Human Serine Protease HTRA1 Positively Regulates Osteogenesis of Human Bone Marrow-derived Mesenchymal Stem Cells and Mineralization of Differentiating Bone-forming Cells Through the Modulation of Extracellular Matrix Protein†‡§
André N. Tiaden,
Bone and Stem Cell Research Group, CABMM, Zurich, University of Zurich, Switzerland
Author contributions: A.N.T.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; M.B., A.M., F.W., G.B., and S.G.: collection and/or assembly of data; P.C.: data analysis and interpretation; M.E.: data analysis and interpretation and manuscript writing; P.J.R.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, and manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS August 3, 2012.
Mammalian high-temperature requirement serine protease A1 (HTRA1) is a secreted member of the trypsin family of serine proteases which can degrade a variety of bone matrix proteins and as such has been implicated in musculoskeletal development. In this study, we have investigated the role of HTRA1 in mesenchymal stem cell (MSC) osteogenesis and suggest a potential mechanism through which it controls matrix mineralization by differentiating bone-forming cells. Osteogenic induction resulted in a significant elevation in the expression and secretion of HTRA1 in MSCs isolated from human bone marrow-derived MSCs (hBMSCs), mouse adipose-derived stromal cells (mASCs), and mouse embryonic stem cells. Recombinant HTRA1 enhanced the osteogenesis of hBMSCs as evidenced by significant changes in several osteogenic markers including integrin-binding sialoprotein (IBSP), bone morphogenetic protein 5 (BMP5), and sclerostin, and promoted matrix mineralization in differentiating bone-forming osteoblasts. These stimulatory effects were not observed with proteolytically inactive HTRA1 and were abolished by small interfering RNA against HTRA1. Moreover, loss of HTRA1 function resulted in enhanced adipogenesis of hBMSCs. HTRA1 Immunofluorescence studies showed colocalization of HTRA1 with IBSP protein in osteogenic mASC spheroid cultures and was confirmed as being a newly identified HTRA1 substrate in cell cultures and in proteolytic enzyme assays. A role for HTRA1 in bone regeneration in vivo was also alluded to in bone fracture repair studies where HTRA1 was found localized predominantly to areas of new bone formation in association with IBSP. These data therefore implicate HTRA1 as having a central role in osteogenesis through modification of proteins within the extracellular matrix. STEM Cells2012;30:2271–2282
Mammalian high-temperature requirement serine protease A1 (HTRA1) belongs to a well-defined group of serine proteases originally identified in bacteria [1, 2]. They share many common features including a highly conserved trypsin-like serine protease domain and at least one Post synaptic density protein, Drosophila disc large tumor suppressor, and Zonula occludens-1 protein domain at the C terminus. To date, four HTRA1 family members have been identified and are termed HTRA1, HTRA2, HTRA3, and HTRA4. HTRA2 is the best characterized of the four and exists as a membrane protein primarily involved in mitochondrial quality control [2, 3]. Both HTRA1 and HTRA2 are primarily regarded as being key regulators of tumor development and subsequent malignancies [4–6], although a growing body of evidence now exists to suggest that HTRA1 may also play a central role in musculoskeletal development and disease through its proteolytic actions on proteins within the extracellular matrix (ECM) [7–10].
The skeletal ECM is a structurally dynamic scaffold that provides a well-organized framework for mineralization and orchestrates many of the cellular processes required for maintaining bone integrity. The bone tissue ECM comprises 70%–90% mineral and 10%–30% protein, the major proportion of which being glycoproteins such as collagen, fibronectin, and noncollagenous small integrin-binding ligand, N-linked glycoproteins (SIBLINGs) . A direct interaction between these proteins and bone progenitor cells is essential for the development of a mineralized matrix [12–14]. Furthermore, degradation of specific glycoproteins within the ECM by serine proteases is also thought to play a central role in controlling mineral deposition by osteoblasts .
Bone remodeling and regeneration is a strictly regulated process, being reliant on the activities of resident osteoblasts originating from bone marrow-derived mesenchymal stem cells (BMSCs) through the process of osteogenic differentiation . Osteogenesis is governed by a complex series of well-orchestrated events involving numerous different transcription factors, signaling pathways, and growth factors [17, 18]. The potential involvement of HTRA1 in osteogenesis and bone development has already been alluded to in studies examining the expression profile of HTRA1 in developing mouse embryos and adult mouse bone where it was identified in osteocytes and osteoblasts within the bone matrix [10, 19]. More recently, it was suggested that HTRA1 may actually play a negative role in bone and mineral development through its ability to inhibit osteoblastic differentiation in the mouse 2T3 cell line . However, no studies have yet sought to determine the influence of HTRA1 on the osteogenic differentiation potential of MSCs.
