SEARCH

SEARCH BY CITATION

Keywords:

  • extracellular matrix proteins;
  • HtrA1;
  • HtrA3;
  • serine proteases;
  • TGF-β signaling

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The expression of mouse HtrA1 is developmentally regulated and restricted in embryo tissues which depend largely on TGF-β signaling for their differentiation. We examined whether mouse HtrA3, another HtrA family member very close to HtrA1, shows similar expression patterns. HtrA3 and -1 were expressed mostly in the same embryonic organs but exhibited complementary patterns in various tissues; the lens epithelial cells in day 12.5 embryo expressed HtrA3 whereas the ciliary body and pigment retina expressed HtrA1. In the vertebrae of day 14.5 embryo, HtrA3 was expressed in the tail region, but HtrA1 was predominantly expressed in the thoracic and lumbar regions.

Similar to HtrA1, HtrA3 bound to various TGF-β proteins and inhibited the signaling of BMP-4, -2 and TGF-β1. HtrA3 did not inhibit signaling originated from a constitutively active BMP receptor, indicating that the inhibition occurred upstream of the cell surface receptor. HtrA3 also showed proteolytic activities indistinguishable from those of HtrA1 toward β-casein and some extracellular matrix (ECM) proteoglycans. The protease activity was absolutely required for the TGF-β signal inhibition activity.

All these data suggest that HtrA3 and -1 have the overlapping biological activities but can function in complementary fashion in certain types of tissues.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

HtrA is a highly conserved family of serine proteases found in species ranging from bacteria to human. The defining feature of the HtrA family is the combination of a catalytic domain with at least one C-terminal PDZ domain (Pallen & Wren 1997). HtrA was initially identified in E. coli by two phenotypes of null mutants. Mutants either did not grow at elevated temperatures (HtrA, high temperature requirement) or failed to degrade misfolded proteins in the periplasm (DegP). It was later shown that bacterial HtrA had a molecular chaperon activity at low temperatures and a serine protease activity that degraded misfolded proteins at high temperatures (Spiess et al. 1999).

Mammals have four HtrA genes; HtrA1, -2, -3 and -4. HtrA2 contains a mitochondrial localization signal and a transmembrane region in the N-terminal region and localized in the intermembrane space of mitochondria. HtrA2 was originally described as an apoptosis inducer that binds to inhibitor of apoptosis proteins (IAPs), thereby suppressing the caspase-inhibitory activity of IAPs (Suzuki et al. 2001; Martins et al. 2002). Recently, it was reported that mutation of mouse HtrA2 caused a neurodegenerative disease due to progressive mitochondrial damage (Vaux & Silke 2003). The primary function of HtrA2 therefore seems to handle misfolded proteins in mitochondria.

Structure of N-terminal regions of mammalian HtrA1, 3 and 4 are distinct from that of HtrA2 (Clausen et al. 2002). They commonly contain secretory signals at the N-terminus followed by two domains; one similar to the insulin-like growth factor (IGF) binding protein and the other to the Kazal-type serine protease inhibitor. In contrast with HtrA2, the precise function of HtrA1, -3 and -4 are less understood. HtrA1 was initially identified as a gene which was down-regulated in SV40-transformed human fibroblasts (Zumbrunn & Trueb 1996). The expression of HtrA1 was also decreased in progression and invasion of ovarian cancers and melanomas (Baldi et al. 2002; Shridhar et al. 2002). HtrA1 thus may function as a tumor suppressor gene. Another interesting aspect of human HtrA1 may be its possible involvement in the pathogenesis of osteoarthritis. Hu et al. reported that the content of HtrA1 protein was increased in the joint cartilage of human osteoarthritis patients (Hu et al. 1998).

We have investigated the expression pattern of mouse HtrA1 during embryogenesis in detail and found that HtrA1 was characteristically expressed in a distinct set of embryo tissues where the development was largely regulated by TGF-β family proteins (Oka et al. 2004). For instance, HtrA1 is expressed in skeletal tissues, such as rudimentary tendons and ligaments, cells surrounding mesenchymal condensations that later form bones and cells in future joint areas. Development of these tissues is regulated by BMPs, GDFs and TGF-βs (Brunet et al. 1998; Francis-West et al. 1999; Capdevila & Belmonte 2001; Schweitzer et al. 2001). HtrA1 was also expressed in the developing endocardial cushion where epithelial-to-mesenchymal transformation (EMT) was mainly regulated by TGF-β2 (Nakajima et al. 2000; Sandford et al. 1997). TGF-β-regulated EMT is also a fundamental process involved in the malignant progression and metastasis of cancers (Moustakas et al. 2002). Maintenance of normal chondrocytes in the joint cartilage requires intricate regulation by TGF-βs, and transgenic mice with disturbed TGF-β signaling show phenotypes very similar to human osteoarthritis (Serra et al. 1997; Yang et al. 2001). All these findings suggest close association of HtrA1 and TGF-β signaling. In fact, the N-terminal region of HtrA1 shows homology to Mac25, which in turn shares structural similarity to follistatin, an activin antagonist (Kato 2000). Our previous investigation has proved that HtrA1 binds to a variety of TGF-β family proteins, and that HtrA1 inhibits signaling of at least BMP2, -4 and TGF-β1 not only in vitro but also in vivo. Surprisingly, the signal inhibition activity depends totally on the proteolytic activity of HtrA1 (Oka et al. 2004). The actual mechanism of how the secretory protease inhibits TGF-β signaling remains elusive. HtrA1 seems not to degrade TGF-β molecules or their receptors. The most plausible mechanism may be degradation by HtrA1 of some components of ECM which support signaling of various TGF-β family proteins (Kresse & Schönherr 2001).

