Syndecan-1 Expression in Epithelial Cells Is Induced by Transforming Growth Factor β through a PKA-dependent Pathway*

Syndecans comprise a major family of cell surface heparan sulfate proteoglycans (HSPGs). Syndecans bind and modulate a wide variety of biological molecules through their heparan sulfate (HS) moiety. Although all syndecans contain the ligand binding HS chains, they likely perform specific functions in vivo because their temporal and spatial expression patterns are different. However, how syndecan expression is regulated has yet to be clearly defined. In this study, we examined how syndecan-1 expression is regulated in epithelial cells. Our results showed that among several bioactive agents tested, only forskolin and three isoforms of TGFβ (TGFβ1-TGFβ3) significantly induced syndecan-1, but not syndecan-4, expression on various epithelial cells. Steady-state syndecan-1 mRNA was not increased by TGFβ treatment and cycloheximide did not inhibit syndecan-1 induction by TGFβ, indicating that TGFβ induces syndecan-1 in a post-translational manner. However, TGFβ induction of syndecan-1 was inhibited by transient expression of a dominant-negative construct of protein kinase A (PKA) and by specific inhibitors of PKA. Further (i) syndecan-1 cytoplasmic domains were Ser-phosphorylated when cells were treated with TGFβ and this was inhibited by a PKA inhibitor, (ii) PKA was co-immunoprecipitated from cell lysates by anti-syndecan-1 antibodies, (iii) PKA phosphorylated recombinant syndecan-1 cytoplasmic domains in vitro, and (iv) expression of a syndecan-1 construct with its invariant Ser286 replaced with a Gly was not induced by TGFβ. Together, these findings define a regulatory mechanism where TGFβ signals through PKA to phosphorylate the syndecan-1 cytoplasmic domain and increases syndecan-1 expression on epithelial cells.

Syndecans are a family of cell surface heparan sulfate proteoglycans (HSPGs), 2 which, along with the glypicans, are the major source of cell surface HS (1,2). Four members designated syndecan-1-4, each encoded by distinct genes, comprise the syndecan family. Syndecans are expressed in a cell-, tissue-, and development-specific manner. For example, expression of syndecan-1, the founding member of the syndecans, is first detected at the 4-cell stage in mouse embryos, suggesting that its expression is zygotically activated (3). In adult tissue, syndecan-1 is the major HSPG expressed on the surface of stratified epithelial cells and plasma cells, and on the basolateral surface of simple epithelial cells (4). Other cell types, such as fibroblasts, vascular smooth muscle cells, and endothelial cells, also express syndecan-1 to a lesser degree.
Syndecans bind and regulate a wide variety of soluble and insoluble ligands, such as growth factors, extracellular matrix (ECM) components, and chemokines, through their HS chains (1,2,4,5). At the cell surface, syndecans frequently serve as coreceptors that catalyze the interaction between ligands and their respective signaling receptors. Cell surface syndecans bind ligands and increase ligand concentration in the pericellular vicinity of their respective signaling receptors. Cell surface syndecans can also regulate receptor-ligand interactions by affecting the stability, conformation or oligomerization state of both ligands and receptors through their HS chains (2). Further, syndecans can function as soluble HSPGs because their extracellular ectodomains replete with all their HS chains can be proteolytically cleaved and shed into the extracellular milieu by a process known as ectodomain shedding (1,2).
The physiological significance of syndecans has yet to be clearly defined, but results from gene deletion studies in mice have provided some valuable insights. Mice made null for the syndecan-1 (6 -8), -3 (9, 10), and -4 (11,12) genes have been generated. These mutant mice are healthy, fertile, and do not show apparent pathologies, indicating that these syndecans are not essential for development or that other HSPGs have compensated for their loss during development. In contrast, syndecan-null mice show major phenotypes when subjected to various pathological conditions (6,7,(12)(13)(14)(15)(16). For example, syndecan-1-null mice resist Pseudomonas aeruginosa lung and burned skin infection (7,16), but are more susceptible to allergen-induced (14) and bleomycin-induced airway inflammation (15) relative to wild-type mice. These results indicate that certain post-developmental functions of syndecans cannot be compensated by other HSPGs.
How syndecans function specifically in post-developmental processes is not known. One of the potential underlying mechanisms is differential regulation of syndecan expression. Syndecan expression is regulated in a temporal and spatial manner, and these features may enable these HSPGs to function specifically in vivo. For instance, syndecan-1 expression in mesenchymal cells is induced at the transcriptional level by platelet derived growth factor (17), angiotensin II (17), the antimicrobial peptide PR-39 (18), and FGF-2 (19). Syndecan-1 expression can also be up-regulated post-transcriptionally in stratifying keratinocytes (20) and in mesenchymal cells during kidney formation (21), and in resident peritoneal macrophages (22). These findings suggest that regulation of expression is one of the fundamental mechanisms that allows syndecans to perform specific functions in a spatial and temporal manner in vivo.