In this study, we examined the effects of recombinant HTRA1 on the osteogenesis of human BMSCs (hBMSCs) and matrix mineralization by differentiating bone-forming cells with an aim to establish a role for HTRA1 in bone formation. Our results demonstrate that HTRA1 is an essential requirement for osteogenesis and that its effects are mediated primarily through its proteolytic actions on ECM proteins.
MATERIALS AND METHODS
4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF) was from Sigma-Aldrich (Buchs, Switzerland, http://www.sigmaaldrich.com/). Recombinant integrin-binding sialoprotein (IBSP) was purchased from R&D Systems (Abingdon, U.K., http://www.rndsystems.com/). Monoclonal mouse anti-IBSP, C-terminal region (clone ID1.2) and polyclonal donkey anti-collagen type 1 were from LabForce (Nunningen, Switzerland, http://www.labforce.ch/site/site.asp), polyclonal rabbit anti-IBSP, N-terminal region was from Enzo Life Sciences (Lausen, Switzerland, http://www.enzolifesciences.com/), and polyclonal rabbit anti-HTRA1 was generated as previously described . The biotinylated polyclonal swine anti-rabbit IgG was from DAKO (Baar, Switzerland http://www.dako.com/ch/). All anti-IgG horseradish peroxidase (HRP)-conjugated and fluorescence-conjugated secondary antibodies were from Jackson ImmunoResearch (Suffolk, U.K., http://www.jireurope.com/).
Experiments were performed using 5-month-old senescence-accelerated resistant mice (SAMR1) (n = 27) (Harlan, Netherlands, http://www.harlan.com/). All animal research procedures were approved by the Animal Experimentation Committee of the Veterinary Office of the Canton of Zürich, Switzerland (Project License 140/2005 and 151/2010), and followed the guidelines of the Swiss Federal Veterinary Office for the use and care of laboratory animals.
Mouse adipose-derived stromal cells (mASCs) were isolated from the inguinal fat pads of SAMR1 mice (n = 4) and purity confirmed by fluorescence-activated cell sorting (Sca1, 98%; CD29, 100%; CD105, 42%; CD34, 5%; CD45, 0.5%) as previously reported . Mouse embryonic stem cells (mESCs) (E14 129/Ola) are a well-established cell line and were maintained as previously described [22, 23]. hBMSCs were obtained from Lonza (Verviers, Belgium, http://www.lonza.com/) and were confirmed as being positive for CD105, CD166, CD44, and CD26, and negative for CD14, CD34, and CD45 as stated by the manufacturer. Cells were maintained in normal growth medium consisting of Dulbecco's modified Eagle's medium-low glucose (with Glutamax) (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/), supplemented with 10% fetal bovine serum (Invitrogen AG), penicillin (50 units/ml), and streptomycin (50 μg/ml) (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/). Cells were used between passages 2 and 6 unless otherwise stated. In some cases, hBMSCs undergoing osteogenic differentiation were treated with recombinant active or inactive HTRA1 (5 μg/ml) for up to 18 days. For protease inhibition studies, AEBSF (20 μg/ml) was also included where indicated. For three-dimensional (3D)-spheroids, mASCs were cultured in 25 μl hanging drops in normal growth medium as described above, using Terasaki plates (VWR, Dietikon, Switzerland, https://ch.vwr.com/) at 2,500 cells per drop.
