Dentin sialoprotein facilitates dental mesenchymal cell differentiation and dentin formation

Dentin sialoprotein (DSP) is a dentin extracellular matrix protein. It is involved in dental mesenchymal cell lineages and dentin formation through regulation of its target gene expression. DSP mutations cause dentin genetic diseases. However, mechanisms of DSP in controlling dental mesenchymal cell differentiation are unknown. Using DSP as bait, we screened a protein library from mouse odontoblastic cells and found that DSP is a ligand and binds to cell surface receptor, occludin. Further study identified that the C-terminal DSP domainaa 363–458 interacts with the occludin extracellular loop 2aa 194–241. The C-terminal DSP domain induced phosphorylation of occludin Ser490 and focal adhesion kinase (FAK) Ser722 and Tyr576. Coexpression of DSP, occludin and FAK was detected in dental mesenchymal cells during tooth development. Occludin physically interacts with FAK, and occludin and FAK phosphorylation can be blocked by DSP and occludin antibodies. This DSP domain facilitates dental mesenchymal cell differentiation and mineralization. Furthermore, transplantation and pulp-capping procedures revealed that this DSP domain induces endogenous dental pulp mesenchymal cell proliferation, differentiation and migration, while stimulating blood vessel proliferation. This study elucidates the mechanism of DSP in dental mesenchymal lineages and implies that DSP may serve as a therapeutic agent for dentin-pulp complex regeneration in dental caries.

DSPP in odontoblasts and dentin is approximately 400 fold higher than that of osteoblasts and bone 7 . Although DSPP is transcribed from a single gene 8,9 , full length of DSPP protein has scarcely been isolated from cells or tissues 10,11 , whereas its cleavage products, dentin sialoprotein (DSP) and dentin phosphoprotein (DPP), are most abundant NCPs in odontoblasts and dentin 12 . DSP is further processed into small molecular fragments 11,[13][14][15] . Cleaved DSP fragments segregate into specific compartments within odontoblasts and dentin 14,16 .
DSP and DPP play unique biological functions during tooth development 17,18 . Mutations of either DSP or DPP domain in humans caused dentinogenesis imperfecta (DGI) type II (DGI-II, OMIM #125490) and type III (DGI-III, OMIM 125500) and dentin dysplasia (DD) type II (DD-II, OMIM 125420) [19][20][21] , the most common dentin genetic diseases. Mouse DSPP knock-out exhibited similar phenotype to that of DSPP gene mutations in human 22 . DPP contains an RGD domain, acting as a ligand, and binds to integrin as well as triggers intracellular signals via DPP-RGD/integrin-αvβ3 interactions 23,24 . By contrast, DSP lacks a RGD domain 9 , and many DSPP gene mutations occur in DSP region 19,20,25 . DSP and peptides derived from DSP are able to regulate gene expression and protein phosphorylation as well as induce dental primary/stem cell differentiation 9,16,26 . Recently, we have identified that 36 amino acids of DSP domain aa 183-219 bind to integrin β6 and the DSP-integrin β6 complex stimulated phosphorylation of Smad1/5/8 proteins through p38 and Erk 1/2 protein kinases. The phosphorylated Smad1/5/8 proteins were translocalized into nuclei and bind to DSPP gene promoter, activating expression of DSPP and DMP1 genes and inducing dental mesenchymal cell differentiation and biomineralization 9 . However, the molecular mechanisms of DSP controlling gene expression and cell differentiation have not been completely understood.
Occludin (Ocln) is an integral membrane protein associated with the tight junctions (TJs) of cells and mainly comprises four transmembrane domains, NH 2 -and COOH-terminal cytoplasmic regions and two extracellular loops 27,28 . The COOH-terminal domain is rich in serine, threonine and tyrosine residues, which are frequently phosphorylated by various protein kinases 29 . The extracellular loops of Ocln interact with a variety of cellular signaling molecules and are dynamically involved in intracellular signal transductions including protein phosphorylation/dephosphorylation and ion flux 28,30,31 . The cytoplasmic tail of Ocln is necessary for binding to its partners 32 . Ocln mutations in humans are involved in the pathogenesis of malformations of cortical development with band-like brain calcification and chronic kidney dysfunction 15,[33][34][35] . Ocln deficient mice developed deafness with dislocalization of tricellulin in cochlea 36 and displayed other signs of pathological disorders such as growth retardation, dysfunction of the salivary gland, calcification in the brain and thinning of the compact bone 37 . Although Ocln is one of the TJ proteins, unexpectedly, Ocln knock-out mice exhibited the normal TJ strand formation, and visceral endoderm cells originating from Ocln-deficient embryonic stem cells have well-developed networks of tight-junction strands 37,38 . It suggested that Ocln is also important for other biological functions. However, whether the signaling pathways by which DSP regulates intracellular activity via Ocln is not clearly understood.
Focal adhesion kinase (FAK or Ptk2) is the non-receptor bound tyrosine kinase and is involved in mediating both integrin and Ocln signaling for regulating intracellular transduction with the ECM on the outside of the cells 39,40 . Phosphorylation events occurring within focal adhesions influence numerous processes including cell adhesion, shape, motility as well as cell differentiation and tissue development 41,42 . FAK knockout studies showed an early embryonic lethal phenotype with extensive mesodermal deficiency and Ptk2 −/− embryonic fibroblasts from these mice exhibited profound defects in migration 43 .
Our hypothesis is that DSP regulates intracellular signal transductions and dental mesenchymal cell differentiations via matrix-cell surface interaction. Here, we found that DSP as a ligand is capable of binding to its cell surface receptor, Ocln. Further researches revealed that the COOH-terminal DSP domain interacts with the extracellular loop 2 of Ocln. This peptide phosphorylates Ocln and FAK. Immunohistochemistry showed that the spatial-temporal distribution of DSP, Ocln and FAK is co-localized in odontoblasts during dentinogenesis. The DSP domain was capable of inducing differentiation and mineralization of dental pulp stem cells. Furthermore, this peptide induced endogenous mouse dental pulp cell proliferation and differentiation and enhanced biomineralization when this DSP peptide was implanted into mouse dental pulp chambers.

