Dentin sialophosphoprotein (DSPP) is cleaved into its two natural dentin matrix products by three isoforms of bone morphogenetic protein-1 (BMP1)

Abstract

The protease that cleaves the most abundant non-collagenous protein of dentin matrix, dentin sialophosphoprotein (DSPP), into its two final dentin matrix products, dentin sialoprotein (DSP) and dentin phosphoprotein (DPP), has not been directly identified. In this study, full-length recombinant mouse DSPP was made for the first time in furin-deficient mammalian LoVo cells and used to test the ability of three different isoforms of one candidate protease, bone morphogenetic protein-1 (BMP1) to cleave DSPP at the appropriate site. Furthermore, two reported enhancers of BMP1/mTLD activity (procollagen C-endopeptidase enhancer-1, PCPE-1, and secreted frizzled-related protein-2, sFRP2) were tested for their abilities to modulate BMP1-mediated processing of both DSPP and another SIBLING family member with a similar cleavage motif, dentin matrix protein-1 (DMP1). Three splice variants of BMP1 (classic BMP1, the full-length mTolloid (mTLD), and the shorter isoform lacking the CUB3 domain, BMP1-5) were all shown to cleave the recombinant DSPP in vitro although mTLD was relatively inefficient at processing both DSPP and DMP1. Mutation of the MQGDD peptide motif to IEGDD completely eliminated the ability of all three recombinant isoforms to process full-length recombinant DSPP in vitro thereby verifying the single predicted cleavage site. Furthermore when human bone marrow stromal cells (which naturally express furin-activated BMP1) were transduced with the adenovirus-encoding either wild-type or mutant DSPP, they were observed to fully cleave wild-type DSPP but failed to process the mutant DSPP(MQDeltaIE) during biogenesis. All three BMP1 isoforms were shown to process type I procollagen as well as DSPP and DMP1 much more efficiently in low-salt buffer (< or = 50 mM NaCl) compared to commonly used normal saline buffers (150 mM NaCl). Neither PCPE-1 nor sFRP2 were able to enhance any of the three BMP1 isoforms in cleaving either DSPP or DMP1 under either low or normal saline conditions. Interestingly, we were unable to reproduce sFRP2's reported ability to enhance the processing of type I procollagen by BMP1/mTLD. In summary, three isoforms of BMP1 process both DSPP and DMP1 at the MQX/DDP motif, but the identity of a protein that can enhance the cleavage of the two SIBLING proteins remains elusive.

Fig.1

Mutation of the Met-Gln to Ile-Glu in DSPP’s proposed BMP1-cleavage site completely prevented its processing by all three isoforms of BMP1. A. Recombinant wild-type DSPP^wt and mutant protein DSPP^MQΔIE were incubated 21 h at 37 °C in low-salt reaction buffer (25 mM HEPES, 10 mM NaCl, 5 mM CaCl₂, 0.01% Brij) alone or in the presence of the indicated BMP1 isoforms. Samples were electrophoresed on a SDS 4-12% NuPAGE gel, transferred to a PVDF membrane, and immunodetected with anti-DSP domain (LF-153) as detected by fluorescent second antibody and monitored on the Li-COR Odyssey Imager. Note that DSPP^wt was processed by all three BMP1 isoforms while DSPP^MQΔIE was not. Left side numbers indicate the location of molecular weight standards (in kDa). B. Schematic illustrating the domain structures of the three mature BMP1 isoforms used in this study. Domains are as follows: MP, metalloproteinase catalytic domain; C1-C5, CUB domains; E1-E2, EGF-like domains; S, isoform-specific regions.

Fig. 2

Compared to their respective PCP activities, BMP1 isoforms exhibit differential effectiveness in cleaving both DSPP and DMP1. Type I procollagen, DSPP, and DMP1 were incubated for 1h at 37 °C alone or in the presence of the same relative amounts of either BMP1-5, or BMP1 (BMP1-1), or mTLD (BMP1-3) in low-salt reaction buffer (25 mM HEPES, 10 mM NaCl, 5 mM CaCl₂, 0.01% Brij). Samples were then electrophoresed on the appropriate polyacrylamide gel, transferred to membranes when necessary, and detected with (A) antibody against the human collagen α1(I) amino-propeptide (LF-39), (B) antibody against the DSP portion of DSPP (LF-153), or (C) stained with Stains-All to detect DMP1/fragments. Note the differential loss of proα1(I) bands (with concurrent increase in pNα1(I) bands) with BMP1 being the most effective, followed by mTLD, and then BMP1-5(A). Note that BMP1 was the most effective isoform to cleave both DSPP (B) and DMP1 (C), while mTLD (BMP1-3) was ineffective, and BMP1-5 was of intermediate activity. Left side numbers indicate the approximate molecular weights in kDa for relevant prestained standards within each specific gel type. The arrowhead indicates a false-positive protein band variably detected in some preparations of LoVo-derived media on Western blots by the LF-153 antibody. (Note that as would be expected for a false-positive DSPP band, it is not affected by any BMP1 protease.)