Induction and Analysis of Differentiation
Well-established differentiation protocols were used to induce and analyze either osteogenesis or adipogenesis in mASCs and E14 129/Ola cells  and hBMSCs . For differentiation assays, a starting density of 5,000 and 10,000 cells per square centimeter was used for human and mouse cells, respectively. For osteogenic 3D-spheroid preparations, mASCs were predifferentiated in two-dimensional (2D) cultures for 3 days and cultured in hanging drops containing osteogenic culture medium for a further 2, 4, or 6 days, after which time they were harvested and processed for histological analysis using previously described techniques . Mineralization induced by osteogenic differentiation was identified using Alizarin Red S (Sigma-Aldrich, Buchs, Switzerland, http://www.sigmaaldrich.com/). Adipocyte formation was confirmed by positive staining for oil red O (Sigma-Aldrich, Buchs, Switzerland, http://www.sigmaaldrich.com/). Differentiation markers specific to either osteogenesis or adipogenesis were also quantified by reverse-transcription polymerase chain reaction (qRT-PCR) using TaqMan Gene Expression Assays (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/) (Supporting Information Table S1). Total RNA was harvested from cells at given time points during differentiation, 0.5 μg total RNA was reverse-transcribed using Superscript II (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/), and an equivalent of 10 ng total RNA was applied as cDNA template in the successive qRT-PCR reaction using the StepOnePlus (LifeTechnologies, Zug, Switzerland, http://www.lifetechnologies.com/).
Recombinant HTRA1 Production
Purified recombinant active HTRA1 (HTRA1Δmac) and inactive HTRA1 (HTRA1ΔmacSA) were produced in Escherichia coli and purified using metal-affinity chromatography as previously described .
SDS-PAGE and Western Blotting
Protein was analyzed by SDS-PAGE using 4%–15% precast Tris-HCl gels (BioRad, Reinach, Switzerland, http://www3.bio-rad.com/) under reducing conditions and electroblotted onto PVDF membranes using the Trans-Blot Turbo blotting system (BioRad, Reinach, Switzerland, http://www3.bio-rad.com/). IBSP was identified using antibodies raised against either the N-terminal (1:800) or C-terminal (1:500) regions of human IBSP and detected using HRP-conjugated secondary antibodies (1:10,000) followed by incubation in SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, Lausanne, Switzerland, http://www.piercenet.com/) and exposure to x-ray film.
Cells in either 2D or 3D cultures were fixed in phosphate buffered saline (PBS) buffered formaldehyde (4%) or ice-cold methanol, blocked with normal goat serum (1:10) and incubated with polyclonal rabbit anti-HTRA1 (1:50), polyclonal donkey anti-collagen type 1 (1:50), or monoclonal mouse anti-IBSP (1:50) in PBS containing 1% bovine serum albumin (BSA) overnight at 4°C. For paraffin wax sections, staining reactions were performed at 37°C for 1 hour. In the case of double immunostaining procedures, samples were incubated with both polyclonal rabbit anti-HTRA1 (1:50) and monoclonal mouse anti-IBSP (1:50) using the conditions described above. Samples were then washed and incubated with either goat anti-rabbit-Cy3 (1:400), goat anti-donkey-Cy3 (1:400), or goat anti-mouse-Cy5 (1:400) for 1 hour and mounted in 4,6-diamidino-2-phenylindole containing mounting solution and images captured using the Leica DMI6000B automated inverted research microscope system (Leica Microsystems, Heerbrugg, Switzerland, http://www.leica-microsystems.com/).
Dewaxed paraffin sections (8 μm) were rehydrated and blocked with normal swine serum (Dako Baar, Switzerland, http://www.dako.com/ch/) for 30 minutes. Sections were then incubated with polyclonal rabbit anti-HTRA1 (1:50) for 1 hour at 37°C. Sections were then washed in PBS and incubated with biotinylated swine anti-rabbit IgG (1:500) for 1 hour at room temperature followed by washing and a further incubation for 30 minutes with Vectastain (Reactolab, Servion, Switzerland, http://www.vectorlabs.com/). Sections were then developed using 3,3′ diaminobenzidine tetrahydrochloride, counterstained with Harris' Hematoxylin and mounted in dibutylphthalate polystyrene xylene (DPX).
Proteolytic Enzyme Assays
Degradation of IBSP by HTRA1 was determined using methods previously described . Briefly, HTRA1 (45 nM) and IBSP (476 nM) were incubated in 50 mM Tris-HCl, pH 8.5, and 150 mM NaCl for up to 24 hours at 37°C. In some reactions, active HTRA1 was replaced by proteolytically inactive HTRA1. Proteins were separated on a 4%–15% SDS-PAGE gel and analyzed by Western blot as described above. The proteolytic activity of HTRA1 was also investigated in hBMSC cultures using immunofluorescence staining. Cells were incubated in osteogenic induction medium either without or with active or inactive HTRA1 (5 μg/ml), fixed in methanol and IBSP or collagen type 1 detected with specific antibodies using methods as described above.