Results
DSP as a ligand binds to its cell surface receptor, Ocln. To identify whether DSP interacts with other proteins, we generated GST-DSP fusion protein (Fig. 1A,B), and cell lysis was isolated from mouse odontoblast-like cells. The DSP fusion protein was used as bait to screen the protein library. This result showed that four proteins among 110 candidates interact with DSP, including integrin β6, CD105 (endoglin), collagen IV and Ocln (Fig. 1C (Fig. 1E-H). To further verify specific and saturable interactions and to evaluate the binding affinity between DSPf5 and OclnL2, we used a concentration range of DSPf5 and OclnL2 as analyzed by biotin-labeled proteins. These results manifested that the binding of DSPf5 to OclnL2 was dose-and time-dependent manners (Fig. 1I). For in vivo study, mammalian expression of DSP and Ocln genes was co-transfected into mammalian human embryonic kidney (HEK) 293 cells and coimmunoprecipitation assay showed that DSP binds to Ocln, and binding of DSP to Ocln was narrow to DSPf5 and OclnL2 (Fig. 1J), whereas the DSP-Ocln interaction could be blocked by the DSP and Ocln antibodies, but not IgG as control ( Supplementary Fig. 1). Using the gene runner computer software analysis of amino acids of the OclnL2 and DSPf5 reveals that these regions across many species are highly homologous (Supplementary Fig. 2A,B). This result indicates that DSP acts as a ligand and binds to its cell surface receptor, Ocln.
Scientific RepoRts | 7: 300 | DOI:10.1038/s41598-017-00339-w Expression of DSP and Ocln during mouse tooth formation. To determine whether the two proteins are coexpressed in tooth tissues, their distribution profiles during developing mouse teeth were examined using double labeling immunohistochemistry. At embryonic day (E) 13.5, DSP and Ocln expressions were not detected in craniofacial tissues (Fig. 2b,c). At postnatal day 1 (PN1), DSP was expressed in odontoblasts and weakly in ameloblasts (Fig. 2g). At PN5, DSP expression was strongly observed in odontoblasts and dentin and moderately in ameloblasts (Fig. 2l). Similar to PN5, DSP distribution exhibited the same expression pattern at PN15 in mouse  72-191, 182-295, 263-371, 363-458 were expressed in Escherichia coli BL21 and purified according to the manufacturer's instruction as described by "Materials and methods". The purified DSP fusion proteins were confirmed by Coomassie blue staining (A,D) and Western blot assays using either an anti-GST or anti-DSP antibody (B). Interaction between DSP polypeptides and Ocln by GST pull down was detected by Western blot using anti-DSP and anti-Ocln antibodies (C,E,F). Arrow shows Ocln band (C). Mixture of different fragments of DSP and Ocln fusion proteins was pulled down by Ocln antibody and interaction of DSP with Ocln was detected using DSP antibody (E) and vice versa (F). The GST-DSP fusion protein constructs and the localization of the DSP protein stretch that contains Ocln-binding domain are illustrated (G). The DSP number starts at the translational start site of DSPP (Met) as No. 1. (H) Schematic representation of mouse Ocln protein structure from the initiation of the translation start site 1 to the end 522. OclnL1 and OclnL2 indicate the extracellular loop 1 (aa 85-183) and loop 2 (aa 194-241). (I) Either recombinant OclnL2 protein or BSA as control was coated in 96-well plates. Serial diluted biotinylated DSPf5 was added to the wells and incubated with the unlabeled recombinant OclnL2 protein or BSA, respectively. Bound DSPf5 was detected using AP-conjugated streptavidin and 1 mg/ml PNPP as substrate at 405 nm using a microplate reader. Binding affinity of different concentrations of DSPf5 to its substrate at the given time periods were calculated. Using the same method, binding of the biotinylated OclnL2 to the unlabeled DSPf5 was examined. Data point represent the mean ± S.D. (n = 3). (J) For in vivo studies, different fragments of DSP and Ocln cDNAs were subcloned into a CMV mammalian expression plasmid tagged with Flag or Myc peptides, respectively. Myc-Ocln and Flag-DSP as well as Myc-DSP and Flag-Ocln expression vectors were transfected into HEK-293 cells, respectively. After 48 h transfection, the proteins were harvested and protein-protein interactions were immunoprecipitated using Myc antibody. Protein-protein interaction was detected by Western blotting using anti-Myc, anti-DSP or anti-Ocln antibody. teeth (Fig. 2q). Ocln signal coincided with DSP expression in odontoblasts (Fig. 2h,m,r), but Ocln expression pattern was also seen in stratum intermedium (SI). Although Ocln is an integral membrane protein, the NH 2 -and COOH domains of Ocln are localized in the cytoplasm 22 . Therefore, Ocln expression was observed in the cytoplasm besides the cell membrane. No signal was detected in the negative control tissue sections, where normal IgG was used instead of the primary antibodies of DSP and Ocln ( Supplementary Fig. 3b,c).

DSP induces phosphorylation of Ocln and FAK and two protein expressions in odontoblasts.
To elucidate whether DSP regulates modification of Ocln and its partner activity, mouse dental papilla mesenchymal cells were treated with DSPf5 protein. We found that DSPf5 phosphorylates Ocln at Ser 490 at a dose-dependent manner (Fig. 3A,B). As several genes are Ocln partners 29 , we further investigated whether DSPf5 also regulates Ocln partner activity. Interestingly, we found that DSPf5 also induces FAK phosphorylation at Ser 722 and Tyr 576 , but no effect on FAK at Tyr 397 (data not shown) and Akt at Ser 473 (Fig. 3A,B). Maximal induction of protein phosphorylation of Ocln-Ser 490 and FAK-Ser 722 , FAK-Tyr 675 by DSPf5 was at 2 h, 1 h and 1 h, respectively (Fig. 3C,D).