To better understand how syndecan expression is regulated, we investigated how syndecan-1 is regulated in epithelial cells. Syndecan-1 is the predominant HSPG of epithelial cells, but the regulation of its expression in epithelial cells has not been studied in detail. Our results showed that among several bioactive agents tested only forskolin and TGF␤ significantly induced syndecan-1 expression on epithelial cells. Induction of syndecan-1 by TGF␤ was inhibited by transfection of a dominantnegative construct of PKA and by specific inhibitors of PKA. Our results also showed that the syndecan-1 cytoplasmic domain is Ser-phosphorylated by PKA when epithelial cells are treated with TGF␤. Further, substitution of the invariant Ser 286 with Gly in the syndecan-1 cytoplasmic domain abrogated its induction by TGF␤. These data support a regulatory mechanism in which TGF␤, PKA, and the syndecan-1 cytoplasmic domain coordinate to induce cell surface expression of syndecan-1 in epithelial cells.
Cells and Immunochemicals-Normal murine mammary gland (NMuMG), human lung carcinoma (A549), human keratinocyte (A431), and mouse mammary gland (C127) epithelial cells were from our culture collection, and cultured as described previously (23). 293A cells were from Invitrogen. The 281-2 rat monoclonal anti-mouse syndecan-1 ectodomain antibody was from BD Pharmingen (San Diego, CA) and the B-B4 mouse monoclonal anti-human syndecan-1 ectodomain antibody was from Serotec Inc. (Raleigh, NC). The rabbit antimouse syndecan-1 cytoplasmic domain antibody directed against the C-terminal 16 amino acids was generated and affinity-purified as described previously (24). Rabbit polyclonal antibodies against phosphoserine or phosphothreonine were from Chemicon (Temecula, CA). Rabbit anti-PKA regulatory subunits I␣ or II␣ antibodies were purchased from Calbiochem (La Jolla, CA). Horseradish peroxidase-conjugated secondary antibodies were from Jackson ImmunoResearch Lab (West Grove, PA).
Measurement of Cell Surface Syndecans-Quantification of cell surface syndecan-1 was performed by mild trypsinization and detection of the released syndecan-1 ectodomains by dot immunoblotting as described previously (23)(24)(25). Briefly, confluent or 1-day post-confluent NMuMG and C127 cells in 96-well plates and A549 cells in 24-well plates were washed once with their respective culture medium and various test samples diluted in culture medium were added to cells. Cells were incubated at 37°C for the indicated times as described in the figure legends. Cells were then washed once with ice-cold TBS (50 mM Tris, pH 7.5, 150 mM NaCl) containing 0.5 mM EDTA and then incubated for 10 min at 4°C with 10 g/ml TPCK-treated trypsin in TBS with 0.5 mM EDTA. Trypsin was subsequently inactivated by addition of an equal volume of 100 g/ml soybean trypsin inhibitor. Trypsinates were spun down and acidified by adding NaOAc (pH 4.5) to a final concentration of 50 mM, and blotted onto Immobilon Nyϩ membranes using a dot blotting apparatus. By acidifying the samples, only highly anionic molecules, such as syndecan-1, are retained by the cationic Immobilon Nyϩ membrane while most proteins pass through the membrane during dot blotting. To obtain measurements in the linear range, various sample volumes were blotted onto Immobilon Nyϩ membranes. The blotted membranes were developed with 281-2 anti-mouse or B-B4 antihuman syndecan-1 antibodies and ECL as described previously (23)(24)(25). The developed blots were scanned and quantified using the public domain NIH Image software.
Cell surface levels of syndecan-4 were measured by FACS because trypsin digestion destroyed the antigenic epitope of the Ky8.2 anti-mouse syndecan-4 antibody. Briefly, NMuMG cells in 6-well plates were incubated with or without 10 ng/ml TGF␤2 for 2 days. Cells were detached by incubating with 5 mM EDTA in PBS for 30 min at 4°C, washed, and fixed in 4% paraformaldehyde/PBS for 30 min on ice. Fixed cells were incubated with or without 20 g/ml Ky8.2 anti-syndecan-4 antibodies in 1% bovine serum albumin/PBS at room temperature for 2 h. Cells were washed and incubated with fluorescein isothiocyanate-labeled anti-rat IgG mouse antibodies. Cells were fixed again with 4% paraformaldehyde/PBS and washed three times with 1% bovine serum albumin/PBS. The fluorescence associated with cells was measured by a flow cytometer (EPICS-XL, Beckman-Coulter, Fullerton, CA).