HTRA1 Enzyme-Linked Immunosorbent Assay
The HTRA1 enzyme-linked immunosorbent assay (ELISA) was performed using HTRA1 specific monoclonal and polyclonal antibodies as previously described .
HTRA1 Small Interfering RNA
Specific knockdown of HTRA1 expression was performed with Silencer Select small interfering RNA (siRNA) oligos (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/), according to the manufacturer's protocol. hBMSCs (1 × 105 cells) were transfected with 100 nM HTRA1-specific (s11279, s11280) or negative control siRNA (Negative Control-1) using the NEON Transfection System (Life Technologies, Zug, Switzerland, http://www.lifetechnologies.com/). Following transfection, cells were seeded in cell culture plates with fresh growth medium (without antibiotics) and incubated for 24 hours at 37°C, 5% CO2. Medium was then replaced with either fresh growth medium or differentiation medium and total RNA and supernatants were harvested at selected time points for further analysis.
A rigidly stabilized, unilateral mid-diaphyseal osteotomy gap was created in the femurs of SAMR1 mice (n = 5–6 per time point) using a 0.22 mm gigli saw as previously described . The immunohistochemical and immunofluorescence techniques described above were used to detect both HTRA1 and IBSP protein in decalcified paraffin wax sections (8 μm) from mouse femora at various stages of repair (n = 3 per time point).
All statistical analyses were carried out using SPSS18.0 (SPSS Inc., Chicago, IL). Parametric analysis of normally distributed data was performed using the two-tailed unpaired Student's t test for comparison of two groups or one-way analysis of variance followed by Tukey's post hoc test for multiple group comparisons. The sample size used in each study was based upon the observed level of variation between individual experiments and the need for sufficient sample numbers to allow for accurate statistical analyses. In all cases, a p-value of <.05 was considered statistically significant, and all data were expressed as mean ± SD.
HTRA1 Production Is Upregulated During Osteogenic Induction of MSCs
Differentiation studies conducted using hBMSCs demonstrated a small but significant increase in the expression of HTRA1 during the early phases of osteogenic induction with a greater than 1.5-fold (p <.01) increase in gene expression being attained by day 14 (Fig. 1A). However, unlike HTRA1, expression levels of the closely related HTRA family member, HTRA3, decreased during osteogenesis and remained significantly downregulated throughout the course of the study. The early induction of HTRA1 gene expression was accompanied by significant increases in the expression of the well-known osteogenic markers runt-related transcription factor 2 (RUNX2), alkaline phosphatase (ALP), IBSP, and Collagen type 1A1 (Fig. 1A). The expression level of the late osteogenic marker secreted phosphoprotein 1/osteopontin (SPP1) was actually downregulated during early osteogenesis but became significantly increased during the late phase of osteogenesis at day 21, along with noticeable increases in mineralized matrix deposition beginning at day 10 as determined by Alizarin red staining (Fig. 1B). As HTRA1 functions primarily as a secreted protease, we also investigated the effect of osteogenic differentiation on HTRA1 protein secretion from hBMSCs using an HTRA1-specific ELISA and immunofluorescence staining. In accordance with increases in its gene expression, secreted HTRA1 protein levels were also elevated in both the supernatant (Fig. 1C) and in the cell matrices (Fig. 1D) of hBMSC cultures undergoing osteogenic differentiation. Clearly, therefore, induction of osteogenesis in hBMSCs has a major stimulatory effect on HTRA1 production resulting in both enhanced gene expression as well as protein secretion.
In order to investigate whether such effects were limited to adult human stem cells only, we extended these studies to include both mESCs and mASCs. HtrA1 expression was also significantly upregulated in a time-dependent manner in mESCs (Fig. 2A) and mASCs grown in either 2D- (Fig. 2B) or 3D- (Fig. 2C) culture systems in response to osteogenic stimuli. These increases in HtrA1 expression were associated with mineralization of differentiating bone-forming cells derived from 2D-culutres of either mESCs (Fig. 2D) or mASCs (Fig. 2E) as well as 3D-cultures composed of osteogenic mASCs (Fig. 2F). Furthermore, immunofluorescence staining of paraffin wax sections from mASC 3D-spheroids identified HtrA1 protein in tissues undergoing mineralization, but not in uninduced controls (Fig. 2G), thus giving a first hint as to its role in stem cell osteogenesis.
