-PAGE analysis showed that based on the molecular weight (the theoretical molecular weight of SrtA-OPGT is 41.250 kD, while the OPGT without SrtA is 21.880 kD), the expressed protein in inclusion body was mainly composed of OPGT that was separated from SrtA via the self-cleavage of the SrtA fusion protein during protein expression (Fig. 2b). After the dissolution of the inclusion bodies and refolding, the refolded protein was purified using size-exclusion chromatography with the collection of the 15.11-ml peak (Fig. 2c). The result of the SDS-PAGE analysis confirmed the existence and purity of the desired OPGT (Fig. 2d). The precise molecular weight of the prepared protein (21.888 kD) measured by the MALDI-TOF mass spectrometer confirmed the integrity of the desired OPGT, whose molecular weight is 21.880 kD (Fig. 2e). Using this expression system, the yield of purified recombinant OPGT was approximately 10 mg per liter of LB medium in our laboratory.

Fig. 2
figure 2

Expression and purification of OPGT using the pET-28a-SrtA vector. a Schematic gene structure of the constructed pET-28a-SrtA-LPETG-OPGT. b Analysis of the inclusion bodies by the SDS-PAGE gel. The cells were crushed by sonication and washed extensively with washing buffer. After centrifugation, the pellet was loaded onto 12% SDS-PAGE gel for analysis. c Purification of the refolded proteins using the size-exclusion chromatography. The inclusion bodies were dissolved in guanidine hydrochloride solution and dialyzed against the dialysis buffer. The refolded proteins were subjected to the size-exclusion chromatography (Superdex 200, GE Healthcare). d Identification of the purified OPGT. The two components collected from 14.50 to 15.20 ml (P1), and 15.20 to 15.90 ml (P2) of the size-exclusion chromatography were analyzed by 12% SDS-PAGE. e The precise molecular weight of the purified OPGT was determined by the MALDI-TOF mass spectrometer

Full size image

High affinities of OPGT in binding with RANKL and TRAIL

The affinity between OPGT and RANKL (or TRAIL) is an important index at the molecular level to evaluate the inhibition ability of OPGT to its targeted protein. A very effective method for detecting the affinity between molecules is surface plasmon resonance (SPR) (Nguyen et al. 2015). In this study, the affinity of the prepared OPGT to RANKL or TRAIL was measured using Biacore 3000, a highly sensitive SPR detecting system. RANKL and TRAIL were immobilized in different channels on a CM5 sensor chip. Recombinant TNFRSF9, a nonspecific protein control, was immobilized in another channel on the same chip. Different concentrations of OPGT were passed through the chip one at a time to record the interaction signals. As seen in Fig. 3, the average KD of OPGT for TRAIL was 4.39 × 10−9 M (Fig. 3a) and for RANKL was 1.19 × 10−9 M (Fig.3b).

Fig. 3
figure 3

The affinities of OPGT to TRAIL and RANKL. The affinities of TRAIL and hRANKL for binding to OPGT were measured by BIAcore analyses using a chelating CM5 sensor chip. Recombinant TNFRSF9 was used as a negative control. Varying concentrations of OPGT (0, 3.125, 6.25, 12.5, 25, 50, and 100 nM) were injected into channels and the signals were recorded as sensorgrams. The equilibrium-dissociation constants were calculated using BIAevaluation software 4.1

Full size image

E. coli purified OPGT could inhibit the differentiation of RAW264.7 into osteoclast

The differentiation assay of the murine monocytic cell line RAW264.7 into osteoclast was carried out to assess the ability of the prepared OPGT to inhibit RANKL-induced differentiation. Osteoclast differentiation and maturation induced by osteotropic factors including RANKL is characterized by the formation of giant multinucleated cells (MNCs) and the expression of tartrate-resistant acidic phosphatase (TRAP) in MNCs (Filgueira 2004; Simonet et al. 1997). The number of TRAP-positive MNCs and the level of TRAP activity, which usually are detected using histochemical methods for light microscopy or spectroscopy are used to estimate the degree of osteoclast differentiation and maturation (Miyamoto 2011; Simonet et al. 1997). In this study, as shown in Fig. 4a, b, there were no TRAP-positive cells or mature osteoclasts (TRAP-positive, multinuclear cells with diameter >100 μm) in the control group, which contained neither of the proteins. In the RANKL alone group, the human-derived RANKL used in this study successfully induced the formation of TRAP-positive cells and mature osteoclasts due to evolutionary conservation of RANKL protein. Accompanying the increase of molar ratio of OPGT to RANKL, less TRAP-positive cells were observed. The numbers of mature osteoclasts decreased and approached zero when the molar ratio of OPGT to RANKL was 5:1 or more. Using spectroscopic measurement with a wavelength of 405 nm, we further assessed the TRAP activity of the RAW264.7 cells cultured with RANKL and different concentrations of OPGT. The results showed that with increased concentrations of OPGT, the TRAP activity decreased (Fig. 4c). These results indicated that the prepared OPGT could inhibit RANKL-induced osteoclast differentiation.

