DSP-PP Precursor Protein Cleavage by Tolloid-Related-1 Protein and by Bone Morphogenetic Protein-1

Dentin sialoprotein (DSP) and phosphophoryn (PP), acidic proteins critical to dentin mineralization, are translated from a single transcript as a DSP-PP precursor that undergoes specific proteolytic processing to generate DSP and PP. The cleavage mechanism continues to be controversial, in part because of the difficulty of obtaining DSP-PP from mammalian cells and dentin matrix. We have infected Sf9 cells with a recombinant baculovirus to produce large amounts of secreted DSP-PP240, a variant form of rat DSP-PP. Mass spectrometric analysis shows that DSP-PP240 secreted by Sf9 cells undergoes specific cleavage at the site predicted from the N-terminal sequence of PP extracted from dentin matrix: SMQG447↓D448DPN. DSP-PP240 is cleaved after secretion by a zinc-dependent activity secreted by Sf9 cells, generating DSP430 and PP240 products that are stable in the medium. DSP-PP processing activity is constitutively secreted by Sf9 cells, but secretion is diminished 3 days after infection. Using primers corresponding to the highly conserved catalytic domain of Drosophila melanogaster tolloid (a mammalian BMP1 homolog), we isolated a partial cDNA for a Spodopotera frugiperda tolloid-related-1 protein (TLR1) that is 78% identical to Drosophila TLR1 but only 65% identical to Drosophila tolloid. Tlr1 mRNA decreased rapidly in Sf9 cells after baculovirus infection and was undetectable 4d after infection, paralleling the observed decrease in secretion of the DSP-PP240 processing activity after infection. Human BMP1 is more similar to Sf9 and Drosophila TLR1 than to tolloid, and Sf9 TLR1 is more similar to BMP1 than to other mammalian homologs. Recombinant human BMP1 correctly processed baculovirus-expressed DSP-PP240 in a dose-dependent manner. Together, these data suggest that the physiologically accurate cleavage of mammalian DSP-PP240 in the Sf9 cell system represents the action of a conserved processing enzyme and support the proposed role of BMP1 in processing DSP-PP in dentin matrix.


Introduction
Dentin sialoprotein (DSP) and phosphophoryn (PP) are two major noncollagenous dentin proteins derived from a single copy of DSP-PP gene (also referred to as DSPP) whose expression is tightly regulated during dentinogenesis [1]. DSP-PP transcripts are also expressed in other tissues, including inner ear and jaw tissue [2,3]. The recent demonstration of DSP-PP promoter-driven lacZ expression in multiple tissues, such as bone, kidney [4], hair follicles [5,6], salivary gland, and lung (Ritchie unpublished data), further suggests that DSP and PP may have physiological roles in several other organs besides teeth.
It is believed that DSP-PP precursor proteins are the immediate translation products of DSP-PP mRNA (here we use ''DSP-PP'' to refer generically to the precursor secreted in dentin matrix and ''DSP-PP 240 '' to refer specifically to the shorter variant form of rat DSP-PP used in these experiments.). DSP-PP undergoes multiple post-translational modifications, including signal peptide cleavage, Asn-linked glycosylation, phosphorylation and proteolytic processing between the DSP and PP domains, to produce mature DSP and PP proteins required for dentin mineralization. Recent research on DSP-PP cleavage has focused on the identification of its initial cleavage site and the protease(s) responsible for cleavage. For example, Qin et al. reported Y438 as the major cleavage site (2001) using tryptic fragments from native, purified DSP [7]. Qin and colleagues proposed that a transmembrane endopeptidase, Phex, was the protease responsible for processing of both DMP1 (e.g., dentin matrix protein 1) and DSP-PP [8,9]. More recently, Sun et al. reported that a D448A mutation blocked cleavage of recombinant mouse DSP-PP in a cultured human cell system and concluded that the key cleavage site is G 447 QD 448 [10]. However, the exact cleavage site of the wild type (wt) mouse DSP-PP precursor in cultured cells was not determined by direct sequencing or mass spectrometry (MS); rather identification of cleavage products relied on gel mobility of immuno-stained or Stains-All-stained bands separated by SDS-PAGE. Thus, the D448A mutant also did not provide direct evidence that G 447 QD 448 was the cleavage site.