Silencing of the HTRA1 Gene Impairs hBMSC Osteogenesis and Enhances Adipogenesis
In order to investigate the role of HTRA1 gene expression in the regulation of osteogenesis, we next performed loss of function studies by means of RNA interference in hBMSC cultures. The transfection of hBMSCs with siRNA specific for HTRA1 prior to osteogenic induction resulted in a significant downregulation of HTRA1 gene expression (Fig. 3A) and HTRA1 protein secretion (Fig. 3B) that was sustained throughout the first 2 weeks of osteogenic differentiation. HTRA1 protein levels both within cells and the ECM were also markedly reduced (Supporting Information Fig. S1). This was accompanied by a noticeable reduction in Alizarin red staining at day 14, indicative of reduced mineral deposition in differentiating bone-forming cells (Fig. 3C). Biochemical analysis using the ALP activity assay revealed that these deficits in mineralization were accompanied by significant reductions in ALP protein activity, which were maintained for up to 14 days (Fig. 3D). Furthermore, ALP expression was also significantly downregulated in these cells at an early time point, although expression levels of the osteogenic markers SPP1 and RUNX2 were not significantly affected by HTRA1 silencing (Fig. 3E).
Through the course of our studies, it became apparent that not only was osteogenesis reduced in hBMSCs following HTRA1 gene silencing but also that the adipogenesis appeared to actually be increased. Evidence of this was first realized in late-stage osteogenic cultures of hBMSC treated with HTRA1 siRNA, where phase-contrast microscopy revealed more cells containing lipid droplets, reminiscent of mature adipocytes although control siRNA cultures continued to demonstrate matrix mineralization (Fig. 3F). In order to confirm this finding, hBMSCs transfected with either HTRA1 siRNA or control siRNA were induced to undergo adipogenic differentiation and adipogenic gene expression measured by qRT-PCR (Fig. 3G) and triglyceride accumulation visualized by oil red O staining (Fig. 3H). As anticipated, silencing of the HTRA1 gene greatly enhanced the number of oil red O-positive cells present with the hBMSC culture and significantly enhanced the expression of specific adipogenic gene markers, peroxisome proliferator-activated receptor gamma 2, and fatty acid binding protein 4, thereby supporting a differential role for HTRA1 in hBMSC lineage commitment.
Recombinant HTRA1 Protein Enhances hBMSC Osteogenesis
As previously mentioned, HTRA1 also exists as a secreted protease and thus may have an extracellular role in the osteogenic differentiation of hBMSCs. We therefore next investigated whether exogenously added recombinant human HTRA1 protein could also influence osteogenic differentiation. hBMSCs undergoing osteogenic differentiation were initially treated every 3 days with either active or inactive HTRA1 for up to 14 days and mineralization determined by Alizarin red staining. Indeed, mineral deposition by differentiating bone-forming cells was greatly enhanced following stimulation with active, but not inactive, HTRA1 (Fig. 4B). Additional studies confirmed that these effects could be reproduced in late-stage (day 10) cultures of osteogenic hBMSCs treated for only 4 days with HTRA1 (Fig. 4C). However, short-term treatment with HTRA1 during the first week of osteogenic differentiation only failed to elicit any noticeable increase in mineralization (Supporting Information Fig. S2), thus implying that the stimulatory actions of HTRA1 were dependent not only on its protease activity but also on the differentiation status of the cells. Further studies using the broad spectrum serine protease inhibitor, AEBSF, confirmed the stimulatory effects of HTRA1 to be dependent on its proteolytic activity (Fig. 4D). Although matrix mineralization was clearly enhanced in differentiating bone-forming cells treated with HTRA1, the expression of several well-known osteogenic markers including ALP, RUNX2, and SPP1 remained unaltered at the time point tested (day 10) (Fig. 4E). However, significant increases were observed in the expression levels of IBSP and bone morphogenetic protein 5 (BMP5). Furthermore, a significant decrease in the expression level of sclerostin (SOST) was also observed in cells treated with HTRA1.