To determine relationship between Ocln and FAK within odontoblasts during odontogenesis, expression of Ocln and FAK in odontoblasts was performed using double labeling fluorescent immunohistochemistry. Our study showed that expression of Ocln and FAK was observed in odontoblasts at PN1, 5, and 10 ( Fig. 4A). No signals were detected in the control tissue sections when the antibodies of Ocln and FAK were replaced by normal IgG as negative control ( Supplementary Fig. 4b,c). Also, expression of Ocln and FAK proteins was seen in mouse dental papilla mesenchymal cells (Fig. 4B) and HEK 293 cells ( Supplementary Fig. 4g,h), but no immunostaining was seen in the control groups ( Supplementary Fig. 4l,m,q,r). To further elucidate if Ocln interacts with FAK, mammalian expression vectors of Ocln and FAK genes were transfected into mammalian cells. Coimmunoprecipitation assay demonstrated that Ocln physically interacts with FAK in vivo (Fig. 4C).

Effect of DSPf5 on phosphorylation of Ocln and FAK is disrupted by DSP and Ocln antibodies.
To further assess phosphorylation of Ocln and FAK mediated by DSP-Ocln signaling, we sought to block DSP signal pathway using DSP and Ocln antibodies. Mouse dental papilla mesenchymal cells were treated by DSPf5 with or without the DSP or Ocln antibody and the cells were harvested for immunoblot analysis. It was noted that the DSP and Ocln antibodies could block phosphorylation of Ocln at Ser 490 and FAK at Ser 722 and Tyr 576 compared to the control group ( Fig. 5A-D), but IgG as control had no effect on block of the two protein phosphorylations induced by DSPf5. Also, disruption of the two protein phosphorylation by the DSP and Ocln antibodies was dose-dependent. With increase of the DSP and Ocln antibodies, phosphorylation of p-Ocln at Ser 490 , p-FAK at Ser 722 and p-FAK at Tyr 576 was dramatically decreased. Immunohistochemistry further indicated that phosphorylated levels of the two proteins were attenuated by the DSP and Ocln antibodies ( DSP induces dental cell differentiation and mineralization. As DSP is abundant in dentin ECM during odontoblast differentiation, we further addressed the possibility that Ocln-FAK signaling could be involved in controlling functional activities of dental cell differentiation and mineralization. To directly assess DSP effect on dental cell differentiation, we firstly examined differentiation of human dental pulp stem cells and mouse dental papilla mesenchymal cells in cultured systems. ALP assay revealed that DSPf5 has high effect on human dental pulp  Alizarin red S assay also demonstrated that DSPf5 stimulates more calcium deposition of the two dental cells after 14 days of culture than that of the control groups ( Fig. 7E-H). The cell differentiation and mineralization induced by DSPf5 protein was attenuated by the DSP and Ocln antibodies, whereas IgG as control had no effect on the cell differentiation and mineralization stimulated by DSPf5 ( Supplementary Fig. 6). Furthermore, gainand loss-Ocln gene in the mouse dental papilla mesenchymal cells up-and down-regulated the cell differentiation and mineralization ( Supplementary Fig. 7). The data indicate that DSP promotes the dental cell differentiation and mineralization partially through the Ocln signaling.
To further assess whether this DSP peptide modulate endogenous dental mesenchymal cell differentiation and biomineralization in an in vivo context, we mixed the peptide with Affi-Gel Blue Gel. Data showed that DSPf5 highly binds to the Affi-Gel Blue Beads, forming a DSP-bead compound and this DSPf5 peptide is also easily released from the compound (Fig. 8A). Then, the compound was implanted into mouse dental pulp chambers. Tissue morphology after 1-, 3-, and 5-week surgery was observe by histological chemistry, and H&E stained tooth tissue sections revealed that there were not significant differences of tooth morphology between the control and DSPf5 treated groups after 1 week surgery (Supplementary Fig. 8a-f). However, after 3 week treatment of DSPf5, DSPf5 was able to induce the dental pulp mesenchymal cell proliferation and differentiation. Dental pulp mesenchymal cells in the DSP-treated group secrete dental ECM at the top "artificial hole" between the resin and dental pulp chamber ( Fig. 8Ba-g). More interestingly, in the DSP-treated group, there were many blood vessels and less inflammatory cells around the agarose beads and dental pulp cells. Also, the agarose beads were resorbed as well as dental pulp cells and blood vessels were invaded into the agarose beads. Similar to the 3-week DSP treatment group, in the 5-week DSP induced groups, ECM covered major space at the wound area and less inflammatory cells were seen in the dental pulp cavity. Many blood vessels were apparent around the agarose beads ( Fig. 8Ca-h). However, formation of reparative dentin was not seen in the DSP treated groups, the mechanism of the DSP domain in the reparative dentin formation needs to be further investigated in the future. These findings indicate that the DSP domain induces differentiation of dental pulp mesenchymal cells into odontoblast-like cells and differentiating cells are capable of synthesizing and secreting ECM.

Figure 2.
Expression of dentin sialoprotein and occludin in developing mouse teeth. At embryonic day (E) 13.5, DSP and Ocln expressions were not detected in tooth tissues (b,c) using double fluorescent immunohistochemistry, but at postnatal day (PN) 1, DSP expression (red) was observed in odontoblasts, ameloblasts and dental pulp cells (g), Ocln expression (green) was detected in these areas overlapped with DSP (h). At PN5 and PN15, expression of DSP and Ocln was apparently seen in odontoblasts and ameloblasts (l,m,q,r). However, Ocln expression was relatively wider than that of DSP. (a,f,k and p) show bright images. The cells were stained with Hoechst for nuclei (d,i,n,s). Images were merged (e,j,o,t). (a-t) show higher magnifications from the yellow boxes.