Other attempts to localize the initial DSP-PP cleavage site have met with limited success because of the lack of sufficient DSP-PP precursor protein available for quantitative studies. Steiglitz et al. proposed in 2004 bone morphogenetic protein-1 (BMP1) as the candidate enzyme for proteolytic cleavage of both DMP1 and DSP-PP [11]. This work demonstrated that BMP1 was responsible for DMP1 cleavage but did not test DSP-PP cleavage because the DSP-PP precursor protein was not available [11]. Recently, Fisher and co-workers identified a mouse DSP-PP precursor protein by expression in transfected human colon carcinoma LoVo cells which lack furin and therefore should be defective in BMP1 activation [12]. LoVo cells were shown to secrete an intact mouse DSP-PP precursor that could be cleaved by exogenously added BMP1 and by the homologous proteins tolloid-like 1 (TLL1) and tolloid-like 2 (TLL2) to generate a DSP-sized band, as shown by Western blot analysis [12]. However, because of the low amounts of DSP-PP and DSP that were detected, no mass spectrometry or N-terminal sequence data were available to identify the precise cleavage site. A mutant form of mouse DSP-PP (with a substitution equivalent to M 445 Q 446 to I 445 E 446 in rat DSP-PP 240 ) was not cleaved by BMP1, TLL1 or TLL2. Again, analysis of this mutant did not provide direct evidence that G 447 QD 448 was the cleavage site. Yamakoshi and co-workers [13], showed that a DSP-PP preparation obtained from porcine dentin matrix was cleaved by the matrix metalloproteinases MMP-2 and MMP-20 at multiple sites making them unlikely candidates for processing enzymes. More recently, similar preparations were shown to be cleaved by BMP1 and MEP1A to generate a product similar in mobility to PP [14]. Taken together, despite intensive investigations since 2001, there is no direct evidence proving that G 447 QD 448 is the DSP-PP cleavage site and only Western blot analysis to substantiate the claim that BMP1 can correctly cleave DSP-PP.
Previously, we used a baculovirus expression system that was capable of producing high yields of DSP-PP precursor protein.
The secreted DSP-PP 240 protein could be visualized with Stains-All staining and could be identified unambiguously by mass spectrometry [15]. From MS/MS analysis of isolated tryptic fragments of the PP 240 band, we proposed that the DSP-PP cleavage site was G 447 QD 448 [15]. In this report, we used MS and MS/MS to analyze a smaller, chymotryptic fragment of the PP 240 band which has permitted direct determination of the amino acid sequence of this peptide by ion trap/fragmentation MS and firmly establishes that the initial cleavage site in DSP-PP 240 is G 447 QD 448 .
We also show that cleavage of DSP-PP 240 at this site occurs after secretion into the conditioned medium of Sf9 cells and that cleavage is catalyzed by an endogenous Zn-dependent proteolytic activity secreted by Sf9 cells. Secretion of this activity is suppressed by baculoviral infection. We show, furthermore, that Sf9 cells transcribe a tolloid-related-1 (TLR1) peptidase gene (Spodoptera frugiperda tlr1) and that expression of this gene is similarly suppressed by baculoviral infection. We show that the human homolog of tolloid-related-1, BMP1, can also cleave DSP-PP 240 to release DSP 430 and PP 240 .

Materials and Methods
Mass Spectrometric (MS) Analysis of PP 240 to Determine the Cleavage Site in Secreted DSP-PP 240 PP 240 was partially purified by the polyanion extraction procedure as previously described [15]. Briefly, conditioned medium (5 ml) was collected from 1610 6 Sf9 cells (Novagen, Madison, WI) 4 days after infection with baculovirus containing the DSP-PP 240 cDNA (hereafter termed DSP-PP 240 virus) [15]. Trichloroacetic acid (TCA) was added to the medium to a final concentration of 5% (w/v) and incubated at room temperature for 2 min to precipitate the majority of culture medium proteins. The TCA-soluble fraction, collected after centrifugation in an Eppendorf rotor at 12,000 rpm at room temperature for 5 min, was neutralized with one-fifth volume of 3 M Tris-HCl, pH 8.8, and precipitated with one-tenth volume of 1 M CaCl 2 . The calcium precipitate, collected after centrifugation in an Eppendorf rotor at 12,000 rpm for 5 min at room temperature, was dissolved in 1 ml of 5% trichloroacetic acid, neutralized with 3 M Tris-HCl, pH 8.8, and re-precipitated with 1 M CaCl 2 . This second CaCl 2 precipitate, containing recombinant DSP-PP 240 and products of processing, was dissolved in 0.5 ml of 0.1 M EDTA. The dissolved sample (20 ml) was fractionated on a non-denaturing polyacrylamide gel (7.5%). PP 240 protein was stained with Stains-All and analyzed by MS and MS/MS analyses as described below.