To understand functions of other mammalian HtrA family members, we analyzed the expression pattern and biochemical activities of HtrA3. The expression patterns of humans HtrA1, -2, and -3 have been examined and compared in detail mostly in adult tissues using human multiple tissue expression arrays (Nie et al. 2003a). In view of the fact that HtrA1 plays important roles in the developmental regulation in close association with the TGF-β signaling pathway, we focused our attention on the expression analysis of HtrA3 during mouse development. In this paper we describe that HtrA3 and HtrA1 show similar, but sometimes complementary expression patterns in mouse embryos and that HtrA3 shares the same inhibitory activity on TGF-β signaling with HtrA1.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Plasmids and reagents

cDNA encoding the full length of mouse HtrA3 was a gift of T. Matsuguchi, Nagoya University. The full length cDNAs encoding HtrA3 and HtrA1 were tagged with tandem-aligned three myc epitopes at the C-terminus (termed HtrA3-myc and HtrA1-myc, respectively) and cloned into the pcDNA3 mammalian expression vector. To produce the protease activity-deficient mutant of HtrA3 (HtrA3 S311A), serine at amino acid position 311 was substituted with alanine by PCR mutagenesis. A mutant of HtrA3 with a truncation of N-terminal 137 amino acids (ΔN HtrA3) was constructed by removing a 5′-cDNA region with EcoRI and XhoI. The protease activity-deficient HtrA1 mutant (HtrA1 S328A) and a mutant with a truncation of N-terminal 150 amino acids (ΔN HtrA1) were described previously (Oka et al. 2004). Plasmid DNA used in cotransfection assays were obtained as follows: pGL3-Id985WT and caBMPR-IB cDNA in pEF-BOS from T. Katagiri, Showa University, Tokyo; pGL3ti-(SBE)4 from W. Kruij, Biological Center, Kerklaan; mouse Smad cDNA from K. Miyazono, University of Tokyo; mouse BMP4 cDNA from D. Constam, Swiss Institute for Experimental Cancer Research. Human recombinant BMP-2 protein was provided by Yamanouchi Pharmaceuticals, Tokyo and human recombinant TGF-β1 was purchased from Pepro Tech Inc. (Rocky Hill, NJ, USA). β-casein (bovine milk), biglycan (bovine articular artilage), decorin (bovine articular cartilage), bovine serum alubumin (BSA, fraction V), collagen (type I insoluble, bovine achilles tendon) and fibronectin (bovine plasma) were purchased from Sigma-Aldrich Japan (Tokyo).

Preparation of HtrA1 and HtrA3 probes

The HtrA1 probe (554 bp, from nucleotide 1197–1750) has been described previously (Oka et al. 2004). Two fragments of HtrA3 cDNA cloned in pBSKS (Stratagene) were used to prepare hybridization probes; one fragment (500 bp, from nucleotide 219–718, refer to the sequence registered as AY037300 for nucleotide number) was derived from the 5′-coding region and the other fragment (790 bp, from 1730 to 2519) was derived from the 3′ non-coding region. The 5′-coding probe was used in this report for northern blot analysis after 32P-labeling and for in situ hybridization after digoxigenin-labeling. The 3′ non-coding probe was used to confirm the specificity of in situ hybridization signals.

Sense and antisense digoxigenin-labeled RNA probes were synthesized with a kit (Roche Diagnostics) using T3 and T7 RNA polymerases (Boehringer Mannheim). Specificity of the HtrA1 and HtrA3 probes was confirmed by genomic Southern blotting; each probe detected only the corresponding gene fragments under standard conditions.

RNA isolation and northern blot analysis

RNA was isolated from mouse embryos or from tissues of adult mice (8–11 weeks old) using ISOGEN (Wako Pure Chemical Industries, Osaka Japan) following manufacturer's instructions. Total RNA (20 µg) was size-fractionated by electrophoresis. The RNA was transferred to Hybond-N membrane (Amersham Pharmacia Biotech, Buckinghamshire, UK). The membrane was hybridized with a 32P-labeled probe prepared by random priming, followed by washing three times with 2 × SSC/0.1% SDS for 20 min at 65°C and twice with 1 × SSC/0.1% SDS for 20 min at 65°C. The membrane was exposed to X-ray film (Fuji Film, Tokyo). The probes were stripped off by boiling the membranes in 0.1 × SSC/1% SDS for 30 min then the membranes were rehybridized with a β-actin probe for RNA loading control. The β-actin probe used was a 410 base-pair BglII-XbaI fragment of mouse β-actin cDNA (nucleotide 173–582).

Preparation of tissue samples and immunohistochemical analysis

Mouse tissues were fixed in 4% paraformaldehyde in a phosphate buffered saline (PBS) at 4°C overnight. The tissue specimens were embedded in OCT compound or in paraffin and sectioned at 6–20 µm thick.

Bone samples were fixed in 4% paraformaldehyde in PBS at 4°C, and decalcified in 10% EDTA (pH 7.4) for 2 weeks at 4°C. The EDTA solution was changed every 2 days. The bone specimens were then dehydrated in grading concentrations of methanol and chloroform and finally embedded in paraffin.

These sections were used for in situ hybridization and immunohistochemical staining. Immunohistochemical staining was performed using the tyramide signal amplification-avidin-biotin-complex method as previously reported (Toda et al. 1999).

Whole mount and section in situ hybridization

Mouse embryo was fixed in 4% paraformaldehyde in PBS at 4°C overnight. The fixed embryos were dehydrated in methanol and stored at -80°C before use. In situ hybridization was carried out essentially as described previously (Sasaki & Hogan 1994; Morimoto et al. 1996). Briefly, the whole embryos were rehydrated in PBS and treated with 50 µg/mL proteinase K for 1 h. The embryos were then fixed in a mixture of 0.2% glutaldehyde and 4% paraformaldehyde in PBST (0.1% triton X-100 in PBS) for 20 min. After two washes with PBST for 5 min each, the embryos were incubated in a prehybridization solution (5 × SSC, 50% formamide and 1% SDS) at 70°C for 1 h. Hybridization was carried out at 70°C overnight in the prehybridization solution supplemented with 1 µg/mL digoxigenin-labeled probe, 50 µg/mL tRNA and 50 µg/mL heparin. After hybridization, excess probes were removed by washing in 5 × SSC containing 50% formamide. The embryos were further washed with 2 × SSC containing 50% formamide at 70°C and 65°C for 1 h each. The embryos were then treated with 1.5% blocking reagent (Boehringer Mannheim, Germany) containing 20% sheep serum for 2 h and then incubated overnight at 4°C with alkaline phosphatase-conjugated antidigoxigenin antibody (Boehringer Mannheim) diluted 1 : 1000 in the blocking reagent. The next day the embryos were washed and incubated with NBT/BCIP to detect bound phosphatase.

