Dynamic Processing of Recombinant Dentin Sialoprotein-Phosphophoryn Protein*

Dentin sialoprotein (DSP) and phosphophoryn (PP) are the two noncollagenous proteins classically linked to dentin but more recently found in bone, kidney, and salivary glands. These two proteins are derived from a single copy DSP-PP gene. Although this suggests that the DSP-PP gene is first transcribed into DSP-PP mRNAs, which later undergo processing to yield the DSP and PP proteins, this mechanism has not yet been demonstrated because of the inability to identify a DSP-PP precursor protein from any cell or tissue sample. To study this problem, we utilized a baculovirus expression system to produce recombinant DSP-PP precursor proteins from a DSP-PP240 cDNA, which represents one of several endogenous DSP-PP transcripts that influence various tooth mineralization phases. Our in vitro results demonstrate that DSP-PP240 precursor proteins are produced by this system and are capable of self-processing to yield both DSP and PP proteins. We further demonstrated that purified recombinant DSP-PP240, purified recombinant PP240, and the native highly phosphorylated protein (equivalent to the PP523 isoform) have proteolytic activity. These newly identified tissue proteases may play key roles in tissue modeling during organogenesis.

Dentin sialoprotein (DSP) 2 and phosphophoryn (PP) are the two most abundant noncollagenous proteins (NCPs) in dentin. In rats, DSP and PP coding sequences are derived from DSP-PP transcripts (1,2), a finding confirmed in the mouse (3) and in human DSP-PP transcripts (4). Immunohistochemistry and in situ studies showed that DSP/PP proteins and DSP-PP mRNA expression are tightly associated with dentin mineralization (5)(6)(7)(8). Mutations in the DSP-PP gene are linked to dentinogenesis imperfecta II and hearing loss (9,10), whereas DSP-PP null mouse exhibit hypomineralization and dentin dysplasia (11).
These studies, taken together with the early finding that PP binds Ca 2ϩ and can initiate hydroxyapatite formation in vitro (12)(13)(14), strongly support the assertion that DSP and PP proteins play significant roles in dentin mineralization. However, more recent findings by Godovikova et al. (15) demonstrate DSP-PP mRNA expression occurring not only in teeth but also in bone, kidney, and salivary glands, suggesting that the DSP-PP gene may participate in a variety of processes during organogenesis.
Since DSP and PP proteins are derived from a single copy DSP-PP gene, it is commonly assumed that the DSP-PP gene is first transcribed into DSP-PP mRNAs, translated to become DSP-PP precursor proteins, and then enzymatically cleaved to yield DSP and PP proteins found in dentin. However, dentin DSP protein has been estimated at 5-8% of the dentin NCP content (16), and PP has been estimated to be Ͼ50% of dentin NCP content (17). To date, this discrepancy between the observed 1:6 DSP/PP dentin protein ratio and the expected 1:1 ratio has not been explained. A key missing element in this story has been the inability to identify a DSP-PP precursor protein from any cell or tissue sample, and, without this putative DSP-PP precursor protein, it has not been possible to study DSP-PP post-translational processing and cleavage, leaving unanswered such questions as where DSP-PP cleavage occurs (i.e. intracellularly or extracellularly) and what cleavage enzyme(s) may be involved.
To answer these DSP-PP protein processing questions, we utilized a baculovirus expression system to produce recombinant DSP-PP precursor proteins from a DSP-PP 240 cDNA, which represents one of several endogenous DSP-PP transcripts (18,19) believed to play different roles during dentin mineralization. Our in vitro results demonstrate that DSP-PP 240 precursor proteins are produced by this system and are capable of self-processing to yield both DSP and PP proteins. 240 cDNA Using the pVL941 and pVL1392 Baculovirus Expression Systems-A diagram of the DSP-PP gene, showing the relative positions of DSP-PP 240 cDNA and DSP 370 is shown in Fig. 1A. DSP-PP 240 cDNA contains the 17-amino acid leader sequence, DSP-PP 240 coding sequence, a stop codon, and a 200-bp 3Ј noncoding sequence (see Fig. 1A). DSP-PP 240 cDNA was subcloned into the baculovirus expression vector pVL1392 at XbaI and BamHI sites. The DSP 370 cDNA, containing a signal sequence encoding the 17-amino acid leader sequence and the partial DSP coding sequence for the first 370 amino acids, was sub-cloned into the baculovirus expression vector pVL941 at the BamHI site. This DSP 370 cDNA construct yielded a fusion protein, which contains a 370-amino acid DSP peptide sequence as well as an additional 18-amino acid peptide derived from the viral sequence (see Fig. 1C). An antisense DSP 370 cDNA construct was produced as a control.