7.5% non-denaturing polyacrylamide gel samples were stained with Stains-All and then PP 240 band was 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 chymotrysin. The peptides were extracted from the gel plugs with 2% aceonitrile and 1% formic acid. The extracted peptides (30 ml) were transferred to another 96-well plate, where 5 ml of matrix (a-Cyano) was added to the sample well. The samples were then vaporized to dryness and redissolved in 5 ml of 60% aceonitrile and 0.1% trifuoroacetic 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.
Preparation of 0-day to 3-day Conditioned Medium (CM 0-3d ) Sf9 cells cultured at 28uC in T-25 flasks in 5 ml of Grace's medium with 10% FBS and 50 mg/ml of Gentamycin were infected with DSP-PP 240 virus for 3 days. A mock infection served as a control. Conditioned medium (CM) was harvested by centrifugation (5006g). CM containing DSP-PP 240 precursor protein was designated ''CM 0-3d with viral infection'' and CM from uninfected Sf9 cells was designated ''CM 0-3d without viral infection''.

Preparation of 3-day to 7-day Conditioned Medium (CM 3-7d )
Sf9 cells culture and infection with DSP-PP 240 virus as described above. After three day viral infection, the conditioned medium was removed and discarded. Cells were then gently washed with 16PBS and cultured in 5 ml of fresh Grace's medium with10% FBS and 50 mg/ml Gentamycin for an additional 4 days (from day 3 to day 7 after infections) at 28uC. Then the CM, containing largely uncleaved DSP-PP 240 precursor protein, was collected on day 7 by centrifugation (5006g) and designated CM 3-7d .
Testing for Proteolytic Activity in Condition Medium of Sf9 Cells CM 3-7d (5 ml) from virus-infected cells was incubated with CM 0-3d (5 ml) from uninfected cells for different time periods at 28uC. The CM 3-7d from infected cells contained uncleaved DSP-PP 240 while the CM 0-3d from uninfected cells contained endogenous proteolytic activity secreted by Sf9 cells. In parallel, CM 3-7d (5 ml) from virus infected cells was incubated with Grace's medium containing 10% FBS and 50 mg/ml Gentamycin (5 ml) as a control. At each time point, 1 ml samples of mixture was processed using a standard purification protocol [15]. Briefly, 1 ml medium was adjusted to 5% TCA, in which DSP-PP 240 , DSP 430 and PP 240 are soluble but other proteins in conditioned medium are precipitated, incubated at room temperature for 2 min and then centrifuges in an Eppendorf rotor at 12,000 rpm for 5 min at room temperature. The supernatant was neutralized with one-fifth volume of 3 M Tris-HCl pH 8.8 and then precipitated by adding one-tenth volume of 1 M CaCl 2 and centrifuging in an Eppendorf rotor at 12,000 rpm for 5 min at room temperature. The pellet from 1 ml medium was resuspended in 100 ml of 0.1 M EDTA. 15 ml of resuspended samples were resolved onto native PAGE as described previously [15], and the gels were fixed and stained with Stains-All to visualize DSP-PP 240 and the cleavage products DSP 430 and PP 240 . Stained gels were subjected to quantitative densitometry. Because the PP 240 portion of DSP-PP 240 is responsible for the majority of staining by the Stains-All reagent, percentage of protein processing was defined as PP 240 /(PP 240 + DSP-PP 240 ).