IBSP Represents a Novel HTRA1 Substrate During Osteogenic Differentiation of MSCs In Vitro
In this study, we demonstrated that although IBSP expression was significantly enhanced in hBMSCs following the addition of HTRA1, IBSP protein was almost completely abolished from the cellular matrix of these cultures, an effect not observed with inactive HTRA1 (Fig. 5A). This was regarded as being specifically due to the HTRA1 protein added due to the fact that collagen Type 1, a proteoglycan not degraded by HTRA1 , remained unaffected within the cell matrices (Supporting Information Fig. S3). The potential for proteolytically active HTRA1 to degrade IBSP was further confirmed by in vitro enzyme assays, where a noticeable reduction in recombinant human IBSP (476 nM) was evident after 8 hours incubation with HTRA1 (45 nM) as determined by Western blot analysis (Fig. 5B). We next used the mASC spheroid culture system to investigate whether endogenous HTRA1 protein could be localized to IBSP directly within differentiating MSCs. Paraffin wax sections of mASCs were stained for HTRA1 and IBSP at various stages of osteogenic differentiation and overlays produced to determine whether HTRA1 could be localized to areas of IBSP protein expression. IBSP was detected within mASC spheroids during the early phase of osteogenesis (day 5), although HTRA1 remained at very low levels and mineral deposition was not detected using Alizarin red staining (Fig. 5C, upper panel and Supporting Information Fig. S4A). By day 9, levels of HTRA1 had greatly increased and it was regularly found colocalized with IBSP in the tissue matrix (Fig. 5C, lower panel). In addition, mineralization of differentiating bone-forming cells within 3D-spheroids was also apparent at this time point as identified by intense Alizarin red staining (Supporting Information Fig. S4B). Sections incubated with isotype-matched IgG antibodies stained negative and served as controls (Supporting Information Fig. S4C).
Localization of HTRA1 and IBSP to Areas of New Bone Formation in Mice
In order to assess the potential involvement of HTRA1 in the reparative process of bone tissue, we used a previously well-established mouse femur osteotomy model (Fig. 6A) . HTRA1 protein was identified in thin paraffin wax bone sections taken from mice at various stages of fracture repair using immunohistochemical staining. At day 7, early osteoid formation was observed within the bone marrow at the osteotomy site and was associated with intense staining for HTRA1 (Fig. 6B). At this time point, HTRA1 was also located within the cells and tissue of the overlying periosteal layer (Supporting Information Fig. S5A). By day 14, the nonmineralized callus had increased greatly in size with numerous HTRA1-positive fibroblast-like cells present throughout the tissue as well as in cells of the newly forming callus (Fig. 6C and Supporting Information Fig. S5B). Large numbers of HTRA1-positive chondrocytes were also evident within intact lacunae and represented the early stages of endochondral ossification. By day 28, a substantial proportion of the callus had been replaced by new bone and HTRA1 was now localized to specific areas of active bone formation (Fig. 6D) in association with cuboidal osteoblasts (Supporting Information Fig. S5C). By day 42, the osteotomy gap was almost completely healed (Fig. 6E) and the bone resembled that of a nonoperative control femur (Fig. 6F). In both cases, HTRA1 was almost completely absent, being localized to only a small number of osteocytes within the bone matrix.
Having confirmed that HTRA1 protein was indeed present within actively forming bone tissue, we conducted further investigations to determine whether HTRA1 could also be localized to IBSP within these bone sections, thus providing a potential mode of action for HTRA1 in the context of osteogenesis in vivo. Indeed, colocalization of HTRA1 and IBSP was identified in paraffin wax bone sections from 28-day postoperative mice by double immunofluorescence staining, being confined to areas previously identified as undergoing active regeneration (Fig. 7 and Supporting Information Fig. S6).
A complex molecular network of signaling pathways and regulatory factors governs the osteogenesis of multipotent MSCs [17, 18, 27], being reliant on the activation and regulation of a number of key molecular targets including RUNX2, ALP, SPP1, and IBSP . Although there have been significant advances in our understanding of the molecular mechanisms involved in controlling osteogenic differentiation, the influence of serine proteases on such processes remains relatively obscure. In this study, we have shown for the first time that the serine protease HTRA1 is a positive regulator of MSC osteogenesis and its presence is required for the efficient mineralization of differentiating bone cells.