Fig. 4
figure 4

The inhibition of osteoclastogenesis mediated by OPGT. a Different ratios of RANKL and OPGT were added to the RAW264.7 cell cultures. The concentration of hRANKL was 50 ng/ml and the concentrations of OPGT were 50, 125, 250, 500, and 1000 ng/ml. RAW264.7 cells without RANKL and OPGT served as the control. Scale bar, 300 μm. b The multinucleated, TRAP staining-positive, and large diameter (>100 μm) cells were counted. The bar represents the average of four independent experiments; the data are shown as the mean ± SD. d TRAP activity of the RAW264.7 cells cultured with RANKL and different concentrations of OPGT. All groups were cultured in α-MEM medium containing 50 ng/ml of RANKL. The curve represents the average of four independent experiments with SD bars

Full size image

E. coli purified OPGT could inhibit the TRAIL-induced apoptosis

We then used sub-G1 DNA content analysis to measure the inhibitory effects of the prepared OPGT on TRAIL-mediated colo205 cell apoptosis. Colo205 cells are highly sensitive to TRAIL and therefore are used widely in detections of TRAIL-induced apoptosis (Lippa et al. 2007). The sub-G1 DNA content analysis is also the classical method for measuring the modulation effects of agents on TRAIL-induced apoptosis (Nakata et al. 2004). As shown in Fig. 5a, b, after treating the colo205 cells with 100 ng/ml TRAIL for 3 h, the percentage of sub-G1 phase cells increased from approximately 5 to 40%. Relative to the treatment with TRAIL alone, the co-treatment of cells with TRAIL and different concentrations of OPGT resulted in decreased amounts of sub-G1 phase cells. This decrease was in a OPGT dose-dependent manner, thus indicating the inhibitory effects of the prepared OPGT on TRAIL.

Fig. 5
figure 5

Flow cytometric analysis of apoptotic cells. Colo205 cells were treated with sTRAIL (100 ng/ml) in either the absence or the presence of OPGT (molar ratios of OPGT to sTRAIL varying from 1:1 to 125:1) for 3 h. Apoptosis was analyzed as a sub-G1 fraction by fluorescence-activated cell sorting. b The percentages of the sub-G1 fraction were calculated. The bar represents the average of three independent experiments; the data are shown as the mean ± SD

Full size image

Discussion

OPG was discovered unexpectedly by researchers at Amgen in studies designed to find novel tumor necrosis factor receptor (TNFR)-related molecules that could interfere with TNF signaling (Boyce and Xing 2008). RANKL promotes osteoclast development through RANK activation, while OPG inhibits this process by competitively binding to RANKL. Its antagonistic activity makes this naturally occurring RANKL inhibitor a promising therapeutic candidate for the treatment of disorders related to the abnormal expression of RANKL.