Identification of Tolloid-related-1 Protein mRNA in Sf9 Cells
We aligned tolloid protein sequences from Drosophila melanogaster (GeneBank accession number NM_079763), Culex quinquefasciatus (GeneBank accession number XM_001861486), Aedes aegypti (GeneBank accession number XM_001653501) and Tribolium castaneum (GeneBank accession number XM_965069), and obtained consensus sequences, [i.e., QAMRHWE and IMHYA(R/ K)N(T/S)) as shown in Fig. 1. Sense and anti-sense oligonucleotide primers were generated according to these consensus peptide sequences. Total RNA from Sf9 cells infected with DSP-PP 240 virus for 3 days was extracted with TRIzolH Reagent (Invitrogen, Carlsbad, CA). Using RT-PCR, tolloid-related-1 cDNA fragments were generated from this RNA preparation. After cloning into a TOPO cloning vector (Invitrogen, San Diego, CA), candidate clones were identified by PCR and confirmed by DNA sequencing. The cDNA sequences were then used to search the NCBI non-redundant protein sequence database for homologous peptide sequences.

BMP1 Cleavage of DSP-PP 240
CM 3-7d from virus infected cells (5 ml) was incubated at 28uC with an equal volume of unconditioned (fresh) Grace's medium in the presence of 33 ng/ml or 170 ng/ml of recombinant human BMP1 or with no enzyme addition for various times from 1 h to 72 h as indicated in the corresponding figure. At each of the time points, 1 ml of mixed medium was processed as described above. 15 ml of each resuspended sample was analyzed by native PAGE, fixed and stained with Stains-All and the percentage of protein processed was determined as described above.

Quantification of DSP-PP 240 and PP 240
Stained gels were dried, scanned to produce TIFF files and NIH ImageJ was used to quantify intensity of Stains-All stained bands. The stained protein band images were converted to grayscale and inverted and rectangular areas were used to integrate the intensity of the DSP-PP 240 and PP 240 bands. These intensity values were then used to calculate the ratio of PP 240 /(DSP-PP 240 +PP 240 ). Each experiment was conducted at least 2 times and in most cases 3 or more times with comparable results. Plotted PP 240 /(DSP-PP 240 +PP 240 ) ratios expressed as a percentage, represent the mean of at least two experiments and error bars represent the standard deviation of the mean. Data were analyzed and plotted using KaleidaGraph (Synergy Software, Reading, PA).

DSP-PP 240 Expressed in Sf9 Cells is Cleaved in
Conditioned Medium at the Physiological Processing Site: SMQG 447 QD 448 DPN Previously, we showed that full-length DSP-PP 240 can be expressed in Sf9 cells using a baculovirus vector and that the precursor is partially processed into stable fragments of sizes expected for the physiological products DSP 430 and PP 240 [15]. Based on mass spectrometric analysis, we identified a 7.7 kDa polyacrylamide gel band from trypsinized recombinant PP 240 that likely represented a 76 residue peptide containing the presumed N-terminal sequence of PP 240 . Due to its large molecular mass, it was not possible to determine the sequence of the peptide directly [15]. To overcome this problem, we used chymotrypsin digestion of baculovirus-expressed PP 240 , which would be expected to produce a 34 amino acid peptide corresponding to the N-terminus of PP 240 (DSP-PP 240 residues 448-481) if the precursor was This finding demonstrates that baculovirus-produced recombinant PP 240 protein starts with DDPN, which corresponds to the Nterminal sequence of phosphophoryn purified from dental matrix (see Fig. 2). Thus cleavage of DSP-PP 240 precursor protein in the baculovirus expression system occurs at the physiological site, generating stable fragments that do not undergo further cleavage.