Air-dried sections were fixed in 4% paraformaldehyde in PBS for 30 min. The sections were incubated with proteinase K (10 µg/mL) in 10 mM Tris-HCl (pH 7.6) and 1 mM EDTA for 10 min at 37°C and postfixed in 4% paraformaldehyde, in PBS for 10 min. The sections were briefly washed in PBS and acetylated two times for 10 min each with 0.4% (v/v) acetic anhydride in 0.1 M triethanolamine. The sections were rinsed in PBS, incubated with 0.2 M HCl for 10 min, and then washed in PBS for 5 min before hybridization. Hybridization was carried out in a buffer containing 5 × SSC, 50% formamide, 0.1 mg/mL tRNA, and 1 µg/mL antisense probe overnight at 55°C. Hybridized sections were washed and the hybridized probe was detected as described for whole-mount hybridization. For control, embryos and sections were hybridized with an equivalent amount of sense probe in the same way as described previously. The two non-overlapping HtrA3 probes gave rise to the same results, indicating the specificity of the signals.

Preparation of antibody against HtrA3

An N-terminal peptide of mouse HtrA3 (amino acid 25–128) was produced in E. coli as a histidine-fusion protein using pET-28 vector (Novagen, Madison, WI, US) and purified using Ni-NTA agarose (Qiagen, Chatsworth, CA, US). The purified protein (500 µg/rabbit) was mixed with adjuvants and injected intracutaneously into two rabbits (NZW) every four weeks. Two weeks after the fourth injection, blood was withdrawn and serum was prepared. The unpurified serum was used for western blotting and immunohistochemical analyses at 1 : 1000 and 1 : 2000 dilution, respectively. This antiserum did not cross-react with HtrA1 (Fig. 7B).

image

Figure 7. (A) Increased expression of HtrA3 in joint cartilage affected by arthritis. The joint samples were prepared from mouse 6 days after lipopolysaccharide injection. The joints sectioned were stained with HtrA3 antibody (a and b). The superficial chondrocytes in the articular cartilage (a) and chondrocytes in the epiphyseal cartilage (b) displayed increased staining as compared with Fig. 6 (e and f). (B) Western blotting showing specificity of anti-HtrA3 antibody. HtrA3-myc and HtrA1-myc proteins were produced in sf21 cell using the baculovirus system. Culture supernatants were electrophoresed and blotted on membranes. The membrane was stained with anti-HtrA3 antibody (upper panel) or with antimyc antibody (lower panel) to show the amounts of loaded HtrA3-myc and HtrA1-myc proteins. (C) Increase in HtrA3 contents in arthritic joints. Knee joints from each mouse were prepared 6 or 9 days after lipopolysaccharide injection. The joint cartilage was scraped off and extracted with a guanidine buffer. The same amounts of the extracted proteins (8.0 µg each lane) were separated by SDS electrophoresis, blotted on a membrane, and probed with anti-HtrA3 antibody. The joints from two mice prepared at day 6 both showed 1.5-fold increase, and the joints of a mouse prepared at day 9 showed 2.5-fold increase in HtrA3 protein. ac, articular cartilage; bm, bone marrow; m, bone matrix; ec, epiphyseal cartilage.

Download figure to PowerPoint

Western blot analysis

Samples containing equal amount of protein were separated by SDS/PAGE on 10% polyacrylamide gel. The proteins were electro-transferred to a nitrocellulose membrane (Schleicher & Schuell, USA). The membrane was incubated with a blocking buffer (5% skimmed milk in TBS (10 mM Tris-HCl (pH 7.5), 150 mM NaCl)) for 1 h at room temperature. The membrane was then probed with primary antibody in the blocking buffer for 2 h at room temperature, washed five times with TBS. The membrane was then incubated with horseradish peroxidase-conjugated antirabbit IgG antibody (Amersham Biosciences) in the blocking buffer, followed by washing with TBS. The peroxidase on the membrane was detected by ECL according to the manufacturer's instruction (Amersham Biosciences) using X-ray film (Fuji Film, Tokyo).

Expression of HtrA3 in mammalian cells

A cDNA fragment encoding HtrA3-myc was cloned into pFASTBAC1 (Gibco, Paisley, UK), yielding a bacmid transfer vector. The transfer vector was used for transformation of the DH10Bac cell (Gibco). Identification and isolation of the recombinant bacmid were performed according to the instructions for the Bac-to-Bac baculovirus expression kit (Gibco).

Sf21 cells (derived from Spodoptera frugiperda) were propagated as monolayer at 27°C in Grace's medium (Gibco) containing 8% fetal bovine serum (FBS) (Wako), 3.3 g/L of lactalbumin hydrolysate and 3.3 g/L of yeastolate, 50 mg/L penicillin, and 80 mg/L streptomycin. Transfection of sf21 cells with recombinant bacmid DNA was performed by using Cellfectin Reagent (Gibco) according to the manufacturer's instructions. Positive viral clones were identified by their ability to direct the expression of secreted HtrA3 protein that was detected by western blot analysis of culture supernatants using antimyc monoclonal antibody (unpurified ascites of 9E10 hybridoma cells).

For expression of HtrA3, Sf21 cells (2 × 106 cells/mL) were infected with the recombinant baculovirus. After 48 h, the culture medium was harvested, clarified by centrifugation at 1000 g for 10 min and stored at -80°C.

293T cell was grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, 50 mg/L penicillin, 80 mg/L streptomycin. The cells were transiently transfected with pCDNA3 HtrA1-myc using calcium phosphate precipitation method (Jordan et al. 1996). The culture medium containing secreted HtrA1-myc protein was collected 48 h after transfection and used for binding assays.

GST pull-down assay

Production of active forms of TGF-β1, TGFβ2, BMP4 and GDF5 and GST-pull down assay were carried out as described previously (Oka et al. 2004). Briefly, cDNA encoding an active form of these growth factors was cloned into pGEX-4T-1 and transformed into DH5α. Expression of GST-fusion proteins was induced with 0.2 mM isopropyl-1-thiol-β-D-galactopyranoside. After incubation at 37°C for 5 h, the cells were disrupted by sonication in 20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 0.05% NP-40 and 10% glycerol. Insoluble GST-fusion proteins were solubilized as described (Groppe et al. 1998).

GST-TGF-β fusion proteins or control GST protein were incubated at 4°C overnight with HtrA3-myc produced by sf21 cells or HtrA1-myc produced by HEK293T in the pull-down buffer which contained 20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 0.05% NP-40 and 10% glycerol. After incubation, the mixture was clarified by centrifugation for 10 min at 21800 g at 4°C. Supernatant was incubated with glutathione 4B beads (Pharmacia, Uppsala, Sweden) for 1 h at 4°C. Beads were then gently centrifuged, washed five times with the pull-down buffer, and boiled in 2 × SDS sample buffer (100 mM Tris-HCl (pH 6.8), 200 mM DTT, 4% SDS, 0.2% Bromophenol Blue, 20% glycerol). The eluted proteins were separated by SDS gel electrophoresis and detected by Coomassie Brilliant Blue (CBB) staining or by western blotting using antimyc monoclonal antibody (9E10 ascites) diluted at 1 : 1000.