Discussion
DSP is highly expressed in dentin ECM during odontoblast differentiation and dentin formation. Its mutations are associated with dentin hereditary diseases [19][20][21]44 . Previous studies showed that DSP and its fragments derived from DSP regulate intracellular signal transductions 9,16,26 . Unlike other SILING members, DSP domain does not contain RGD 8,45 . Therefore, mechanisms of DSP in regulation of intracellular signaling during dentinogenesis have not well been understood. Recently, our and other laboratories found that DSP is processed by MMP20 and MMP9 into the given fragments 11,15 and the middle domain of DSP aa 183-219 binds to integrin β6, forming a complex. This complex phosphorylated transcription factors, Smad1/5/8, through p38 and Erk1/2 protein kinases and up-regulated expression of DSPP and DMP1 genes as well as induced dental mesenchymal cell attachment, differentiation and mineralization 9 . In this study, we used DSP as bait to screen protein library from mouse odontoblast-like cells and identified that several cell membrane proteins were bound by DSP including Ocln (Table 1). Previous studies showed that Ocln interacts with several cellular molecules and activates intracellular signal transductions 22  immunohistochemistry showed that at PN1, 5, 10 during mouse tooth development, Ocln (red) and FAK (green) were co-expressed in odontoblasts by the double immunofluorescent histochemistry using antibodies specific to Ocln and FAK (b,c,g,h,l,m). (a, f and k) show bright images. The cells were stained with Hoechst for nuclei (d,i,n). Images were merged (e,j,o). Am, Ameloblasts; Od, Odontoblasts. (B) Ocln (red) and FAK (green) were expressed in iMDP-3 cells using double immunofluorescent assay (b,c). a shows bright image. Hoechst was used for nuclear staining (d). Image was merged (e). (C) Ocln interacts with FAK in vivo. Myc-Ocln and GFP-FAK mammalian expression vectors were cotransfected into HEK-293 cells. After 48 h transfection, the cells were harvested. Ocln and FAK were immune-precipitated using Myc or GFP antibody. Interaction of Ocln with FAK was detected by Western blot assay using Ocln and FAK antibodies.
Scientific RepoRts | 7: 300 | DOI:10.1038/s41598-017-00339-w through HCV glycoproteins (HCVpp) and OCLN is an essential entry factor. They furthermore found that the specific determinants of OCLN's HCV entry factor functions are entirely contained within the OCLN extracellular loop 2 31 . As DSP is able to bind to Ocln in vitro and in vivo, we further studied which domain of DSP interacts with Ocln and found that the COOH-DSP domain aa 363-458 is a ligand and binds to the extracellular loop 2 (OclnL2) of Ocln. This DSP domain regulates intracellular signaling through a direct matrix-cell interaction. Immunohistochemistry showed that expression of DSP and Ocln is stage-specific in odontoblasts during dentinogenesis.
Ocln is an integral membrane protein expressed in epithelial and other cells and was originally predicted to confer barrier properties to the TJs [27][28][29] . During dentinogenesis, adjacent odontoblasts are tightly attached together and freeze-fracture studies showed that TJs between odontoblasts appear as short rows of fused particles during early mantle dentin mineralization and become more complex, eventually forming networks of fused particles in advanced mineralization [46][47][48][49] . The transport of calcium ions through odontoblasts is mediated by specific cell organelles and cell membrane domains 47,48,[50][51][52][53] . TJs between odontoblasts are mainly composed of transmembrane proteins including claudin and Ocln as well as cytoplasmic proteins [such as zona occludens-1 (ZO-1), ZO-2 and ZO-3, FAK] 48,49,54 . Expression of the TJ-associated proteins was clearly detected in odontoblasts, however, temporal and spatial distributions of these proteins such as Ocln in differentiating odontoblasts are different 48,49,[55][56][57][58] . The cause of the differences is still uncertain, and it may be due to the animals used and the methods for specimen preparation as well as antibody concentrations in each study 59 . Lee et al. observed that Ocln and ZO-1 proteins are strongly expressed in odontoblasts at PNs from 7 to 18 examined in the wild type mouse teeth whereas expression of Ocln and ZO-1 is hardly detected in odontoblasts of Nfic-deficient mice at the same ages. Nfic-deficient teeth failed to differentiate into normal odontoblasts and these aberrant odontoblasts were round and dissociated and lost their polarity. Electron microscopy of aberrant odontoblasts exhibited no intercellular junctional complex between them 58 . Lungrw et al. described that MRPC-11 cells derived from rat dental pulp cell line exhibit an odontoblast-like phenotype and express DSPP and Ocln. TJ complex was seen in the intercellular spaces of these cells and the transcellular Ca 2+ flux was inhibited by nifedipine, giving evidence for an active intracellular Ca 2+ transport through voltage-gated channels 60 . In this study, we observed that expression of DSPP/DSP and Ocln was not detectable in tooth tissues at embryonic days 13.5 (Fig. 2Ab,c). With  odontoblast cytodifferentiation, both of these proteins were expressed in odontoblasts at PNs from 1 to 15 examined (Fig. 2Ag,h,l,m,q,r). Our results are in agreement with previous studies by other groups 5, 55-61 .
Ocln mutations in humans cause band-like brain calcification, polymicrogyria and advanced chronic renal disease 15,[33][34][35] . Ocln deficient mice developed deafness with dislocalization of tricellulin in cochlea 36 . Tricellulin is a recently identified constituent of TJs, and is the first marker of the tricellular tight junction (tTJ) 62 . Also, Ocln knock-out mice showed calcification in the brain, testicular atrophy, loss of cytoplasmic granules in striated duct cells of the salivary gland, and thinning of the compact bone, but in Ocln knockout-mice, morphology of the TJs does not appeared to be affected and barrier function of epithelium is normal 37 and cells originating from Ocln-deficient embryonic stem cells have well-developed networks of tight-junction strands 38 , suggesting that the functions of Ocln are more complicated than previously supposed. These contradictory data led us to ask if Ocln contributes to a regulatory function in intracellular signal transductions. Our current study demonstrated that a novel role of Ocln acts as a receptor interacting with DSP for modulating intracellular functions via DSP-Ocln-FAK axis. It was found that several ECMs and molecules interacting with either extracellular loop 1 or 2 of Ocln regulate intracellular signaling through phosphorylation/dephosphorylation of Ocln [28][29][30][31]63 . In the present study, we found that DSP aa 363-458 binds to the extracellular loop 2 aa 194-241 of Ocln. Both of the DSP aa 363-458 and OclnL2 aa 194-241 are high homologous across species lines. In human, the mutation of DSP at codon 45 (c.133C > T, p.Q45X) introduces a premature terminal signal and would results in a truncated protein without the COOH-terminal DSP domain 21,25,64,65 . As the mouse COOH-terminal DSP domain aa 363-458 is highly homologous to the human COOH-terminal DSP domain aa 374-469 66 and this domain regulates dental mesenchymal cell lineages and mineralization through Ocln-FAK signal. It is assured that the lack of the domain of the human DSP in these cases (p.Q45X) may be relevant to DGI.