The expression of HTRA1 was upregulated in hBMSCs upon osteogenic induction and was associated with significant increases in the levels of its secreted protein product. Furthermore, HTRA1 production was increased in a time-dependent manner throughout the course of osteogenic differentiation and was closely associated with the appearance of mineralized matrix. Interestingly, expression levels of the closely related HTRA family member, HTRA3, were actually downregulated during osteogenesis. This was of particular interest as HTRA1 and HTRA3 are considered to have overlapping functions due to close structural similarities . It may well be therefore that HTRA1 and HTRA3 play differential roles in mediating the osteogenesis of hBMSCs. This pattern of HTRA1 production during osteogenesis of hBMSCs was recapitulated in cultures composed of either mASCs or mESCs, where significant increases in HtrA1 expression coincided with the appearance of HtrA1 protein and mineralized matrix. Although previous investigations have alluded to the involvement of HTRA1 in mouse osteogenesis [19, 20], our study is the first to report the potential role of HTRA1 in human osteogenesis. Based on earlier reports of HTRA1 imparting a negative influence over mineral deposition by murine 2T3 osteoblasts , initial expectations were for HTRA1 to inhibit osteogenesis of hBMSCs and thus prevent mineralization of differentiating human bone-forming cells. On the contrary, we found that siRNA-induced loss of HTRA1 function in hBMSCs, effectively suppressed the stimulatory effects of osteogenic culture medium on ALP expression and intracellular ALP protein activity by hBMSCs as well as the formation of a mineralized matrix. Although ALP expression levels were compromised in cells lacking HTRA1, some of the key regulatory factors involved in osteogenesis, including RUNX2 and SPP1, remained intact. It is possible therefore that HTRA1 mediates its effects through specific pathways, independent of these particular elements. In addition to its effects on osteogenesis, HTRA1 silencing had a profound effect on hBMSC adipogenesis as evidenced by increases in adipogenic gene expression and oil droplet formation. Such observations would suggest that HTRA1 may in fact represent a decisive factor in determining stem cell lineage commitment and that its involvement in stem cell differentiation go somewhat beyond it simply being a mediator of osteoblast formation. Further investigations into the role of HTRA1 in other differentiation pathways, such as chondrogenesis, myogenesis, or even tenogenesis, would therefore represent a logical progression of these studies and may hold particular relevance with regards to the involvement of HTRA1 in musculoskeletal development as has previously been inferred .
Based on the fact that HTRA1 is a secreted serine protease, we next directed our attention to the possible influence of HTRA1 on hBMSC osteogenesis by way of its extracellular effects. A specific role for HTRA1 in the osteogenesis of hBMSCs and mineralization of differentiating bone-forming cells was confirmed in studies using both proteolytically active and inactive recombinant forms of human HTRA1 as well the broad spectrum serine protease inhibitor AEBSF. An increase in mineralized matrix formation was observed in cells treated with active HTRA1 only, and was dependent on HTRA1 being present during the initiation phase of mineralization, as short duration treatments prior to the onset of mineralization proved ineffective. This would therefore infer that the stimulatory effects of exogenously added HTRA1 protein on hBMSC osteogenesis and mineralization of differentiating bone-forming cells may be reliant on its ability to regulate factors present within the actively mineralizing ECM. In addition to its ability to affect ECM mineral content, HTRA1 also induced a significant upregulation in the expression of several positive regulators of mineralization, including BMP5 and IBSP [30, 31], although a large proportion of well-known osteogenic genes such as RUNX2, ALP, and SPP1 remained unaffected. In addition, the expression level of a potent negative-regulator of mineral formation, SOST [32, 33], was markedly downregulated in hBMSCs treated with HTRA1 and may thus represent an additional mechanism by which mineral deposition is enhanced in differentiating bone-forming cells. It is possible, therefore, that the effects of HTRA1 protein on hBMSC mineralization may be mediated indirectly through its ability to modulate the production of both activators and inhibitors of osteogenic differentiation and/or mineralization. These results are in stark contrast to the findings presented in a previous report by Hadfield et al. , where HTRA1 was shown to impart a negative influence over mineralized matrix formation by mouse 2T3 osteoblasts. One possible explanation may lie within the fact that 2T3 cells are derived from transgenic mice overexpressing the simian virus 40 (SV40) T antigen and as such are immortalized . The fact that HTRA1 gene expression is known to be significantly affected in SV40 transformed cells  would imply that modifications of the osteogenic status of these cells brought about by changes in HTRA1 expression and activity may not be truly representative of nonimmortalized, primary cells. Further studies aimed at determining the effects of HtrA1 gene silencing or exogenously added recombinant HTRA1 on primary mASC osteogenesis may help to clarify this point.