Processing of DSP-PP 240 into DSP 430 and PP 240 Occurs after Secretion into Conditioned Medium and is Catalyzed by a Proteolytic Activity Secreted by Sf9 Cells  (Fig. 3A, lane 1, Fig. 3C). After an additional 3 day incubation, DSP-PP 240 precursor protein underwent further cleavage, resulting in accumulation of much higher levels of DSP 430 and PP 240 (Fig. 3A, lane 2). Because DSP 430 is weakly stained by Stains-All relative to PP 240 , the DSP 430 band was barely visible, and PP 240 alone was therefore used as a quantitative marker for DSP-PP 240 protein cleavage. Quantification of DSP-PP 240 and PP 240 bands showed that 18% of the DSP-PP 240 was cleaved to PP 240 in the CM 0-3d sample and that this increased to 60% after additional 3 day incubation, indicating that significant additional DSP-PP cleavage occurred during the extended 3 day incubation period. Thus DSP-PP 240 cleavage largely occurred after secretion into conditioned medium and that the conditioned medium contained a proteolytic activity that was active during the additional 3 day incubation.
DSP-PP 240 in CM 3-7d undergoes little cleavage into DSP 430 and PP 240 after an additional 3-day incubation at 286C. To determine whether DSP-PP 240 production continued beyond the third day after infection, the culture medium of Sf9 cells infected with the DSP-PP 240 virus for 3 days was replaced with fresh medium and incubation was continued for 4 days. When this 3-to-7 day conditioned medium (CM 3-7d ) was analyzed, even larger amounts of DSP-PP 240 were present than that in CM 0-3d (compare Fig. 3A, lane 1 to Fig. 3B, lane 1), indicating that DSP-PP 240 production continued during the additional 4 day incubation period. However, the PP 240 cleavage product band was barely visible, representing only about 15% cleavage of the precursor (Fig. 3C). Even after an additional 3 day incubation in the same medium, the PP 240 band was still barely visible (Fig. 3B, lane 2), implying that no significant additional cleavage occurred (Fig. 3C). These results demonstrate that, whereas DSP-PP 240 processing was dramatically increased with the extended 3 day incubation of the CM 0-3d sample, DSP-PP 240 secreted into fresh medium after 3 days of infection (the CM 3-7d sample) was largely uncleaved and remained so even when incubated for an additional 3 days (Fig. 3B). Thus the CM 3-7d sample appeared to lack the DSP-PP processing activity.
CM 0-3d from uninfected Sf9 cells contains a proteolytic activity that promotes DSP-PP 240 precursor protein cleavage. Because DSP-PP 240 present in the CM 3-7d media did not undergo cleavage even during extended incubation periods, we used this DSP-PP 240 sample as substrate to test for the production of a DSP-PP 240 processing activity by conditioned medium of uninfected Sf9 cells. To do this, we mixed (1:1) CM 3-7 from infected cells with medium conditioned by incubation for three days with uninfected Sf9 cells and incubated the mixture at 28uC. CM 3-7d from infected cells was incubated with Grace's medium as a control. Fig. 4 shows that, as a function of time, incubation of CM 3-7d from infected cells with CM 0-3d medium from uninfected Sf9 cells resulted in progressive cleavage of DSP-PP 240 to PP 240 that was roughly linear for the first 3 days of incubation and reached near completion (,80% cleavage) after 6 days (Fig. 4, upper curve). Little cleavage was seen in the control containing unconditioned Grace's medium (Fig. 4, lower curve). Therefore, Sf9 cells secrete an activity that correctly processes DSP-PP 240 , but secretion of this activity is greatly diminished or suppressed within 3 days after virus infection.

Identification of a Tolloid-related-1 Transcript in Sf9 Cells
Mammalian BMP1 was reported to be a protease capable of cleaving several extracellular matrix proteins, including the pro-a1 precursors of type I, type II, type III, and type VII collagen, the pro-a2 precursors of type I and type V collagen, the human prolysyl oxidase, and DMP1 [11,[16][17][18][19][20][21]. Because the cleavage site of DMP1 exhibits high sequence similarity to the putative cleavage site in DSP-PP [11], we asked whether a BMP1 equivalent protein was present in Sf9 cells.