Protease assay

An HtrA3 mutant with a truncation of N-terminal 137 amino acids (ΔN HtrA3), a HtrA1 mutant with a truncation of N-terminal 150 amino acids (ΔN HtrA1) and their derivatives with mutation of the active serine residue replaced with alanine, ΔN HtrA3 S311A and ΔN HtrA1 S328A, were produced as histidine-tagged proteins in BL21 cells using the pET 28b vector (Novagen, Madison, WI, US) and purified using Ni-NTA agarose beads (Qiagen, Chatsworth, CA, US).

The protease activities of these proteins were assayed by incubating with various substrate proteins (5–10 µg of β-casein, BSA, decorin, biglycan, or fibronectin) in 20 µL mixture containing 50 mM Tris-HCl (pH 7.6) for 12 h at 37°C. When decorin or biglycan was used, the reaction was terminated by the addition of phenylmethylsulfonyl fluoride (final 5 mM) and then glucosaminoglycan chains were removed by treatment with 0.02 units of Chondrotinase ABC (Seikagaku, Japan) at 37°C for 1 h. The reaction products were boiled for 5 min in the SDS sample buffer, separated by SDS gel electrophoresis and stained with CBB.

Luciferase assay in C2C12 cells

C2C12 myoblasts were maintained in DMEM containing 15% FBS. The C2C12 cells were seeded in 24-well plates at a density of 1.5 × 104 cells/well in 0.5 mL medium. After an overnight incubation, the medium was replaced with DMEM containing 2% FBS. Four hours later, cells were transfected by calcium phosphate method (Jordan et al. 1996). To assay BMP4 signaling, cells were transfected with 250 ng/well of Id985wt-luc reporter plasmid, 125 ng/well of pME-β-gal, 50 ng/well each of pcDEF3-smad1 and pcDEF3-smad4, 50 ng/well of pcDNA-BMP4 and either pcDNA3-HtrA1-myc or pcDNA3-HtrA3-myc (0–200 ng/well). The empty pcDNA3 vector was also added to keep the total amount of DNA constant (725 ng/well). After 16 h of transfection, the medium was refreshed, containing 2% FBS. Twenty-four or 48 hours after medium change, culture medium was collected for western blotting analysis and cell lysates were prepared for assay of luciferase and β-galactosidase. Luciferase activity was measured using a kit (PicaGene, Toyo Ink) and normalized with β-galactosidase activity. For BMP2 signaling assay, pcDNA3-BMP4 was omitted from the transfection and recombinant human BMP2 protein (15 ng/well) was added to the medium at the end of transfection. For TGF-β1 assay, the SBE-luc reporter was used and recombinant human TGF-β1 protein (5 ng/well) was added at the end of transfection.

Induction of mouse arthritis

Arthritis was induced in Balb/c mice (8 weeks old) by injecting a mixture of anticollagen type II monoclonal antibodies followed by lipopolysaccharide using a kit (Arthritogenic mAb Cocktail, Immuno-Biological Laboratorie, Japan). Swelling of joints appeared 3 days after lipopolysaccharide injection and reached maximum after 6–9 days as reported (Wallace et al. 1999). The joints were collected 6, 9 and 13 days after lipopolysaccharide injection and either fixed in 4% paraformaldehyde for immunohistochemical analysis or minced in a guanidine solution to extract proteins as described (Santo et al. 2000).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Expression pattern of mouse HtrA3

To elucidate physiological roles of mouse HtrA3 protein, we examined its expression pattern by northern blotting, in situ hybridization and immunohistochemical analysis by using specific antibody against mouse HtrA3. Northern analysis indicated that a low level of HtrA3 transcripts could be detected as early as embryonic day (E) 9.5 (Fig. 1A), while HtrA1 was first detected at a slightly later (E10.5) stage (Oka et al. 2004). HtrA3 transcript was gradually increased, became most abundant at E16.5 and stayed at the same level until after birth.

image

Figure 1. Northern analysis of HtrA3 expression. (A) Total RNA was isolated from mouse embryos at different embryonic days as indicated. A 32P-labeled fragment of 5′-coding region of HtrA3 cDNA was used as a probe. A 2.8 Kb major band was detected (top panel). After exposure to X-ray film, the filter was stripped and rehybridized with a β-actin probe for an RNA loading control (bottom panel). (B) Total RNA was isolated from tissues of E16.5 embryos and adult (8 weeks) mice. The placenta (10.5 days of gestation) was removed from an 11-week-old mouse. Hybridization was carried out using 32P-labeled HtrA1 probe (top panel) or the HtrA3 5′-coding probe (middle panel). A 2.4 kb major band was detected with HtrA1 probe. The filter was stripped and hybridized again with the β-actin probe (bottom panel).

Download figure to PowerPoint

Whole mount in situ hybridization showed that HtrA3 was first detected in the eye of E10.5 embryo (data not shown). Then the expression expanded to the tissues closely associated with the skeletal system. In E12.5 embryo, HtrA3 was expressed in the vertebral rudiments in the tail region (Fig. 2a and c). This was in contrast with the expression of HtrA1, which showed predominant expression in the thoracic and lumbar parts of the vertebrae (Fig. 2d and e). HtrA1 was strongly expressed in the future knee joint and the tarsal region of the hindlimb (Fig. 2d and f), but expression of HtrA3 in the limbs was weak and localized mostly in the tibial part (Fig. 2c). Expression patterns of HtrA3 and -1 in the embryonic eye were also different. In whole-mount in situ hybridization, the HtrA3 probe showed a round-shaped expression pattern, while the HtrA1 probe stained a ring-shaped area just outside the HtrA3 expression area (Fig. 2a and d). The sections of the embryo eye indicated that HtrA3 was expressed in the anterior surface of the lens but not expressed in the lens fibers (Fig. 3a–c). On the other hand, HtrA1 was expressed strongly at the peripheral margin of the neural retina which later develops to the ciliary body (Fig. 3g-i) and moderately in the pigment layer of the retina. HtrA3 expression in the lens epithelium continued during embryonic stages (Fig. 3d), but in the adult eye HtrA3 could no longer be detected in the lens. The expression of HtrA1 in the ciliary body, on the other hand, continued and expanded to the iris and cornea of E16.5 embryo (Fig. 3j). The high expression in the ciliary body, iris and cornea was retained in the adult mouse (data not shown). In the adult eye, a weak expression of HtrA3 was detected only in the ganglion cell layer and the inner nuclear layer of the retina (Fig. 3e). These two neural layers displayed high levels of HtrA1 expression (Fig. 3k). These data agree well with the northern blot analysis, which showed that HtrA1 was expressed at a much higher level than HtrA3 in the adult eye (Fig. 1B). The expression patterns of HtrA3 and − 1 suggested that these two very similar genes could have complementary functions in the developmental regulation of the eye and the skeletal tissues.