Construction and Expression of DSP 370 cDNA and DSP-PP
The pVL941-DSP 370 cDNA, pVL941-antisense-DSP 370 cDNA, and pVL1392-DSP-PP 240 cDNA constructs were individually cotransfected with a linearized BaculoGold baculovirus DNA (BaculoGold transfection kit; PharMingen, San Diego, CA) (20) into insect Sf9 cells to obtain virus stock. To produce recombinant proteins, insect Sf9 cells infected with recombinant virus stock at a multiplicity of 10, were grown in Grace's insect cell medium (Invitrogen) supplemented with 10% fetal calf serum to a density of 2 ϫ 10 6 cells/T25 flask. Supernatants were harvested on days 1, 2, 3, and 4 after infection and partially purified using a polyanion extraction protocol (see below).
Partial Purification of Recombinant DSP-PP 240 and DSP 370 Proteins Using Polyanion Extraction-This protocol takes advantage of the finding that acidic proteins, such as DSP and PP, are soluble in 5% trichloroacetic acid (21). For DSP 370 purification, the supernatant from pVL941-DSP 370 cDNA-infected Sf9 cells was diluted 1:20 with 100% trichloroacetic acid. The majority of culture medium proteins were precipitated and removed by centrifugation. The trichloroacetic acid-soluble portion was further neutralized with one-fifth original volume of 3 M Tris-HCl, pH 8.8, and precipitated with one-tenth volume of 1 M CaCl 2 . This new precipitate was dissolved again in 5% trichloroacetic acid and precipitated with 3 M Tris-HCl, pH 8.8, and 1 M CaCl 2 . This second CaCl 2 precipitate, containing recombinant DSP 370 , was dissolved in one-tenth original volume of 0.1 M EDTA. Purified DSP-PP 240 was similarly obtained. The recombinant proteins were stored in 0.1 M EDTA at Ϫ20°C and were stable for 2 years.
SDS-Polyacrylamide Gel Electrophoresis-SDS-PAGE was performed using 7.5, 10, and 4 -15% polyacrylamide gels. Samples were dissolved in Laemmli sample buffer (Bio-Rad). Electrophoresis was carried out at 60 mA for 45 min. The gels were stained with Bio-Safe Coomassie Blue R250 (Bio-Rad) or Stains-All (Sigma). The apparent molecular weights of the protein bands were estimated by comparison with Bench Mark prestained protein ladder standards (Invitrogen). The gels were air-dried in a cellophane membrane overnight. Stains-All stains acidic proteins (i.e. DSP or PP) blue, whereas neutral proteins (such as bovine serum albumin) appear orange-red.
Gel Purification of Recombinant DSP-PP 240 Precursor Proteins-The partially purified recombinant DSP-PP 240 precursor proteins were electrophoresed on 7.5 or 10% SDS-polyacrylamide gels. The DSP-PP 240 band was excised, electroeluted, and concentrated. All steps were performed in the presence of a protease inhibitor IP mixture (Sigma).
Highly Phosphorylated Protein Preparation-Highly phosphorylated protein (HP) was prepared from rat incisors following the method of Marsh (23). The purified HP contained 2.9 nmol of P i /g of HP, and the N-terminal sequence was determined as DDPN.
Rabbit Anti-rat DSP-Rat DSP peptide (NH 2 -CPSGQSQN-QGLETEGSSTGN-COOH) was synthesized and purified by reverse phase high pressure liquid chromatography (Genemed Synthesis Inc., San Francisco, CA). This peptide was then conjugated to keyhole limpet hemocyanin and used for generating rabbit anti-rat DSP antibodies. These anti-DSP antibodies (1:200 dilution) were used to perform Western blot analyses to identify the expressed recombinant DSP proteins.