The proteins in Drosophila melanogaster most closely related to human BMP1 are Drosophila tolloid (TLD) and tolloid-related-1 (TLR1). We therefore investigated whether Sf9 cells contain tolloid-related-1 mRNA sequences. We first obtained consensus sequences by aligning tolloid sequences from Drosophila melanogaster, Culex quinquefasciatus, Aedes aegypti and Tribolium castan. Using this alignment, we designed consensus primers corresponding to the most highly conserved regions of the catalytic domains of these enzymes and performed RT-PCR on Sf9 cell mRNA. We obtained a partial cDNA sequence encoding an open reading frame with 65% peptide sequence identity to Drosophila tolloid protein (TLD) and 78% peptide sequence identity to Drosophila tolloid-related-1 protein (TLR1). Because of greater similarity to TLR1, we propose that this clone represents a partial cDNA of a Spodopteria frugiperda tlr1 mRNA. This cDNA encodes part of the astacin protease domain with the known consensus sequence HExxHxxGFxHExxRxDRD containing the zinc-binding motif, and another conserved region, SxMHY, the so-called Met-turn (see Fig. 5).
We next tested whether tlr1 mRNA was expressed by uninfected and DSP-PP 240 viral infected Sf9 cells. Using the sense and antisense primers specific for Sf9 tlr1 cDNA, we performed RT-PCR on RNA from uninfected and DSP-PP 240 viral infected Sf9 cells. In Sf9 cells without viral infection, tlr1 mRNA expression increased steadily from day 1 to day 4 (Fig. 7A). In Sf9 cells infected with DSP-PP 240 virus, tlr1 mRNA expression was highest at 1 day post-infection but dropped dramatically from day 2 to day 4 post-infection (Fig. 7B). The decrease seen in expression of Sf9 tlr1 mRNA after virus infection is consistent with the lack of secretion of DSP-PP 240 processing activity by 3 days after viral infection (Fig. 3B).

Cleavage of DSP-PP 240 Precursor Protein by BMP1
We next tested whether purified BMP1 could cleave DSP-PP 240 . For these studies we utilized CM 3-7d medium samples, which contained significant quantities of stable DSP-PP 240 precursor protein as substrate. Fig. 8 shows the kinetics of DSP-PP 240 cleavage by BMP1 added to the CM 3-7d substrate at concentrations of 33 ng/ml or 170 ng/ml. BMP1 cleaved DSP-PP 240 precursor protein in a concentration-dependent manner yielding mature PP 240 , while control incubation with Grace's medium yielded minimal cleavage.
Taken together, the results presented here show clearly that DSP-PP 240 precursor secreted by viral infected Sf9 cells is cleaved after secretion, at the known physiological cleavage site, by an endogenous Zn-metalloproteinase activity secreted by Sf9 cells to yield DSP 430 and PP 240 proteins. We have shown that BMP1 is also capable of correct cleavage of DSP-PP 240 produced in the baculovirus system. We have shown further that Sf9 cells express an mRNA encoding the BMP1 homolog, TLR1. A reasonable interpretation of these results is that the specific cleavage of DSP-    PP 240 in the Sf9 cells expression system represents the action of a highly conserved processing enzyme, TLR1, on the mammalian substrate.

Discussion
Expression of the DSP-PP precursor protein is essential for normal tooth development due to requirement for PP in dental matrix mineralization. As a result, elucidating the mechanism and regulation of DSP-PP maturation is fundamental to understanding matrix formation. As shown here, baculovirus expression of DSP-PP 240 can provide a valuable model with which to address key questions about the maturation of DSP-PP. The baculovirus expression system permits the purification of substantial quantities of DSP-PP 240 precursor for biochemical analysis of processing. Like the precursor produced by odontoblasts, DSP-PP 240 produced in Sf9 cells is extensively glycosylated and phosphorylated (Ritchie HH, unpublished results).
Previously [15], we found that recombinant DSP-PP 240 , isolated using TCA precipitation, neutralization, calcium precipitation, and EDTA resuspension followed by preparative native PAGE, underwent spontaneous cleavage into DSP 430 and PP 240 when incubated at 37uC. We considered it unlikely that a contaminating protease in gel-purified DSP-PP 240 was responsible for cleavage because of the harsh, denaturing conditions used in the purification. Indeed, the low pH and chelation steps should have inactivated Zn-dependent peptidases. Regardless of these results, we have shown here that DSP-PP 240 is secreted into the conditioned medium intact, where it undergoes accurate and specific processing at the physiological cleavage site by a Znmetalloproteinase secreted by Sf9 cells to generate stable fragments that correspond to the physiological cleavage products DSP 430 and PP 240 .