Of the osteogenic genes that were regulated by HTRA1, IBSP held particular relevance to this study due to the fact that its ability to regulate mineral deposition has previously been reported to be dependent on the actions of serine proteases . Attention was therefore focused on IBSP and its potential involvement in mediating the effects of HTRA1 on both osteogenesis of MSCs and mineralization of differentiating bone-forming cells. IBSP belongs to the family of SIBLING proteins which also includes osteopontin, dentin matrix protein 1, dentin siaolophosphoprotein, and matrix extracellular phosphoglycoprotein . IBSP, along with other SIBLING proteins, is found almost exclusively in bone and dentin and is a major constituent of the mineralizing matrix during new bone formation [37–39]. The primary biological function of IBSP is unclear, although it has been reported to be involved in the initiation of matrix mineralization through the nucleation of hydroxyapatite crystals [40, 41]. Furthermore, it has been proposed that IBSP is required to undergo fragmentation, possibly through the actions of serine proteases, in order to allow for efficient matrix mineralization [15, 42, 43]. In this report, we were able to demonstrate colocalization of HTRA1 and IBSP in osteo-induced mASC spheroid cultures, thus confirming the potential for endogenous HTRA1 to interact with IBSP under osteogenic culture conditions. Additional immunofluorescence analyses of hBMSCs treated with exogenous HTRA1 revealed that this interaction most likely resulted in the digestion and subsequent removal of IBSP protein from the mineralizing matrix. These results would therefore suggest that the stimulatory effects of secreted HTRA1 protein on matrix mineralization in vitro are mediated, at least in part, through its proteolytic actions on IBSP within the ECM. The potential for HTRA1-induced IBSP proteolysis was further confirmed by Western blot analysis where proteolytically active HTRA1 effectively digested recombinant human IBSP protein at equivalent concentrations to those used in the cell culture studies. However, despite us using two different antibodies raised against either the N- or C-terminal regions of IBSP, we were unable to detect any HTRA1 digest fragments. This was somewhat unexpected, as findings from previous studies suggest that IBSP fragmentation plays an important role in its function as an instigator of mineral nodule formation . This is based on the assumption that IBSP facilitates mineralization predominantly through its ability to bind and nucleate hydroxyapatite, although evidence exists to suggest that both intact and fragmented IBSP can also inhibit hydroxyapatite-seeded crystal growth . It is possible therefore that HTRA1 contributes to osteogenesis through the controlled turnover of IBSP within the mineralizing ECM. This is further supported by the observation that IBSP gene expression is upregulated by exogenous HTRA1 at the expense of its protein product and may thus constitute a novel feedback mechanism through which HTRA1 regulates matrix mineralization.
The osteogenic differentiation of BMSCs toward mature bone-forming osteoblasts is a critical step in the development and repair of bone tissue. In this study, we used a previously well-established mouse osteotomy model to investigate the expression of HTRA1 protein during bone healing. Immunohistochemical analysis identified high levels of HTRA1 protein within areas of active bone formation at various stages of bone repair although no positive staining was observed in tissue sections from intact bone, thus confirming it to be primarily involved in de novo bone formation. Furthermore, strong positive staining for HTRA1 was also evident in callus tissue harboring large numbers of chondrocytes and is thus suggestive of its role in both chondrogenesis and endochondral ossification. Such findings could be of significant importance to fracture healing and possibly even cartilage formation, and as such, warrant further investigation. Immunofluorescence analysis of these sections revealed a close association of HTRA1 with IBSP in areas considered to be undergoing active bone regeneration. This lends support to the theory that HTRA1 regulates both osteogenesis of MSCs and mineralization of differentiating bone-forming cells through its interactions with IBSP, although it does not exclude the possibility that HTRA1 mediates its actions through the proteolytic processing of other SIBLING members [45–48].
The findings from this study therefore implicate HTRA1 as a positive regulator of both osteogenesis of MSCs and mineralization of differentiating bone-forming cells, possibly at the expense of adipogenic differentiation. Based on these properties, HTRA1 may be deemed as a key factor in determining the outcome of diseases such as age-related osteoporosis, where impaired BMSC osteogenesis and upregulated adipogenesis is believed to be a major contributor to the underlying pathology [24, 49].
This study was supported in part by the Swiss National Science Foundation Grant 31003A-134935; CABMM Start-up Grant; Forschungskredit of the University of Zurich; Novartis Foundation, formerly Ciba-Geigy-Jubilee-Foundation; and Uniscientia Foundation.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.