Fig. 3

Secreted frizzled-related protein-2 (sFRP2) does not enhance SIBLINGs processing by BMP1 isoforms nor does it enhance their PCP activities. DSPP (A), DMP1 (B) and type I procollagen (C) samples were incubated for 2 h at 37 °C in low-salt reaction buffer (25 mM HEPES, 10 mM NaCl, 5 mM CaCl₂, 0.01% Brij) in the presence of the same amounts (relative to that used for their respective PCP activities) of each isoform and indicated amounts of recombinant sFRP2. Samples without proteases and samples containing only sFRP2 were used as controls. Completed reactions were electrophoresed and detection was performed as described above in Methods. Shown are representative immunoblots for DSPP (A) and type I procollagen (C) as well as a representative Stains-All staining used to detect DMP1. Note that all BMP1 isoforms cleave each substrate but the addition of 50, 100 or 200 ng sFRP2 (12.5, 25 and 50 nM, respectively) did not enhance or inhibit their proteolytic effectiveness. The arrowhead indicates a false-positive protein band variably detected in some preparations of LoVo-derived media on Western blots by the mouse DSP antibody. Left side numbers indicate the approximate molecular weights in kDa for relevant prestained standards.

Fig. 4

The three BMP1 isoforms cleave human type I procollagen in a salt-dependent manner. A. Type I procollagen samples were incubated for noted times with mTLD (BMP1-3) in low-salt (25 mM HEPES, 10 mM NaCl, 5 mM CaCl₂, 0.01% Brij) or high-salt (50 mM Tris, 150 mM NaCl, 5 mM CaCl₂) reaction buffers. Samples collected after 2, 4, 6, and 21 h were electrophoresed with SDS on a 3-8% Tris-acetate NuPAGE gels followed by immunoblotting with antibody against the human collagen α1(I) amino-propeptide (LF-39). Note that the same amount of mTLD in low-salt conditions processes much more procollagen in shorter periods of time than it does in normal saline. B. Type I procollagen samples were incubated separately with BMP1-5, BMP1, and mTLD under increasing NaCl concentration for 1 h at 37 °C before being subjected to immunoblotting with LF-39 antibody. Immunoblots demonstrate dose-dependent decrease in enzymatic activity by all three BMP1 isoforms as the salt concentration reached concentrations above ~50-100 mM NaCl. Left side numbers indicate the approximate molecular weights in kDa for relevant prestained standards.

Fig. 5

Secreted frizzled-related protein-2 (sFRP2) does not enhance the PCP activity of the three BMP1 isoforms in normal saline conditions. Type I procollagen was incubated at 37 °C for 21 h alone or with each of the three BMP1 isoforms plus 50, 100, and 200 ng sFRP2 (~ 12.5, 25, and 50 nM, respectively) under the normal saline conditions (50 mM Tris, 150 mM NaCl, 5 mM CaCl₂) reported earlier (Kobayashi et al., 2009). The processing of type I procollagen (proα1(I)) into pNα1(I) bands was detected by means of Western blot using antibody against the human collagen α1(I) amino-propeptide (LF-39). Representative immunoblots show that increasing sFRP2 concentrations have no effect on any BMP1 isoform’s processing of type I procollagen even in normal saline (high-salt) conditions previously reported. Left side numbers indicate the approximate molecular weights in kDa for relevant prestained standards.

Fig. 6

Cleavage of DSPP and DMP1 by BMP1 isoforms is not enhanced by PCPE-1. Type I procollagen (A), DSPP (B), and DMP1 (C) samples were incubated alone or separately with BMP1-5, BMP1, or mTLD (BMP1-3) for 2 h with or without PCPE-1 in low-salt reaction buffer (25 mM HEPES, 10 mM NaCl, 5 mM CaCl₂, 0.01% Brij) before electrophoresis on their appropriate gels (see methods). The processing of type I procollagen (proα1(I)) into pNα1(I) bands was detected by means of Western blot using antibody against the human collagen α1(I) amino-propeptide (LF-39). Intact DSPP and DSP cleavage fragments were detected by Western blot using anti-DSP antibody described in Fig. 1 whereas the DMP1 and its cleavage products were directly visualized in the gel with Stains-All. Shown are representative results. Note that the processing of procollagen by both BMP1 and mTLD (BMP1-3) is enhanced by PCPE-1 while the reaction using BMP1-5 is not affected. The processing of both SIBLINGs by BMP1 isoforms is unaffected by PCPE-1. The arrowhead indicates a false-positive protein band variably detected in some preparations of LoVo-derived media on Western blots by the LF-153 antibody. Left side numbers indicate the approximate molecular weights in kDa of relevant prestained standards for each gel type.