Evidence for Ocln phosphorylation has previously been described [28][29][30] . However, the mechanisms by which these modifications through DSP influence intracellular signaling during dentinogenesis have not been Figure 7. Effect of DSP domain on dental cell differentiation and mineralization. Mouse dental papilla mesenchymal (iMDP-3) (A) and human dental pulp stem (hDPSC) cells (B) were treated with or without 10 μg/ml of DSPf5 in calcifying medium for 7 days. ALP activity was analyzed using in situ ALP staining (A, B). (C,D) Quantitative ALP activity of the cell lysates was assayed using ρ-nitrophenyl phosphate as a substrate. Protein concentration was determined using the BCA protein assay reagent as described in "Materials and methods". There were significantly different between the DSPf5 treated groups and DSPf5 untreated groups. For cell biomineralization, iMDP-3 (E) and hDPSC cells (F) were maintained in the same condition for 14 days. The cells were fixed and stained for Alizarin red S dye. (G,H) The amount of calcium deposition was quantified by destaining with 10% cetylpyridinium chloride in 10 mM sodium phosphate at room temperature. Mineralization deposits were determined and data represent mean ± S.D. (n = 3). Con, control. *p < 0.05; **p < 0.01. elucidated. We demonstrated that the DSP peptide phosphorylates Ocln at Ser 490 in mouse dental papilla mesenchymal cells and the Ocln phosphorylation at Ser 490 induced by DSPf5 was blocked by DSP and Ocln antibodies, implying Ocln phosphorylation at Ser 490 through DSP signaling. Previously, it was found that vessel endothelial growth factor (VEGF) induces Ocln phosphorylation at Ser 490 through activating protein kinase C (PKC) β and regulation of vascular permeability 63 . We observed that besides induction of Ocln phosphorylation, DSP also activated FAK phosphorylation at Ser 722 and Tyr 576 in mouse dental papilla mesenchymal cells. Maximal induction of Ocln and FAK phosphorylation was at 1-and 2-h of DSPf5 treatment. FAK is ubiquitously expressed in a variety of cells. Previous study showed that FAK is structurally associated with Ocln, but not other tight junction proteins, JAM-A and claudin-11 in Sertoli cells. FAK knock-down resulted in a significant loss of Ocln-ZO-1 interaction and interruption of the TJs 67 . Our studies demonstrated that similar to coexpression of DSP and Ocln, expression of both Ocln and FAK is stage-specific in odontoblasts during tooth development and in mouse dental papilla mesenchymal cells as well as Ocln physically interacts with FAK (Fig. 4). Expression of Ocln and FAK in odontoblastic cells during tooth formation was also observed by other groups 24,58,60 . Thus, coexpression of the three of DSP, Ocln and FAK exists in odontoblasts during tooth development. FAK bridges the cytoskeleton on the inside of cells with components of the ECM on the outside of cells via the cell surface receptors such as integrin and others 68,69 . FAK protein contains many phosphorylated sites and phosphorylation events occurring within FAK influence numerous processes including mitogenic signaling, cell migration, proliferation and The complex of DSPf5 coated to agarose beads was implanted into mouse dental pulp chambers. Compared to the control group after 3 week operation (Ba-c), in the DSPf5-treated group, DSPf5 was able to induce the dental pulp cell proliferation and differentiation. Dental pulp cells in the DSPf5-treated group secrete ECM (arrow) at the top "artificial hole" between resin and dental pulp. Blood vessels (BV) proliferate and migrate near agarose beads (AB). Arrowheads show that dental pulp cells differentiate and secrete ECM. b, c and e-g are enlarged from the boxes in the b' , c' and e'-g' . In DSPf5-treated group after 5 week operation (C), ECM secreted by dental pulp cells forms a layer covering "the artificial hole" between resin and dental pup cells (Ce, f). Dental pulp cells surround the agarose beads (AB) and AB was resorbed and dental pulp cells and blood vessels invade into AB (Cg, h). A number of inflammatory cells were decreased compared to the control group. b-d and f-h are enlarged from the boxes in the b'-d' and f '-h". B, alveolar bone; BV, blood vessels; DP, dental pulp cells; ECM, extracellular matrix; IC, inflammatory cells; R, resin; T, teeth.
differentiation. Its activity is modulated by phosphorylation, either in a positive or negative fashion 70,71 . FAK phosphorylation could be induced by adhesion of cell surface receptors such as integrin to ECM and by a variety of other extracellular factors 72 . Our study demonstrated that FAK phosphorylation at Ser 722 and Tyr 576 was blocked by either DSP or Ocln antibody. These findings suggest that FAK is a substrate of Ocln. However, the mechanism how Ocln phosphorylates FAK at Ser 722 and Tyr 576 remains unknown and whether the DSP affects binding affinity and structure of Ocln and FAK needs to be further investigated. In fact, recent studies have found that there are multiple potential Tyr, Ser and Thr phosphorylation sites in FAK mapped by mass spectrometry 73 .