image

Figure 2. Whole-mount in situ hybridization of E12.5 embryos showing expression of HtrA3 (a–c) and HtrA1 (d–f). Arrows indicate a round-shaped expression of HtrA3 and a ring-shaped expression of HtrA1 in the eye rudiments. Arrow heads indicate HtrA3 expression in the tail region and HtrA1 expression in the trunk region of the vertebrae. Asterisks show strong expression of HtrA1 in the tarsal region and the knee joint of the hind limb and weak expression of HtrA3 in the tibial region. Insets of a and d show enlarged pictures showing the eyes.

Download figure to PowerPoint

image

Figure 3. Sections of embryonic eyes and adult retina showing expression of HtrA3 (a–e) and HtrA1 (g–k). Sections of eyes from embryos of E12.5 (a–c, f, g–i and l), E16.5 (j), or E18.5 (d) and sections of adult mouse (8 weeks old) retina (e and k) were analyzed by in situ hybridization (a, d, e, g, j, and k) with antisense probes or by immunostaining (b, c, h, and i). f and l show the result of in situ hybridization with control sense probes for HtrA3 and HtrA1, respectively. Arrows in g and h indicate the peripheral retina region where HtrA1 was expressed. c and i show magnified views of the lens epithelium and the peripheral margin of the retina positively stained with anti-HtrA3 and anti-HtrA1 antibodies, respectively. e and k show the adult retina. ls, lens; le, lens epithelium; n, neural retina; ga, ganglion cell layer; in, inner cell layer; ir, iris; o, outer cell layer; p, pigment retina; cb, ciliary body; co, cornea.

Download figure to PowerPoint

The differential expression of HtrA1 and -3 could also be seen in other embryonic tissues. In E16.5 embryo heart, a weak expression of HtrA1 was detected by northern blot but HtrA3 was barely expressed (Fig. 1B). In the E14.5 heart, HtrA1 but no HtrA3 was detected distinctly in the endocardial cushion (Fig. 4a and c). In the trachea of E16.5 embryo, HtrA3 was expressed in the outer layers, whereas HtrA1 was localized predominantly in lamina propria (Figs 4b and d). In the aorta of this stage, HtrA3 was expressed in the adventitia and HtrA1 in the intima.

image

Figure 4. Section in situ hybridization of E14.5 (a and c) and E16.5 (b and d) embryos at the thoracic level showing expression of HtrA3 (a and b) and HtrA1(c and d) in the heart, blood vessels and trachea. at, atrium; ao, aorta; av, atrio-ventricular cushion; v, ventricular septum; vc, vena cava; e, esophagus; t, trachea.

Download figure to PowerPoint

Among the adult organs, the placenta expressed highest levels of both HtrA1 and -3 (Fig. 1B). In situ hybridization revealed that HtrA3 was expressed diffusely in the labyrinthine layer with patchy areas showing strong expression (Fig. 5a and b). HtrA1, on the other hand, showed a patchy expression pattern (Fig. 5e and f). As judged from the distribution, shape and size of the stained cells, it is most likely that HtrA1 expressing cells are trophoblasts (Maekawa et al. 1999).

image

Figure 5. In situ hybridization showing expression of HtrA3 (a–d) and HtrA1 (e–h) in the placenta of day 10.5 of gestation (a–c and e–g) removed from 11-week-old mouse and in the ovary of 8-week-old mouse (d and h). c and g show results of in situ hybridization with control sense probes of HtrA3 and HtrA1, respectively. gr, granulosa cell; o, oocyte; tr, trophoblast.

Download figure to PowerPoint

The ovary expressed a high level of HtrA1 (Fig. 1B). The expression of HtrA3 in the ovary seemed to be relatively lower than HtrA1. In consistent with the result of northern blot analysis, the in situ hybridization signals were much lower for HtrA3 than for HtrA1 in ovary. The granulosa cells were the major source of both HtrA1 and -3 (Fig. 5d and h).

Expression of HtrA3 in the skeletal tissues

Because HtrA1 is characteristically expressed in the skeletal tissues of the embryo and adult mouse (Oka et al. 2004) and because HtrA1 expression was up-regulated in the cartilage of human osteoarthritis patients (Hu et al. 1998), we examined the later stage fetal, infant and adult mice in detail for the expression pattern of HtrA3 in the bone and tissues associated with the bone (Fig. 6).

image

Figure 6. Expression pattern of HtrA3 (a, b, e and f) and HtrA1 (c and d) in various stages of bone development. The forelimb of E14.5 embryo was analyzed by in situ hybridization (a and c). The femur of postnatal day 7 infant (b and d) and the femur of adult (8 weeks) mouse (e and f) were analyzed by immunostaining. Staining of adult femur sections without the antiserum for HtrA3 did not give rise to any signals (g). ac, articular cartilage; bm, bone marrow; m, bone matrix; oc, ossification center; ec, epiphyseal cartilage.

Download figure to PowerPoint

In the forelimb of E14.5 embryos, HtrA1 expression was detected widely in the rudimentary tendons and ligaments and mesenchymal cells surrounding cartilaginous condensations, which later developed to bones (Fig. 6c). Especially, the cells in the future carpal and elbow joints showed strong expression of HtrA1. On the other hand, HtrA3 was expressed in a limited number of the tendons and ligaments (Fig. 6a). Mesenchymal cells in the carpal and elbow joint regions barely expressed HtrA3. Both HtrA3 and -1 were not expressed in the core of the cartilaginous condensations. Interestingly, however, when the blood vessels invaded into the condensations and ossification started in the peri- and postnatal periods, the expression of HtrA3 and -1 were tremendously up-regulated (Fig. 6b and d). Chondrocytes probably undergoing degeneration in the ossification center produced HtrA1 and -3.