Western Immunoblotting of Syndecan-1-NMuMG cells were incubated with or without 10 ng/ml of TGF␤2 for 2 days, and syndecan-1 was released from the cell surface by mild trypsin digestion as described above. Trypsinates were acidified by the addition of NaOAc (pH 4.5) and NaCl to final concentrations of 100 and 300 mM, respectively, and syndecan-1 was partially purified by DEAE ion-exchange chromatography. Samples containing ϳ30 ng of partially-purified syndecan-1 were resuspended in digestion buffer (50 mM Tris, pH 7.5, 50 mM NaOAc, 5 mM EDTA, 2 mM phenylmethylsulfonyl fluoride) and digested with 20 milliunits/ml heparinase III and 30 milliunits/ml CS ABC lyase or with 30 milliunits/ml CS ABC lyase only for 3 h at 37°C. 30 ng of CS ABC lyase-digested and undigested syndecan-1 were fractionated by 6% SDS-PAGE and Western-blotted onto Immobilon Nyϩ membranes. Syndecan-1 digested with heparinase III and CS ABC lyase was fractionated by 5-15% gradient SDS-PAGE, and Western-blotted onto nitrocellulose membranes. Membranes were probed with the 281-2 anti-mouse syndecan-1 antibody.
Transfection of the PKA Dominant-negative Construct-The PKA dominant-negative expression construct, MT-REV ABneo, was kindly provided by Dr. Stanley McKnight (Univ. of Washington, Seattle, WA). MT-REV AB -neo harbors a PKA cDNA insert with point mutations in both cAMP-binding sites (26). Transcription of the dominant-negative construct is driven by the metallothionein promoter and expression can be increased by Ն10-fold by the addition of Zn 2ϩ or Cd 2ϩ . However, we tested the effects of this mutant PKA construct without heavy metal stimulation because basal expression of MT-REV AB -neo was shown to be sufficient to significantly inhibit endogenous PKA activation in T84 intestinal epithelial cells (26) and to avoid unforeseen effects of Zn 2ϩ or Cd 2ϩ . To determine the effects of the PKA dominant-negative construct on TGF␤-induced syndecan-1 expression, NMuMG cells in 96-well plates were transfected with 4 g of MT-REV AB -neo or pUC13 (mock transfection) by Lipofectamine. After 2 days, cells were stimulated with 10 ng/ml of TGF␤2 or incubated with fresh culture medium for 2 days at 37°C, and cell surface levels of syndecan-1 were measured.
Expression of Recombinant Syndecan Constructs-A cDNA construct encoding Leu 17 -Pro 234 in the ectodomain of mouse syndecan-1 was subcloned into the His 6 tag expression vector, pQE30 (Qiagen). Expression and purification were performed with the B-PER His 6 fusion protein purification kit. Briefly, gene expression was induced by incubating transformed Escherichia coli (M15[pREP4]) in logarithmic growth with 1.25 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) for 3 h. Cells were lysed with B-PER reagent, cell lysates were fractionated by Ni 2ϩ -chelated agarose chromatography, and bound proteins were eluted with B-PER elution buffer. A cDNA construct encoding Lys 280 -Ala 311 of the cytoplasmic domain of mouse syndecan-1, and a construct encoding Arg 170 -Ala 202 of the cytoplasmic domain of mouse syndecan-4 were subcloned into the GST fusion protein expression vector, pGEX2T. Expression was induced by incubating transformed E. coli (BL21) in broth containing 1.25 mM IPTG for 3 h. Induced E. coli cells were lysed by incubation in PBS containing 10 mg/ml lysozyme and 0.5 mM EDTA for 30 min at 37°C. Recombinant syndecan proteins were purified from cell lysates by glutathione-Sepharose 4B chromatography. For some experiments, the GST tag was removed from the cytoplasmic domain of syndecan-1 by thrombin cleavage.
Co-immunoprecipitation-NMuMG cells treated with or without TGF␤2 were lysed in PBS containing 0.1% (v/v) Triton X-100 and 0.5 mM EDTA. The lysate was centrifuged at 13,000 ϫ g for 15 min and 150 l of the supernatant was preabsorbed with 20 l of protein G-agarose beads for 2 h at 4°C. The preabsorbed cell lysate was then incubated with 3 g of 281-2 anti-syndecan-1 or Ky8.2 anti-syndecan-4 antibody for 6 h at 4°C with gentle agitation. Protein G-agarose beads (20 l) were added to each sample and incubated for an additional 2 h at 4°C. Beads were pelleted by centrifugation, washed with PBS, and resuspended in non-reducing SDS-PAGE sample buffer. Samples were resolved by SDS-PAGE, Western-blotted onto nitrocellulose, and probed with anti-PKA regulatory subunit antibodies. In some experiments, 281-2 immunoprecipitation was performed in the presence of an excess concentration (150 g/ml) of recombinant syndecan-1 cytoplasmic domain devoid of GST.