In this study, we found that DSP domain aa 363-458 interacts with the extracellular loop 2 of Ocln aa 194-241 . The DSP domain induced phosphorylation of Ocln at Ser 490 and FAK at Ser 722 and Tyr 576 . The phosphorylation of Ocln and FAK proteins could be blocked by the anti-DSP and anti-Ocln antibodies, respectively. Immunohistochemistry analysis showed that co-expression of DSP, Ocln and FAK proteins are present in mouse odontoblasts and dental mesenchymal cells. Furthermore, in vivo study revealed that Ocln binds to FAK, indicating that the DSP domain regulates the intracellular activities through the Ocln-FAK signaling pathway. For the functional study, we observed that the DSP peptide aa 363-458 is sufficient to induce differentiation and mineralization of human dental pulp stem cells and mouse dental papilla mesenchymal cells in the cultured system. Effect of DSP on dental papilla mesenchymal cell differentiation and mineralization could be blocked by the DSP and Ocln antibodies and gain-and loss-Ocln gene in these cells was able to up-and down-regulate the cell differentiation and mineralization. More importantly, when this peptide was implanted into mouse dental pulp chambers, this DSP peptide was able to induce endogenous dental pulp cell proliferation, differentiation. The differentiating dental pulp cells were capable of synthesizing and secreting dental ECM and migrating to the region between resin and dental pulp chamber as well as new synthesized dentin ECM covered the wound area. Also, in DSP-treated groups, more blood vessels were seen surrounding dental pulp cells and agarose beads. Less inflammatory cells were observed in dental pulp chambers compared to the control groups. Although the formation of dentin extracellular matrix between the wound area and dental pulp chamber in the DSP treated groups in 3-and 5-week surgery was seen, the formation of the reparative dentin in both the two groups was not observed. The phenomenon remains to be further studied. In the present study, these findings suggest that this DSP peptide could be a potentially interesting therapeutic molecule by locally regenerating dentin-pulp complex. However, the detail mechanisms by which DSP domain mediates dental cell proliferation, differentiation, migration and blood vessel regeneration through Ocln-FAK signaling transduction needs to be further investigated in the future. Figure 9 depicts a model by which DSP domain aa 363-458 regulates dental cell differentiation and mineralization via DSP-Ocln-FAK signaling based on this report.  Generation of recombinant dentin sialoprotein and occludin. The expression and purification of a recombinant DSP were performed according to previously described protocols 16 . Full-length and different fragments of mouse DSP cDNA were amplified by PCR using a full-length mouse DSPP cDNA as a temple with primers shown in Supplementary Table 2 adding EcoRI sites at both ends for directional ligation into the expression vector pGEX-6P3 with EcoRI sites (Amersham Biosciences, Piscataway, NJ, USA) and named DSP, DSP-N, DSP-C, DSPf1, DSPf2, DSPf3, DSPf4 and DSPf5 (Fig. 1G). Mouse Ocln plasmid (pBluescript SK-Ocln) was kindly provided by Dr. Mikio Furuse (Department of Physiology and Cell Biology, Kobe University Graduate School of Medicine, Kobe, Hyogo, Japan). Full-length and different fragments of Ocln cDNA including the whole length and extracellular loops 1 (OclnL1) and 2 (OclnL2) of Ocln gene were amplified by PCR using a full-length mouse Ocln cDNA as a template. Primers used for generation of Ocln DNA constructs were shown in Supplementary  Table 2. PCR products were subcloned into pGEX-6P3 vector, respectively (Amersham Biosciences, USA). After confirming the right sequence, the resulting plasmid was transformed into Escherichia coli BL21. The protein expression and purification were performed according to the manufacturer's instruction (Amersham Biosciences, USA). Briefly, the fusion protein was induced by the addition of 1 mM of isopropyl b-D-thiogalacto pyranoside (IPTG) at 37 °C for 5 h. The recombinant DSP and Ocln proteins were then purified using a GSTrap 4B protein purification system (Amersham Biosciences, USA, #28-4017-45). The purified recombinant proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by Coomassie brilliant blue staining and Western blotting analysis using GST antibody (1:2,000, Amersham Biosciences, USA, #27-45-77-01). For in vivo study, DSP and Ocln genes were subcloned into CMV expression vector tagged with either Myc or Flag (Sigma-Aldrich, USA, #pp2393, #pp2395). Mammalian mouse FAK with GFP expression plasmid was obtained from Addgene (Cambridge, MA, USA, #50515).

Antibodies and Reagents. A rabbit polyclonal anti-mouse DSP antibody recognizes residues between
Glutathione fusion protein (GST) pull down assay. GST pull down assay was according to the manufacture's instruction (Amersham Biosciences, USA). In brief, GST-DSP fusion protein was incubated with 1 ml of cell lysis from mouse odontoblast-like cells (MO-6G3) 6 in the cold lysis buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, Triton X100 (Sigma-Aldrich, USA) overnight at 4 °C. After the reaction, the 50 μl of a 50% slurry of glutathione agarose beads (Amersham Biosciences, USA) were added for further incubation for 2 h at 4 °C. The samples were centrifuged at 12,000 g for 2 min at 4 °C and the supernatant was removed. After extensive washes containing 10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.025% sodium azide, the DSP binding proteins were eluted by 20 mM reduced glutathione in 50 mM Tris-Cl, pH 8.0. The eluted samples (10 μg) were mixed with 2x SDS-PAGE gel loading buffer after boiling and run onto 7% SDS-PAGE gels and were transferred to Trans-Blot membranes (Bio-Rad Laboratories Inc., Hercules, CA, USA). The membrane was blocked with 5% non-fat milk in TBST buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% Tween-20) for 60 min at room temperature. After blocking, the membranes were incubated overnight at 4 °C with the respective primary antibodies (1:500-1,000 dilutions) shown in Table 1 and Supplementary Table 1. The membranes were washed with 1x TBST and incubated with diluted horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5,000, Pierce, Rockford, IL, USA) for 1 h at room temperature. After three washes, the membranes were detected using an enhanced chemiluminescence (ECL) kit (Millipore, Bedford, MA, USA).