In the adult bone, HtrA3 was largely localized in the bone matrix (Fig. 6e and f) as in the case of HtrA1 (Oka et al. 2004). HtrA3 exhibited significant expression in the articular chondrocytes (Fig. 6e). On the contrary, HtrA1 was hardly expressed by mature articular chondrocytes (Oka et al. 2004). Both HtrA3 (Fig. 6f) and HtrA1 (data not shown) were not expressed in the hypertrophic and proliferating chondrocytes in the epiphyseal cartilage of a normal mouse.

Damages in chondrocytes in osteoarthritis induced expression of HtrA1 several folds in the joint cartilage (Hu et al. 1998; Tsuchiya unpubl. data). The expression of HtrA3 was similarly induced in mouse experimental arthritis but to a lesser extent (1.5–2.5-fold) (Fig. 7C). The size of the HtrA3 detected in the cartilage was smaller than the size expected from the cDNA structure or the size of HtrA3 protein produced by sf21 cells. When western blot analysis was carried out with anti-HtrA3 serum absorbed with bacterially produced HtrA3 protein this smaller band was not detected (data not shown). Western blotting of placenta proteins revealed the presence of both the full length form and a smaller form of HtrA3 (data not shown). These data strongly supported that the detected band was a truncated HtrA3 that was probably produced by proteolytic cleavage. Since the polyclonal antibody against HtrA3 recognizes the N-terminal half of the molecule, it is likely that this smaller molecule is a form of HtrA3 that has intact N-terminal and protease domains but lacks the PDZ domain (predicted Mr = 40 000). Similar degradation products have been reported for human HtrA1 in the joint cartilage (Hu et al. 1998).

Mostly chondrocytes in the superficial layer of the joint cartilage produced HtrA3 after arthritis induction (Fig. 7A, a). Chondrocytes in the epiphyseal cartilage displayed strong elevation in HtrA3 protein production (Fig. 7A, b). These findings were in contrast with the induction of HtrA1 expression, which was most prominent in the chondrocytes of the deep layer in the joint cartilage and minimum in the epiphyseal cartilage (Tsuchiya unpubl. data). Again this result suggested a complementary function of HtrA3 and HtrA1 not only in normal developmental processes but also in pathological conditions.

Binding of HtrA3 to TGF-β family proteins

The above mentioned results on expression patterns indicated that HtrA3 and -1 were expressed in common tissues and in some regions they were expressed in complementary manners. Based on the characteristic expression pattern of HtrA1 and its molecular similarity to follistatin, we have proposed that HtrA1 functions as an inhibitor of TGF-β signaling. Actually, we have shown that HtrA1 binds to various TGF-β family proteins and inhibits their signaling (Oka et al. 2004). To examine if HtrA3 has similar activities, we first tested the binding of HtrA3 to various TGF-β proteins. The myc-tagged HtrA3 (HtrA3-myc) produced by sf21 cells were used for GST pull-down assay with GST-TGF-β1, -TGF-β2, -BMP4, and -GDF5. HtrA3-myc bound to all of these TGF-β proteins (Fig. 8) but not to control GST protein. As judged from the recovery of HtrA3 protein through the pull-down assay, HtrA3-myc showed preferable binding to GDF5 and BMP-4 rather than to TGF-β1 and -β2. This is consistent with the result with HtrA1-myc, which bound to all these TGF-β proteins with the highest affinity for BMP4 (Fig. 8, Oka et al. 2004). Because HtrA1 was recovered less than 3% of input and HtrA3 more than 3% in the pull-down assay with BMP4, it was likely that HtrA3 had the same or slightly higher binding activity to BMP4 than HtrA1.

image

Figure 8. GST pull-down assay of HtrA3 and HtrA1 with various TGF-β proteins. GST protein (control) or GST-fused various TGF-β proteins were incubated with conditioned medium containing HtrA1-myc or HtrA3-myc protein and pulled down with Ni-NTA beads. Pulled-down samples were separated by SDS gel electrophoresis and detected by anti-myc antibody (top panel) to assay the bound HtrA proteins or by CBB staining (bottom panel) to confirm the amount of GST or GST-fusion proteins recovered. The right two lanes contain HtrA1-myc and HtrA3-myc proteins corresponding to 3% of the amounts used for the GST pull-down assay. An unknown protein band derived from sf21 culture medium appeared above the band of HtrA3 protein. Nb: the position of HtrA proteins in the 3% input lanes was sifted downward due to the presence of a large amount of BSA in the culture media.

Download figure to PowerPoint

Inhibition of TGF-β signaling

Next we investigated the effects of HtrA3 on BMP2, -4 and TGF-β1 signaling in C2C12 myoblastic cells. The transfection of C2C12 cells with the BMP4 expression vector resulted in approximately 19-fold induction of a luciferase reporter activity under the control of a BMP-responsive promoter element (Id985wt) (Fig. 9A). Similar to HtrA1, HtrA3 suppressed the BMP4 activation of this reporter in a dose-dependent manner. The effect of HtrA3 on TGF-β1 signaling in C2C12 cells was next examined using a luciferase reporter driven by a TGF-β responsive promoter element (SBE). Both HtrA1 and HtrA3 inhibited moderately the signaling induced by the recombinant TGF-β1 protein (Fig. 9C). HtrA1 displayed moderate inhibitory activity, as compared with noggin, on the activation of the Id promoter induced by recombinant BMP2 protein in C2C12 cells (Oka et al. 2004). HtrA3 also showed a weaker but significant inhibition on the recombinant BMP2 (data not shown).