Western Blot Analyses-Proteins were electrophoresed and transferred to nitrocellulose filters using a semidry apparatus. The nitrocellulose filters were hybridized with 1% blocking agent to block nonspecific antibody binding and then incubated with a 1:200 dilution of primary antibodies (i.e. rabbit anti-rat DSP antibodies) overnight at 4°C. The nitrocellulose filters were washed with TBS three times and incubated with secondary antibodies (goat anti-rabbit antibodies conjugated with alkaline phosphatase at a 1:2000 dilution) for 3 h. The filters were washed, and nitro blue tetrazolium/5-bromo-4-chloro-3indolyl phosphate (Bio-Rad) was added for color development.
N-terminal Amino Acid Sequence Analysis-After gel electrophoresis separation, the proteins were transferred onto Immobilon-P (polyvinylidene difluoride) membrane (Millipore) by the semidry technique in CAPS-methanol buffer (0.8 mA/cm, 1-2 h). After transfer, the membranes were washed and lightly stained with Stains-All, and the protein bands were excised. The N-terminal amino acid sequence of the proteins was determined by Procise Protein Sequencer 494 HT (Applied Biosystems, Foster City, CA) at the Protein Core Facility (University of Michigan), using reagents and methods recommended by the manufacturer.
Mass Spectral Analyses-Gradient 4 -15% SDS-polyacrylamide gel samples were stained with Stains-All and then excised, transferred to a 96-well plate, and destained. The gel samples were then subjected to reduction and alkylation and then washed, dehydrated, and digested with trypsin using a MassPrep robot. The peptides were extracted from the gel plugs with 2% acetonitrile and 1% formic acid. The extracted peptides (30 l) were transferred to another 96-well plate, where 5 l of matrix (␣-Cyano) was added to the sample well. The samples were then vaporized to dryness and redissolved in 5 l of 60% acetonitrile and 0.1% trifluoroacetic acid. Peptide samples were then spotted on a MALDI-TOF/TOF target plate for MS and MS/MS analyses. MS/MS, or tandem mass spectrometry, is a mass spectrometric method in which a peptide is fragmented, and the masses of the resultant fragment ions are recorded in a spectrum. The analyses were performed using the ABI 4800 MALDI-TOF/TOF (Applied Biosystems, Foster City, CA) at the Michigan Protein Consortium. Searches for homologies between the amino acid sequences obtained and those of other known proteins in GenBank TM , GenPept, and SwissProt were performed using BLAST software. The Michigan Proteome Consortium provided proteomics data at the University of Michigan.
Gelatin Zymography-Zymogram gels, prepared with 7.5 or 10% SDS-polyacrylamide containing 0.1% gelatin, were used to detect and characterize protease activity in gel-purified and eluted DSP-PP 240 precursor protein and PP 240 protein samples. Protease activity was also examined in the purified native rat HP. Following electrophoresis, the gel was washed in renaturing buffer (50 mM Tris-HCl, 5 mM CaCl 2 , 2.5% Triton X-100, 0.02% NaN 3 , pH 7.5) with gentle agitation for 30 min at room temperature, equilibrated with developing buffer (50 mM Tris-HCl, 5 mM CaCl 2 , 1% Triton X-100, 0.02% NaN 3 , pH 7.5) for 30 min, replaced with fresh developing buffer, and incubated at 37°C overnight. The gel was then stained with Coomassie Blue, destained, and digitally scanned. 240 and DSP 370 Protein Profiles-Both DSP-PP 240 and DSP 370 baculovirus constructs contained leader sequences; thus, the baculovirus-derived recombinant proteins would be expected to be secreted into the insect culture medium. Insect Sf9 cells were infected with baculovirus containing either DSP-PP 240 cDNA or DSP 370 cDNA. After infection, the cells were incubated for 4 days, and the recombinant proteins in the harvested cell media were partially purified and concentrated using polyanion extraction, separated by SDS-PAGE, and stained with Stains-All (see "Materials and Methods"). The resulting protein profiles are shown in Fig. 2. Five major blue-staining bands (i.e. bands 1-5; BSA stains orange) were present in media obtained from DSP-PP 240 bacteriophage-infected cells, whereas three major blue-staining bands (i.e. bands 1Ј-3Ј; BSA stains orange-red) were present in the DSP 370 cell-infected medium.