Reports from Amgen Incorporation mentioned that Fc-OPG was prepared from E. coli (Bekker et al. 2001; Lacey et al. 2012; Morony et al. 1999); however, there has been no detailed description of their preparation procedure to our knowledge. On the contrary, almost all published reports employ eukaryotic expression systems in the preparations of OPG or its derivative. In 1997, Simonet WS et al. expressed human and mouse OPG using 293E and CHO cells (Simonet et al. 1997). Preparations of recombinant OPG from the pichia pastoris yeast strain, HEK293 cell, or COS-7 cell have also been reported in the following studies (Cundy et al. 2002; Emery et al. 1998; He et al. 2000; LIU et al. 2009; Truneh et al. 2000). In 2012, two research groups expressed OPG using baculovirus-infected insect cells and solved the structure of RANKL/OPG individually (Luan et al. 2012; Nelson et al. 2012). AMGN-0007, an alternative recombinant OPG-Fc developed by Amgen Incorporation in order to improve the half-life through the glycosylation of the Fc region, was also prepared using CHO cells (Body et al. 2003). Thus, in this study, we aimed to prepare OPG using the E. coli expression system. At the beginning stage, we used pET and pGEX vectors, which are the most common vectors used in the E. coli expression system, to express OPG and its truncation, OPGT. There was no obvious OPG or OPGT expression in bacteria with the pET28 vector. An abundant amount of GST-OPGT protein appeared in bacteria with the pGEX-6P vector. However, the expressed GST-OPGT existed mainly as inclusion bodies, and little soluble protein was obtained. This phenomenon indicated a possibility that it is difficult to express OPG or OPGT alone using the E. coli expression system for unknown reasons, and this possibility may account for why OPG and its derivative were rarely prepared using E. coli expression systems in the reports. The high yield produced using the pGEX-6P vector also suggested that the addition of an N-terminal polypeptide which is easier to express may contribute greatly to the expression of fused OPG proteins. Correspondingly, the OPG derivative Fc-OPG developed by the Amgen Incorporation using E. coli is this form of fusion protein (Morony et al. 1999).

Based on our supposition, we chose SrtA as the N-terminal fused polypeptide. SrtA is a type of transpeptidase produced by gram-positive bacteria such as Staphylococcus aureus (Cossart and Jonquieres 2000). It recognizes and specially cleaves the short recognition motif (LPXTG) on the target protein and has been widely used as an in vitro tool to cleave or post-translationally modify engineered proteins. In this study, we adopted one of its recognition sequences, LPETG, to link SrtA and OPGT to separate SrtA from OPGT through the self-cleavage of the fusion protein. The results showed that we successfully obtained an abundance of cleaved OPGT inclusion bodies. After denaturing, refolding, and purification, we obtained the soluble form of OPGT. It was reported that calcium ions (2 mM or more concentrations) are required for efficient SrtA catalysis (Ilangovan et al. 2001). However, in this study, although no additional calcium ion was added, the self-cleavage of SrtA fusion protein was nearly completed based on the result from Fig. 2b, indicating that the inherent calcium ions in the E. coli expression system may be sufficient for the self-cleavage of SrtA-OPGT fusion protein. This supposition is not in conflict with the previous report since there is still some activity of SrtA when incubating in a calcium-free buffer (Ilangovan et al. 2001). Whether the addition of calcium ions promotes the efficiency of self-cleavage or the product yield of OPGT will be investigated in future research.

Subsequently, the Biacore analyses showed that the affinity constants between OPGT and RANKL, and between OPGT and TRAIL were 1.19 × 10 and 4.39 × 10−9 M, respectively. In the previous reports using the baculovirus-infected insect system, the affinity constants of two types of OPG-CRD/OPGT (residues 22 to 186 and residues 22 to 197 of OPG) to RANKL were 6.5 × 10−10 M and 5.14 ± 2.99 × 10−9 M, respectively (Luan et al. 2012; Nelson et al. 2012). Therefore, the OPGT prepared in this study has a similar binding affinity for RANKL as those reported previously using the baculovirus-infected insect system. There are no reports on the affinity constant between OPGT and TRAIL. One report mentioned that the affinity constant between TRAIL and OPG-Fc (the bivalent form of OPG by fusion to the Fc domain) prepared from CHO cells is 3 × 10−9 M (Emery et al. 1998). Therefore, in light of typically enhanced effect of the fused Fc domain on the ligand-receptor binding, the affinity of OPGT prepared using E. coli expression systems in this study for TRAIL should be as strong as the affinity of those prepared from CHO cells. Subsequent cell experiments also confirmed the inhibitory effects of OPGT on RANKL-induced osteoclast differentiation and TRAIL-induced tumor cell apoptosis.

In summary, we successfully constructed an SrtA-aided E. coli expression system for preparing OPGT, a functional truncation of OPG. The prepared OPGT possessed a high affinity for RANKL and TRAIL, and similar to the wild-type OPG, it effectively blocked RANKL-induced osteoclast differentiation and TRAIL-induced tumor cell apoptosis. This study may provide valuable contribution to further studies investigating OPG mutations and modifications and may potentially benefit the commercial application of this protein.