image

Figure 9. Inhibition of TGF-β signaling by HtrA3. (A) Inhibition of BMP4 signaling by HtrA3. C2C12 cells were cotransfected with the BMP4 expression vector (50 ng DNA/well containing 0.5 mL of medium) and various amounts of expression vectors (50–200 ng DNA/well) for HtrA3 and HtrA1 as indicated. Inset depicts the result of western blot analysis of culture medium (10 µL each) showing the expression levels of HtrA proteins; from left, HtrA1 produced with 50, 100 and 200 ng DNA/well, HtrA3 produced with 50, 100 and 200 ng DNA/well. (B) An HtrA3 mutant lacking protease activity was inactive as a signal inhibitor. Co-transfection was carried out as described in (A) except that the expression vector for the protease activity-deficient mutant, HtrA3 S311A, was used. (C) Inhibition of TGF-β1 signaling by HtrA3. Co-transfectoin was carried out as described in (A) except that the SBE reporter plasmid was used and recombinant human TGF-β1 (5 ng protein/well, i.e. 10 ng/mL) was added in place of the BMP4 expression vector. Upper inset shows the expression levels of HtrA proteins as analyzed by western blotting as shown in (A). Note that the expression of HtrA3 was under the detection level by this western blotting, but was detected by immunoprecipitation (arrowhead in lower inset; hc represents IgG heavy chain) with anti-myc antibody followed by western blotting with the same anti-myc antibody. (D) HtrA3 did not inhibit signaling from the constitutively active BMP type 1 receptor. Co-transfection was carried out using the Id985wt reporter, the expression vector for caBMPR-1B, various amounts of HtrA3 expression vector as indicated. The amount of caBMPR-1B plasmid added (3 ng DNA/well) was chosen so that the active receptor gave rise to an induction of Id985wt reporter at a similar level to that caused by the expression of BMP4 (20–50-fold) as shown in Figs 9A and B.

Download figure to PowerPoint

HtrA3 did not inhibit the signaling generated by a constitutively active BMP type-1 receptor, caBMPR-1B, which is able to originate signal transduction without ligand binding (Akiyama et al. 1997; Katagiri et al. 2002) (Fig. 9D). This clearly showed that the inhibition by HtrA3 occurred in the extracellular space and not on the signal transduction pathways in the cytoplasm.

Inhibitory activities of HtrA3 appeared weaker than those of HtrA1. However, if the levels of protein expression in the C2C12 culture media were compared by western blotting (see inlets of Fig. 9A,C), the amounts of HtrA3 expressed were always lower that those of HtrA1. This indicates that HtrA3 may be as potent, if not more potent, an inhibitor of TGF-β proteins as HtrA1.

HtrA1 absolutely required its protease activity to inhibit TGF-β signaling (Oka et al. 2004). Consistent with this finding, a mutation of serine in the protease active center eliminated the inhibitory activity of HtrA3 on the BMP4 signaling (Fig. 9) simultaneously with the proteolytic activity (Fig. 10).

image

Figure 10. Protease activity of bacterially produced HtrA3. (A) Digestion of β-casein by HtrA3 and HtrA1. β-casein (10 µg) was incubated without enzymes (lane 1) or with one of the HtrA proteins in a 20 µL reaction mixture for 30 min at 37°C as follows: lane 2, 0.1 µg of ΔN HtrA1 S328A; lanes 3–5, 0.011, 0.033 and 0.1 µg of ΔN HtrA1; lanes 6–8, 0.011, 0.033, and 0.1 µg of ΔN HtrA3; and lane 9, 0.1 µg of ΔN HtrA3 S311A. The reaction products were subjected to SDS-gel electrophoresis. The gel was stained with CBB. (B) Digestion of BSA, decorin and biglycan by HtrA3 and HtrA1. β-casein (10 µg), BSA (10 µg), decorin (5 µg), biglycan (10 µg) or fibronectin (10 µg) was incubated without enzymes (lanes 1, 4, 7, 10, 13) or with 0.1 µg each of ΔN HtrA1 (lanes 2, 5, 8, 11, 14) or ΔN HtrA3 (lanes 3, 6, 9, 12, 15) for 12 h at 37°C. Biglycan and decorin were treated with chondrotinase ABC after termination of the protease reaction. BSA and some other protein bands appeared in the decorin, and biglycan lanes came from the chondroitinase ABC solution. Samples were subjected to SDS-gel electrophoresis. The gel was stained with CBB.

Download figure to PowerPoint

Proteolytic activity of HtrA3

β-casein could be used as a generic substrate to monitor the protease activity of HtrA3. We produced truncated forms of HtrA3 in bacteria as histidine-fusion proteins; ΔN HtrA3 with deletion in the N-terminal IGF binding protein and KI domains and ΔN HtrA3 S311A which contains the same N-terminal deletion and a mutation at the serine residue in the active center. The proteolytic activities of these proteins were compared with similarly mutated forms of HtrA1. Figure 10A shows that ΔN HtrA3 rapidly degraded β-casein. The protease activity of ΔN HtrA3 was comparable to that of ΔN HtrA1. ΔN HtrA3 S311A and ΔN HtrA1 S328A were totally inert as proteases. Recently our study on HtrA1 (Yano, unpubl. data) revealed that several proteoglycans in ECM served as substrates for HtrA1. We examined decorin, biglycan and a glycoprotein, fibronectin, as candidates for HtrA3 substrates (Fig. 10B). HtrA3 digested decorin and biglycan fairly rapidly, when HtrA3 did not significantly cleave non-denatured BSA, which was a poor substrate for HtrA1 (Yano, unpubl. data). Fibronectin was hardly digested by ΔN HtrA3 or by ΔN HtrA1. HtrA3 and -1 therefore exhibit similar substrate specificity.

The mechanism for the inhibitory activity of HtrA1 on the TGFβ signaling remains elusive. One intriguing possibility is that HtrA1 digests some components of ECM that are essential for efficient TGF-β signaling. The proteolytic activity of HtrA3 on decorin and biglycan may support this notion, but physiological substrates of HtrA3 and -1 should be identified to explore the unique inhibitory activity of these serine proteases on the TGF-β signaling.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Complementary expression of HtrA3 and -1

This study characterized a new member of the mammalian HtrA serine protease family, HtrA3 that shares the similar domain structure and high homology with HtrA1. Base on their structural resemblance, HtrA3 and -1 have been predicted to have similar activities. However, since precise expression patterns of the HtrA3 gene had not been examined and compared with those of HtrA1, there remained the possibility that the two genes could have non-redundant, complementary functions in different regions of organs.