Identification of Baculovirus-derived Recombinant DSP-PP
To characterize the identities of the 120 kDa (band 1) and 95 kDa (band 2) protein bands, the bands were transferred to a polyvinylidene difluoride membrane, excised, and subjected to Edman degradation for N-terminal amino acid sequence analyses. The N-terminal sequence for the 120 kDa band was determined to be IPVPQ, which corresponded correctly to that of the presumed DSP-PP 240 precursor protein.
The N-terminal sequence for the 95-kDa band was determined to be IPVPQ, suggesting that this band represented DSP. We further determined the identity of band 1 with mass spectra analyses. DSP-PP 240 cDNA encodes a 687-amino acid peptide, which contains a 17-amino acid signal peptide, as well as DSP and PP peptide sequences (see Fig. 3B). Mass spectra analyses of band 1, following trypsin digestion, identified a number of tryptic peptides across the presumed DSP-PP 240 precursor protein (see Fig. 3, A and B). The N-terminal sequence of mature PP is DDPN, located at amino acid positions 448 -451, and a PP matching amino acid sequence was detected at positions 524 -542 (i.e. DKDESDNSN-HDNDSDSESK). Thus, band 1 encompasses the DSP coding sequence starting with the N-terminal sequence IPVPQ and an additional six peptides (located between the DSP N-terminal and PP N-terminal sequences) as well as the PP sequence.
To identify the 33 kDa band 4 in Fig. 2, the recombinant protein profile generated from a DSP 370 cDNA construct was compared with that generated from a DSP-PP 240 cDNA construct (Fig. 2, lanes 1 and 2). The DSP 370 cDNA construct encodes a recombinant protein with a size of 388 amino acids (i.e. 370-amino acid DSP protein plus an additional 18amino acid virus-derived sequence). Band 2Ј represents the 388-amino acid DSP protein, which was recognized by anti-DSP antibody (not shown). This DSP 370 recombinant protein does not contain a PP sequence and is shorter than band 2. From Fig. 2, band 4 (33 kDa) and band 5 (29 kDa) are not present in the DSP 370 profile. Thus, these bands most likely represent two PP-related proteins.
As shown in Fig. 2, blue band 2 present in the DSP-PP 240 lane has a higher molecular weight than blue band 2Ј (containing 388 amino acids) present in the DSP 370 lane. Thus, this blue band 2 probably represents the DSP 430 protein, which is composed of DSP 350 and dentin glycoprotein (DGP 80 ). In Fig. 2, blue band 1Ј (180 kDa), present in the DSP 370 lane, most likely represents a dimer of DSP (band 2Ј; 90 kDa). Blue band 3 (70 kDa) and blue band 3Ј, present in both DSP-PP 240 and DSP 370 lanes, respectively, in Fig. 2, were not always present in recombinant protein samples. Mass spectra analyses suggest that blue band 3 is a telokin-like-20 protein (i.e. baculovirus related; data not shown). Thus, blue band 3 could be produced by the baculovirus system. Smaller weak blue bands between 22 and 6 kDa were present in both DSP-PP 240 -derived products and DSP 370 -derived products. These bands are probably derived from DSP.
Dynamic Processing of DSP-PP 240 Protein-Sf9 cells were infected for 4 days with baculovirus containing DSP-PP 240 cDNA. At various times following infection, recombinant DSP-PP 240 and related proteins were purified from the culture medium and run on a 10% SDS-PAGE, which was then stained with Stains-All. After 2 days of infection, bands located at 120 kDa (i.e. DSP-PP 240 ) and 33 kDa (i.e. PP 240 ) were observed (Fig. 4A, lane 2). These two bands increased in intensity on days 3 and 4 (Fig. 4A,  lanes 3 and 4), with two additional minor bands appearing at 95 kDa (i.e. DSP) and 70 kDa (equivalent to blue band 3 in Fig. 2) on day 4. A Western blot, using anti-DSP antibodies of day 4 infected media, recognized both the 120 and 95 kDa bands, confirming that they both contain DSP.