Co-immunoprecipitation. Co-immunoprecipitation assay was performed according to the manufacture's instruction (Active Motif, Carlsbad, CA, USA, #54002). Briefly, HEK 293 cells from a 100-mm Petri dish (70-90% subconfluent) were transfected with 20 μg of Flag and c-Myc plasmids containing DSP and Ocln genes using the Lipofectamine 2000 (Life Technology, USA). After 48 h transfection, the cells were lysed in 3 mL of lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100 and proteinase inhibitor cocktail (Sigma-Aldrich, USA) and mixed with a vortex mixer for 1 h at 4 °C. Insoluble materials were removed by centrifugation at 12,000 g for 10 min at 4 °C. Lysates containing 5 mg of proteins in 1 mL were precleared by incubating with 40 μl of anti-FLAG or anti-Myc Affinity Gel (Sigma-Aldrich, USA, #A4596 and #IP0020) at 4 °C on a rotator overnight. Immunoprecipitates were washed three times with 1 x TBST and heated for 5 min at 95 °C in 20 µl 2x SDS-PAGE sample buffer (62.5 mM Tris HCL, pH 6.8. with 2%SDS, 10% glycerol, 0.002% bromophenol blue). The bound materials were resolved on 7% SDS-PAGE gels for analysis by Western blotting with anti-Ocln (1:1,000, Life Technologies, USA), or anti-Myc (1:1,500; Thermo Fischer Scientific, USA) or anti-DSP (1:1,000, Santa Cruz Biotechnology Inc., USA) antibody. In vitro immunoprecipitation of DSP and Ocln fusion proteins was performed according to the manufacture's protocol (Amersham Biosciences, USA). In brief, 10 μg of GST-DSP and -Ocln fusion proteins were pre-cleared with 10% protein G-strepharose in binding buffer. The samples were then incubated with 10 μg of either primary DSP (Santa Cruz Biotechnology Inc., USA) or Ocln (Life Technologies, USA) antibody, followed by immunoprecipitation with 20 μl protein G-strepharose (Amersham Biosciences, USA). Immunoprecipitations were washed extensively with lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100) and boiled for 5 min. The supernatants were separated by 7% SDS-PAGE gels, transferred to nitrocellulose membranes, immunoblotted with primary DSP (1:1,000, Santa Cruz Biotechnology Inc., USA) or Ocln (1:1,500, Life Technologies, USA) antibody at 4 °C overnight. The membranes were then washed with 1 × TBST for three times and incubated with diluted horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5,000, Pierce, USA) for 1 h at room temperature. After three washes, the membranes were detected using an enhanced chemiluminescence (ECL) kit (Millipore, USA).
Biotinylation of dentin sialoprotein and substrate binding assay. For probing protein-protein interactions in vitro, recombinant DSP aa 363-458 (DSPf5) and Ocln loop2 aa 194-241 (OclnL2) proteins at concentrations of 300 mg/ml was dialyzed against 0.1 M NaHCO 3 and then reacted with 100 mg/ml Sulfo-NHS (N-hydroxysuccinimido)-LC (long-chain)-biotin according to the manufacture's instruction (Pierce, USA) for 20 min at room temperature, followed by 2 h at 4 °C, respectively. Free biotin was removed by dialysis against 50 mM Tris-HCl and 150 mM NaCl, pH 7.4. To characterize the relative DSPf5 interaction with OclnL2, substrate binding assay was performed as described earlier 77 . Briefly, 96-microwell plates were coated with 1 μg/per well of OclnL2 and bovine serum albumin (BSA) as control overnight at 4 °C and non-specific binding sites were blocked with 1% BSA (Sigma-Aldrich, USA) at room temperature for 1 h. After through rinses with 1x PBS, the biotinylated DSPf5 ranged from 0-18 fold molar excesses in PBS with 1% BSA was added into the plates and then incubated at room temperature for different time periods. Bound DSPf5 was reacted with AP-conjugated streptavidin diluted 1:10,000 in 1x PBS for different time points at room temperature using 1 mg/ml PNPP (ƿ-nitrophenyl phosphate disodium) as substrate (Pierce, USA) and quantified at 405 nm (Opsys MR, Dynex, Chantilly, VA, USA). BSA was used as a control group and the binding of DSPf5 to OclnL2 was expressed as relative binding activity compared to the control group. At the same way, binding of the biotinylated OclnL2 to the unlabeled DSPf5 was performed. All experiments were performed from three independent experiments in triplicate.

Protein sequences and data analyses. A database search was performed at the National Center for
Biotechnology Information website (http://www.ncbi.nlm.nih.gov/blast) using the BLAST program. The DSPP and Ocln nucleotides and derived amino acids were aligned with those from different species using the Gene Runner software program (http://www.generunner). Accession numbers of mRNA and protein sequences of DSPP and Ocln were shown in Supplementary Fig. 2.
Alkaline phosphatase (ALP) and mineralization assays. For detection of ALP activity, cultures of the human dental pulp stem and mouse dental papilla mesenchymal cells at 6-well palates with 10 4 cells per well were treated with or without DSPf5 (10 μg) induction in calcifying medium (α-MEM supplemented with 3% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml), 50 μg/ml ascorbic acid, 10 nM dexamethasone and 10 mM sodium β-glycerophosphate) at 37 °C for 7 days and fixed with 70% ethanol for 5 min and washed with 1 x PBS. In situ ALP staining was performed according to the manufacture's instruction (Bio-Rad Laboratories Inc., USA). Quantitative ALP activity of the cell lysates was assayed using ρ-nitrophenyl phosphate as a substrate. Protein concentration was determined using the BCA protein assay reagent (Pierce, USA). The enzyme activity was expressed as nanomoles of ρ-nitrophenol produced per min per mg of protein. For mineralization assay, calcium deposition was determined according to the manufacture's recommendation protocol (Sigma-Aldrich, USA). In brief, these cells were seeded into 6-well plates and treated with or without 10 μg of DSPf5 in calcifying medium at 37 °C for 14 days. The cells were fixed in 10% formaldehyde neutral buffer at room temperature for 15 min. The monolayers were then washed twice with excess ddH 2 O prior to addition of 1 mL of 40 mM alizarin red S dye (pH 4.1) per well (Sigma-Aldrich, USA). The plates were incubated at room temperature for 20 min with gentle shaking. After aspiration of the unincorporated dye, the wells were washed four times with ddH 2 O while shaking for 5 min. Stained monolayers were visualized by phase microscopy using a Nikon Eclipse TE2000S inverted microscope with a filter by means of a digital cooled camera connected to a PC computer and analyzed with NIS-Elements 3.2 software (Melville, NY, USA). The amount of calcium deposition was quantified by de-staining with 1 ml of 10% cetylpyridinium chloride (Sigma-Aldrich, USA) in 10 mM sodium phosphate at room temperature for 1 h. The dye was then removed and 200 μL aliquots were transferred to a 96-well plate prior to reading at 550 nm using a microplate reader (Opsys MR, USA). Experiments were performed in triplicate and repeated in three cultures (n = 3).