Nie et al. analyzed expression of human HtrA1, -2 and -3 using multiple tissue expression arrays and showed that HtrA1 and -3 displayed a very different pattern of expression among human adult tissues (Nie et al. 2003a). Our previous study revealed that the expression of HtrA1 was under intricate developmental regulations (Oka et al. 2004). It is therefore of particular interest to analyze HtrA3 expression in embryonic stages using in situ hybridization or immunohistochemical analysis to identify cell types differentially expressing these two genes. HtrA1 is characteristically expressed in distinct embryonic tissues where TGF-β family proteins play major roles in regulation of differentiation. For instance, HtrA1 is expressed in the developing skeletal tissues, such as rudimentary tendons, ligaments, mesenchymal cells surrounding the future bones and cells in future joint regions. HtrA3 was also expressed in these tissues, but the expression was less robust and restricted to a limited number of them. In vertebral rudiments, HtrA1 and -3 are expressed in non-overlapping regions; HtrA3 at the tail region and HtrA1 at the thoracic and lumbar regions. The embryo eye shows the most striking differential expression patterns for HtrA1 and -3. HtrA1 is expressed in the pigment retina and ciliary body, whereas HtrA3 is expressed in the lens epithelium. Non-overlapping expression patterns were also seen in the blood vessels; the outer adventitial layer expressed HtrA3 and the inner intimal layer HtrA1. No significant expression of HtrA3 was observed in the endocardical cushion (Fig. 4c), the choroids plexus of the brain and the gonad (data not shown), where HtrA1 was expressed at high levels and supposed to play important roles in regulation of TGF-β signaling (Oka et al. 2004). It is likely that HtrA1 and -3 have non-redundant functions at least in some developing tissues.

In adult mouse tissues, both HtrA1 and -3 are expressed at the highest levels in the placenta and ovary. In situ hybridization of the ovary showed that HtrA1 and -3 were produced by the granulosa cells, which are the major source of activin (Harlow et al. 2002). More robust hybridization signals were observed for HtrA1 than HtrA3 in the ovary. In the placenta, HtrA3 displayed more diffuse expression pattern than HtrA1 whose expression seemed to be restricted to the trophoblasts in the labyrinthine region. Northern blot analysis indicated that expression of HtrA3 in the placenta was much higher than in the ovary (Fig. 1B). These data suggest that HtrA3 plays important roles in the placental development and functions, and probably contributes more to them than HtrA1. This is contradictory to the result of human expression patterns (Nie et al. 2003a). In humans, placental HtrA1 is expressed relatively higher than HtrA3. The usage of HtrA genes may be different in mice and humans. Based on the unique expression pattern of HtrA3 in the peri- and postimplantation uterus, Nie et al. proposed that HtrA3 plays an important role in the formation/function of the placenta (Nie et al. 2003a; Nie et al. 2003b); one possible mechanism predicted HtrA3 being an IGF binding and degrading protease. The IGF-IGF binding protein system has been repeatedly reported to participate in the growth and development of placenta. Our preliminary data indicated that HtrA1 bound to, but did not cleave, IGF-I in vitro. Future study will establish the intrinsic function of HtrA1 and -3 in the placenta during pregnancy.

TGF-β signal inhibition activity of HtrA3

We have reported that mouse HtrA1 binds to various TGF-β proteins and inhibits their signaling in C2C12 myoblast cells (Oka et al. 2004). Likewise, HtrA3 also bound to a broad range of TGF-β family proteins in the GST pull-down assay and actually inhibited BMP4, BMP2 and TGFβ1 in cotransfection assay in C2C12 cells. Although HtrA1 is able to inhibit in vivo BMP signaling in the chick eye development, it is still untested if HtrA3 has the same in vivo activity. The TGF-β signal inhibition activities of HtrA3 and -1 were both absolutely dependent on their protease activities. In the case of HtrA1, a deletion of a small linker region between the Kazal protease inhibitor domain and the protease domain eliminated both the protease and TGF-β signal inhibition activities (Oka et al. 2004). This small region is conserved among all mammalian HtrA proteins and is essential for homotrimer formation of HtrA2 (Li et al. 2002) and HtrA1 (M. Yano, unpubl. data). Since this region is also highly conserved in HtrA3, it is likely that HtrA3 forms trimer to become active as a protease and as an inhibitor of TGF-βs.

Proteolytic activity of HtrA3

Bacterially produced ΔN HtrA3 displayed proteolytic activity towards β-casein as potent as ΔN HtrA1. Deletion of the N-terminal end enhanced the protease activity of HtrA1 3× as compared with the intact, full length HtrA1 (M. Yano, unpubl. result). Similarly, ΔN HtrA3 exhibited elevated protease activity (data not shown), suggesting that N-terminal regions exert regulatory function over the catalytic domains of both HtrA1 and -3.

To identify physiological substrates for HtrA1 and -3, we tested if HtrA proteases could degrade components of cartilage, because both HtrA1 and -3 were up-regulated in the arthritic cartilage. We examined fibrillar collagens, proteoglycans, and other glycoproteins whose expression was known to be up- or down-regulated in osteoarthritis (Tsuchiya unpubl. data). Among them, we found that ΔN HtrA3 cleaved decorin and biglycan. TGF-β signaling and small, leucine-rich proteoglycans, such as decorin, biglycan and fibromodulin exhibit a mutual, functional interaction (Hildebrand et al. 1994). The TGF-β signaling induces synthesis of these proteoglycans and they in turn modulate signal transduction of TGF-βs. Decorin and biglycan are known to bind to several specific TGF-β proteins. Decorin can inhibit physiological signaling of TGF-βs by sequestering TGF-β molecules from the vicinity of the receptor, or in some cases decorin enhances the TGF-β signaling probably by effectively concentrating TGF-β molecules near the receptor (Yamaguchi et al. 1990; Markmann et al. 2000). Although we cannot exclude the possibility that the N-terminally truncated forms of HtrA proteins would display modified substrate specificity, our present data seem to suggest that the primary targets of the extracellular HtrA proteases are ECM proteoglycans, including decorin and biglycan.

Decorin and biglycan are ubiquitous component ECM of various tissues, such as the blood vessel, dermis, tendon, ligament, sclera and articular cartilage (Rosenberg et al. 1986). The breakdown of ECM is essential in various developmental processes such as blastocyst implantation, embryonic development, tissue morphogenesis and remodeling as well as in pathological processes such as arthritis or tumor growth and invasion. The breakdown required precisely coordinated and timely controlled expression and activation of growth factors and a host of enzymes that degrade ECM proteins. HtrA3 and -1 may contribute a great deal in these processes as bifunctional molecules; as inhibitors of TGF-β signaling and as degrading enzymes of ECM.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors would like to thank Dr Manabu Fukumoto for valuable advice and Dr Tetsuya Matsuguchi for the generous gift of HtrA3 cDNA. This work was supported by research grants from the Foundation for the Nara Institute of Science and Technology, and the Inamori Foundation, and by grants-in-aid from the Ministry of Education, Culture, Sport, Science and Technology.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References