The Purified DSP-PP 240 Precursor Protein Can Undergo Processing in the Absence of Insect Cell Conditioned Medium-Our
results thus far demonstrate that recombinant DSP-PP 240 protein can be processed to produce both DSP 430 and PP 240 peptides over time (see Fig. 4). To determine whether this processing was due to proteases that were present in the insect culture medium, we obtained an SDS-PAGE-purified DSP-PP 240 protein via electroelution (see "Materials and Methods") and incubated this purified precursor in 25 mM Tris-HCl, pH 7.5, for 30 min at 37°C. The resulting protein profile is shown in Fig. 5. The initial electroeluted DSP-PP 240 is present as a single band (Fig. 5, lane 1), which then undergoes significant protein processing within 30 min to yield a DSP 430 band and a PP 240 band (Fig. 5, lane 2). Thus, DSP-PP 240 precursor protein processing is probably not due to proteases in the insect medium, suggesting that it may instead undergo self-processing.
These findings are summarized in Fig. 6. Here we show that the DSP-PP 240 precursor protein (120 kDa), following secretion into the culture medium, was further processed into a 95-kDa DSP protein (containing 430 amino acids) and a major 33-kDa PP protein (containing 240 amino acids). The DSP 370 protein profile contains a secreted 90-kDa protein (containing 388 amino acids with 370 DSP amino acids and 18 viral amino acids; band 2Ј) and no PP bands. Fig. 6 also displays the cleavage site responsible for removing the leader sequence from the DSP-PP 240 precursor protein and the cleavage site responsible for generating DSP 430 and PP 240 .
DSP-PP 240 Has Gelatinolytic Activity-Our results demonstrate that DSP-PP 240 precursor protein is capable of self-processing to yield both DSP 430 and PP 240 . To test whether this process could be due to proteolysis, gel-purified DSP-PP 240 and PP 240 were electrophoresed on 10% SDS-PAGE gels containing  0.1% gelatin as a proteolytic substrate. Following electrophoresis, the gels were incubated overnight in renaturing buffer (see "Materials and Methods") and then stained with Coomassie Blue. The destained gels showed two white bands at 120 kDa (i.e. DSP-PP 240 ) and 33 kDa (i.e. PP 240 ) (see Fig. 7, A (lane 1) and B (lane 1)) indicative of gelatinolytic activity. In the control groups, no white bands were observed in polyanion extracts of Sf9 insect cell conditioned medium and in polyanion extracts of media derived from Sf9 insect cells infected with the baculovirus containing antisense DSP 370 cDNA (see Fig. 7, A (lanes 2 and 3) and B (lane 2)). This is the first evidence that DSP-PP and PP possess proteolytic activity.
Native HP Also Exhibits Protease Activity-Since DSP-PP 240 and PP 240 possess proteolytic activity, as shown in Fig. 7, purified rat incisor native HP protein (23) was next tested to determine whether naturally occurring HP was capable of digesting gelatin. HP is equivalent to isoform PP 523 , which is derived from the DSP-PP 523 transcript. Rat HP, isolated using the polyanion extraction method, was analyzed by Edman degradation. Its N-terminal sequence was identified as DDPNS-SDESNGSD (24), indicating that it was free of contamination by other proteins. As shown in Fig. 8, lanes 2 and 3, both native HP and heat-denatured HP were stained blue by Stains-All staining. Coomassie Blue did not stain native HP or heat-denatured HP (Fig. 8, lanes  5 and 6). When native rat HP was electrophoresed along with heat-denatured HP on a 7.5% SDS-polyacrylamide gel containing 0.1% gelatin, after a 3-h incubation in renaturing buffer, followed by Coomassie staining, single proteolytic bands were observed around 90 kDa for native rat HP but not for heatdenatured HP (Fig. 8, lanes 7 and 8). Furthermore, under the same conditions, but in the presence of 0.1 M EDTA, no clear band was detected in either the native HP or denatured HP lanes (Fig. 8, lanes 9 and 10). Taken together, these data demonstrate that both PP 240 and PP 523 exhibit proteolytic activity.