Immunohistochemistry. In order to examine the expression of DSP, Ocln and FAK during tooth development and dental papilla mesenchymal cells, the fluorescent immunohistochemistry was performed according to previous descriptions 74 . In brief, tissue sections were dewaxed in xylene for 15 min for 3 times, hydrated in gradually decreasing ethanol concentrations for 2 min each and immersed in ddH 2 O for 5 min for three times. Antigen retrieval was performed by treating the sections with 0.1% (w/v) trypsin (Sigma-Aldrich, USA) at room temperature for 15 min. The sections were successively pretreated with 0.3% hydrogen peroxide for 30 min, blocked with 10% normal donkey serum (Sigma-Aldrich, USA). For double labeling, two primary antibodies were incubated simultaneously overnight at 4 °C at the following dilutions: rabbit anti-FAK (1:100; Santa Cruz Biotechnology Inc., USA, #sc-557), goat anti-DSP (1:100; Santa Cruz Biotechnology Inc., USA, #sc-18328), mouse anti-Ocln (1:100; Life Technologies, USA, #33-150), which were recognized by donkey anti-rabbit, anti-mouse and anti-goat IgG (H + L) antibodies conjugated with Alexa Fluo 488 and Alex Fluo 568 (Molecular Probes, Eugene, Ore., USA, #SA5-10166, #SA5-10168, #A110556, #A21206). After being washed with PBS, the sections were incubated with the secondary antibody (1:500) for 90 min at room temperature and then rinsed in PBS. For nuclear staining, the tissue sections were incubated with 1:5,000 dilution of Hoechst (Sigma-Aldrich, USA, #23491-45-4) for 5 min at room temperature. After being washed, the tissue sections were mounted in Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). As a negative control, the primary antibody was replaced by 10% of mouse IgG (Vector Laboratories, USA, #I-2000). For cell double labeling, cells were fixed in cold acetone and methanol (1:1), permeabilized in 0.2% Triton X-100 at ice for 30 min and blocked with 10% donkey serum for 30 min at room temperature and two primary antibodies recognized by the donkey secondary antibody were incubated simultaneously overnight at 4 °C at the following dilutions: rabbit anti-FAK (1:100; Santa Cruz Biotechnology Inc., USA, #sc-557), goat anti-DSP (1:100; Santa Cruz Biotechnology Inc., USA, #sc-18328), mouse anti-Ocln (1:100; Life Technologies, USA, #33-150). After being washed, the cells were incubated with the secondary antibody conjugated with Alexa Fluo 486 green and Alexa Fluo 568 red (1:500; Molecular Probes, USA, #A21206, #SA5-10168) for 1 h at room temperature. For nuclear staining, the cells were incubated with 1:5,000 dilution of Hoechst (Sigma-Aldrich, USA, #23491-45-4) for 3 min at room temperature. After being washed, the cells were mounted in Vectashield mounting medium (Vector Laboratories, USA). As a negative control, the primary antibody was replaced by 10% mouse IgG (Vector Laboratories, USA, #I-2000). Images of Alexa Fluo 488 and Alex Fluo 568 staining of the various proteins were captured by a Nikon Eclipse TE2000S microscope with a filter by means of a digital cooled camera connected to a PC computer and analyzed with NIS-Elements 3.2 software. Images of nuclear staining with Hoechst were obtained via filter UV-2E/C, C86826. For each experiment, all slides were simultaneously processed for a specific antibody, so that homogeneity in the staining procedure was ensured between samples.
Reparative dentin regeneration. Affi-Gel Blue (cross-linked agarose beads with covalently coupled Cibacron Blue F3-GA) was purchased from Bio-Rad laboratories Inc. and mixed with the DSP peptide (2 mg/ml). The mixture was centrifuged and pellet used for implantation. The experimental procedures were approved by the animal research committee of UTHSCSA. Twenty-five adult mice (C57BL/6) aged 1.5 months were anaesthetized by sodium pentobarbital (Sigma-Aldrich, USA). The first molars on the maxilla were cleaned by sterilized instruments. Exposed pulped cavities were prepared by ¼ diamond cylindrical burs with sterile saline cooling. The exposed pulped cavity on the left first molars was filled onto Affi-Gel blue gel as control and Affi-Gel blue gel with the DSP peptide on the right first molars. The cavities over the implanted materials were filled with Vitremer Glass Ionomer (GI) Core Build-Up/Restorative (Sku# 3303PEDO, 3 M ESPE, Dental Products, St. Paul, NM, USA) and sealed by light curing. Sealant well done in each animal was observed carefully. The animals were euthanatized at 1, 3, and 5 weeks after surgery. Tissue specimen were fixed in 4% paraformaldehyde at 4 °C overnight and embedded in paraffin wax after demineralization with 10% EDTA. The paraffin sections (5 μm in thickness) were morphologically examined after staining with hematoxylin and eosin (HE) (Agilent Technologies, USA). Statistical analysis. Quantitative data were presented as means S.D. from three independent experiments and compared with the results of one-way ANOVA using GraphPad Prism 5 (GraphPad Software, Inc. La Jolla, CA, USA). The differences between groups were statistically significant at *p < 0.05 and **p < 0.01.