In this study, we compared the fate of DSP-PP 240 secreted into conditioned medium in the first 3 days after viral infection (CM 0-3d ) to that of DSP-PP 240 secreted into freshly added medium during days 3-7 after infection (CM 3-7d ). Surprisingly, while significant amounts of mature DSP 430 and PP 240 were generated  by incubation of the CM 0-3d sample, little, if any cleavage of DSP-PP 240 occurred in the CM 3-7d sample on extended incubation, suggesting that a processing enzyme secreted by Sf9 cells within the first 3 days after infection, was no longer secreted after the first three days. Using CM 3-7d media, which contained stable DSP-PP 240 , we could test whether a processing enzyme was secreted by uninfected Sf9 cells. We were able to demonstrate that CM 0-3d media from uninfected Sf9 cells contained a Zn-dependent proteolytic activity that cleaved the DSP-PP 240 into DSP 430 and PP 240 (Fig. 4 and Fig. 6).
Sf9 cells are routinely used for expression of recombinant proteins encoded by baculovirus vectors. Our identification of a partial tlr1 cDNA from Sf9 cells is a new finding that may prove useful to others interested in examining the cleavage of Sf9produced recombinant proteins. Although we do not have definitive evidence that the DSP-PP 240 processing activity is Sf9 TLR1, the fact that the activity is secreted and Zn-dependent and that it cleaves DSP-PP 240 selectively at the physiological cleavage site makes it likely that this is the case. Moreover, the decline in expression of tlr1 mRNA after infection (Fig. 7B), which may be due to baculovirus inhibition of the endogenous gene expression, paralleled the decline seen in DSP-PP 240 cleavage activity in the conditioned medium (Fig. 3B).
As we have shown here, the decrease in endogenous processing activity after infection can be used to produce unprocessed DSP-PP 240 that can be used as substrate in studies using exogenously added protease. Using this approach, we have shown that BMP1 can accurately process baculovirus-expressed DSP-PP 240 to produce stable PP 240 and DSP 430 . These data are consistent with BMP1, or one of the related enzymes TLL1 or TLL2, being the physiological processing enzyme for DSP-PP. BMP1 belongs to the astacin family of metalloproteinases [24,25] and is present in many tissues [26]. BMP1 was reported to be involved in the cleavage of several extracellular matrix proteins such as the pro-a1 precursors of type I, type II, type III, and type VII collagen, the pro-a2 precursors of type I and type V collagen [16][17][18][19][20], and the human prolysyl oxidase [21]. In bone and teeth, as well as in nonmineralized tissues, BMP1 is required for cleavage of such extracellular matrix proteins as procollagen, to provide mature proteins that can then go on to effect both hard and soft tissue maturation. DSP-PP is also found in non-mineralized tissues (i.e., kidney, hair follicle, salivary gland and lung) that also contain BMP1, suggesting a role for BMP1 cleavage in precursor maturation in these tissues as well.
Taken together, our results suggest that the specific and physiologically accurate cleavage of DSP-PP 240 in the Sf9 cell system represents the action of a highly conserved processing enzyme.
Based on our results with the Sf9 cell system, we favor a model in which DSP-PP is cleaved after secretion into dentin matrix by BMP1 (or TLL1 or TLL2). This is suggested by the observation that although DSP-PP 240 and the Zn-dependent processing activity are co-expressed and co-secreted by Sf9 cells, the majority of DSP-PP 240 processing cannot be occurring until after secretion. Key questions raised by this model that can be addressed in the baculovirus system are why processing is delayed until secretion and how cleavage is prevented until after secretion if DSP-PP and BMP1 are co-secreted. Delay of processing may be necessary if the DSP portion of the precursor is required to prevent premature oligomerization of PP, which has been found to form large paracrystalline aggregates when expressed without the DSP portion [27]. Intracellular processing may be prevented if DSP-PP 240 undergoes a conformational change upon secretion, possibly due to high extracellular calcium that promotes efficient cleavage by BMP1 or a related enzyme. Finally, it is of interest to know whether interaction with extracellular calcium protects mature DSP and PP from further proteolysis.