DISCUSSION
To date, three different DSP-PP transcripts (i.e. DSP-PP 523 , DSP-PP 240 , and DSP-PP 171 ), giving rise to three PP isoforms, have been identified in rat tooth extracts (1,18,19). These PP isoforms, which include PP 523 , PP 240 , and PP 171 , are speculated to play different roles during tooth development and mineral-FIGURE 6. Production and processing of recombinant DSP-PP 240 and DSP 370 proteins in a baculovirus expression system. A, recombinant DSP-PP 240 precursor protein, containing a leader sequence, was synthesized from DSP-PP 240 cDNA in a baculovirus expression system. The leader sequence was removed, and DSP-PP 240 precursor protein was secreted into the culture medium (see Fig. 2). During a 30-min incubation at 37°C in 25 mM Tris-HCl, pH 7.5, the DSP-PP 240 precursor protein was further processed into DSP 430 and PP 240 (see Fig.  5). The DSP-PP 240 precursor protein leader sequence is depicted in gray, and the N-terminal sequence is depicted in black boldface letters. The gray triangle indicates the cleavage site responsible for leader sequence removal. SYDEDDESMQGDDPNSSDESN represents the junction amino acid sequence between the DSP and PP sequence. Within this sequence, boldface letters represent amino acids from the DSP sequence, and italic letters represent amino acids from the PP sequence. The black triangle indicates the cleavage site responsible for generating DSP 430 and PP 240 . IPVPQL . . . SYDEDDESMQG represents DSP 430 N-terminal and C-terminal sequences. DDPNSS...GSDSNHSTSDD represents PP 240 N-terminal and C-terminal sequences. B, a recombinant DSP 370 protein-viral sequence (with a leader sequence) was synthesized from DSP 370 cDNA in a baculovirus expression system. The leader sequence was removed, and DSP 370 protein-viral sequence was secreted into the culture medium. The size of the DSP 370 protein-viral sequence (i.e. 388 amino acids) is smaller than DSP 430 , and no PP-related bands were present in the recombinant DSP 370 protein gel profile (see Fig. 2). ization by helping to fine tune mineral nucleation and hydroxyapatite growth at different stages of the mineralization program (19). We used DSP-PP 240 infected Sf9 insect cells to produce and secrete a recombinant 120-kDa protein product into the conditioned cell medium (Fig. 2, band 1). MS and MS/MS analysis identified a number of recognizable tryptic peptides across the DSP portion of the presumed DSP-PP 240 precursor protein, and the detection of a PP 240 matching amino acid sequence at positions 524 -542 (i.e. DKDESDNSNHDNDSDSESK) demonstrated that the 120-kDa protein band contains both DSP and PP 240 sequences. MS identification of the N-terminal DSP-PP 240 amino acid sequence IPVPQLVPLER confirmed that (i) it is likely that the signal peptide sequence (i.e. MKTKIIIYICI-WATAWA) was cleaved from the nascent DSP-PP 240 peptide in the endoplasmic reticulum during the secretory process and (ii) that the DSP-PP 240 precursor protein was secreted into the extracellular medium. These data demonstrate for the first time that DSP-PP 240 transcripts are capable of producing and secreting full-length DSP-PP 240 proteins into the extracellular space. Moreover, we were able to produce recombinant DSP-PP 240 precursor protein in sufficient quantities to allow us to follow its processing over time using standard SDS-PAGE followed by Stains-All staining.
In Vitro DSP-PP 240 Protein Processing-When we infected Sf9 cells with baculovirus containing DSP-PP 240 cDNA, we were able to identify not only the DSP-PP 240 precursor protein in the conditioned medium but also found DSP 430 , PP 240 , and minor amounts of PP 211 . During the 4-day incubation period, both DSP-PP 240 and PP 240 bands increased in intensity. DSP 430 appeared on days 3 and 4 but was significantly weaker than the PP 240 band (Fig. 4A). Interestingly, dentin DSP protein has been estimated at 5-8% of the dentin NCP content, and PP has been estimated to be Ͼ50% of dentin NCP content (16,17). Therefore, the actual DSP/PP ratio in the dentin matrix is estimated to be 1:6 rather than the expected 1:1 ratio. Using the NIH Image J program, we determined the relative densities of recombinant DSP 430 and PP 240 from day 4 culture medium to be ϳ1:6 (see Fig. 4B). Thus, our in vitro DSP-PP processing results agree very well with the measured DSP/PP ratio in dentin tissue.
The cleavage of DSP-PP in baculovirus conditioned medium initially prompted us to consider that dentin matrix metalloproteases (MMPs) might be responsible for this proteolytic activity, since Yamakoshi et al. (25) recently reported on the ability of dentin-resident MMPs to cleave DSP-DGP, where DSP-DGP is a proteoglycan having 457 amino acids. According to these authors, porcine DSP-DGP-PP is first cleaved on the N-terminal side of Asp 458 to split DSP-DGP (equivalent to our DSP 430 ) from PP. They also claimed that this cleavage is rapid, since they were unable to detect intact DSP-DGP-PP protein in the dentin matrix. Without DSP-DGP-PP to use as a substrate, they were unable to identify the protease responsible for the proposed cleavage at Asp 458 . However, they were able to identify 12 different cleavage products from developing porcine molars by N-terminal sequencing. They then compared these fragments with fragments generated from DSP-DGP digested under in vitro conditions with either MMP-2 or MMP-20. They found that both MMP-2 and MMP-20 were capable of cleaving DSP-DGP at specific sites in vitro similar to those identified from in vivo isolations of low molecular weight DSP and DGP. Thus, they concluded that MMP-20 cleaved DSP-DGP from both ends, and MMP-2 cleaved DSP-DGP within the DSP C-terminal region as well as within the DGP region. However, as shown by Jo et al. (26), the baculovirus-Sf9 insect cell system expresses neither gelatinolytic MMP2 or TIMP-2. Furthermore, during our polyanion extraction of Sf9-conditioned culture medium, we found no 72-or 60 -65-kDa protein bands stained with Coomassie Blue that would suggest the presence of MMP-2. And we were able to show that SDS-PAGE-isolated DSP-PP 240 could undergo cleavage in a Tris-buffered salt solution. Thus, it is unlikely that MMP2 participates in DSP-PP 240 precursor protein cleavage in the baculovirus system. DSP-PP 240 , PP 240 , and HP (PP 523 ) Proteolytic Activity-Our in vitro studies, using purified recombinant DSP-PP 240 , demonstrate that the major cleavage products are DSP 430 and PP 240 (see Fig. 5). Edman degradation, Western blot analysis, and comparison of the SDS-PAGE protein profiles derived from DSP-PP 240 cDNA-and DSP 370 cDNA-transfected Sf9 cells support our findings that DSP 430 and PP 240 are products of the proteolytic cleavage of DSP-PP 240 . As mentioned above, our DSP 430 is equivalent to porcine DSP-DGP. We found that no DSP 350 and no DGP cleavage products were produced during the 30-min incubation time used to cleave our purified DSP-PP 240 precursor protein. We also found that no DSP 350 and no DGP cleavage products were produced in 4-day culture medium. Using 0.1% gelatin gels, we also found that DSP-PP 240 , PP 240 , and native HP (i.e. PP 523 ) were capable of degrading gelatin.
Because PP does not stain with Coomassie Blue, there is a possibility that the concentrated PP on the gelatin gel might yield a clear band. To test whether, after Coomassie Blue staining, the clear HP band present on the gelatin gel was caused by the inability of HP to be stained by Coomassie Blue, we ran both native HP and heat-denatured HP on a 0.1% gelatin gel. As shown in Fig. 8, lane 7, after a 3-h incubation in renaturing buffer, 0.5 g of native HP showed a clear band in gelatin zymography, whereas heat-denatured HP displayed no clear band (Fig. 8, lane 8). This finding suggests that native HP possesses proteolytic activity that can be inactivated by heating at 95°C for 5 min. Furthermore, when both native HP and heatdenatured HP were incubated with renaturing buffer in the presence of 0.1 M EDTA, no clear bands were detected in gelatin zymography (Fig. 8, lanes 9 and 10). These data also demonstrate that EDTA can inhibit HP proteolytic activity. These studies demonstrate that the HP clear band detected in gelatin zymography is due to HP proteolytic activity and is not due to the inability of Coomassie Blue to stain HP.
Taken together, these studies suggest that newly synthesized DSP-PP precursor proteins (derived from three DSP-PP multiple transcripts) undergo a rapid self-processing step, which yields DSP 430 plus the associated PP isoforms (i.e. PP 171 , PP 240 , and PP 523 ). Perhaps, over a longer time period, MMP-2 present in dentin then acts on DSP 430 to yield DSP 350 and DGP 80 .
Developmental Implications-DSP-PP promoter-driven LacZ expression appears in kidney in postnatal day 3 transgenic mice (15), in alveolar bone in newborn mice prior to its appearance in the incisor (15), and in salivary glands obtained from newborn mice. 3 Our current data, demonstrating that DSP-PP 240 , PP 240 , and PP 523 all exhibit proteolytic activity, suggest that these proteins may play important roles in tissue modeling during organ development.