To determine if the syndecan-1 cytoplasmic domain is Seror Thr-phosphorylated, cells treated with TGF␤2 in the presence or absence of KT5720 for 2 days were treated with 10 g/ml TPCK-trypsin and lysed with PBS containing 650 mM NaCl, 1% Triton X-100, and a mixture of phosphatase inhibitors. Cell lysates were diluted 5-fold with autoclaved de-ionized H 2 O, and immunoprecipitated with the anti-syndecan-1 cytoplasmic domain antibody and protein A-agarose. The immunoprecipitated material was resolved by SDS-PAGE followed by Western immunoblotting with antiphosphoserine, anti-phosphothreonine, or anti-syndecan-1 cytoplasmic domain antibodies.
In Vitro Phosphorylation of Recombinant Syndecan Proteins by PKA-In vitro phosphorylation of recombinant syndecan-1 and -4 cytoplasmic domains and syndecan-1 ectodomain by PKA was assessed using the PKA assay kit (Upstate). Briefly, syndecan-1 and -4 cytoplasmic domains (10 and 6 g, respectively) or syndecan-1 ectodomain (6 g) were incubated with or without 800 units of the PKA catalytic domain and 10 Ci of [␥-32 P]ATP (ϳ3000 Ci/mmol) for 10 min at 30°C in 60-l reaction buffer provided with the kit. One-fifth of the reaction mixture was analyzed by SDS-PAGE and autoradiography.
Adenoviral Expression of Wild Type and Point-mutated Syndecan-1 Constructs-To generate the S286G syndecan-1 expression construct, a single nucleotide substitution was made in full-length mouse syndecan-1 cDNA by site-directed mutagenesis using the sequential PCR protocol. For replacement of invariant Ser 286 to Gly, we used 5Ј-GACGAAGGCG-GCTACTCCTT-3Ј and its complementary primer (substituted nucleotide is indicated in bold) in the first PCR reaction. PCR products in the second PCR reaction were subcloned into pcDNA3.1V5His.
Adenoviruses bearing wild type (pAd/Sdc1-WT/V5) and S286G-mutated mouse syndecan-1 (pAd/Sdc1-S286G/V5) were prepared with a ViraPower Adenovirus Expression System according to the manufacturer's instructions. Briefly, fulllength cDNAs minus the stop codon of wild type and S286G syndecan-1 were amplified by PCR with Accuprime with corresponding templates described above, and the CDS was subcloned into the pENTER/S.D./D-TOPO vector in-frame with a pENTER Directional TOPO cloning kit. After verifying the correct sequence, the insert was transferred into pAd/CMV/V5-DEST vector by the Gateway system using LR clonase. To obtain virus particles, plasmids were linearized by PacI (New England BioLabs, Ipswich, MA) digestion and transfected into 293A cells with Lipofectamine 2000. When most cells were detached (ϳ1-2 weeks), cells were harvested by gentle pipetting in culture media. Cells were lysed by freeze/thawing, centrifuged, and supernatants were collected as crude viral lysates. To amplify virus titers, 100 l of crude viral lysates were added to fresh 293A cells and cultured for several days until all cells were detached. Virus-enriched supernatants were collected, and viral titers were determined by the plaque-forming assay with 293A cells.
A431 cells in 96-well plates were transduced with adenovirus harboring wild type or mutant S286G syndecan-1 with an MOI of 15 in culture medium. The culture medium was replaced with fresh medium the next day, and after an additional 24 h of incubation, cells were incubated with or without TGF␤2 for 2 days. Cell surface syndecan-1 was quantified by mild trypsinization and dot immunoblotting as described above.

Expression of Cell Surface Syndecan-1 Is Highly Regulated in NMuMG
Epithelial Cells-We initially tested the ability of several bioactive agents to regulate the cell surface expression of syndecan-1 in NMuMG cells (Fig. 1A). NMuMG cells were incubated for 24 h with agents that induce syndecan-1 shedding (PMA, ceramide, EGF), calcium ionophores (ionomycin, A23187), a secondary messenger analogue (dbcGMP), an adenylate cyclase activator that stimulates cAMP generation (forskolin), an inducer of the Toll-like receptor 4-mediated innate host response (LPS), or molecules that have been shown to down-(TNF␣) or up-regulate (TGF␤, FGF-2) syndecan-1 expression in endothelial cells and fibroblasts, respectively. Among these, only forskolin and TGF␤2 markedly increased cell surface levels of syndecan-1 by ϳ3and 5-fold, respectively, over control cells incubated with fresh culture medium. PMA, ceramide, EGF, dbcGMP, ionomycin, A23187, and LPS had marginal effects on cell surface levels of syndecan-1. Surprisingly, TNF␣ and FGF-2 also had little effect on syndecan-1 expression, despite the fact that these molecules have been shown to transcriptionally regulate syndecan-1 expression in other cell types. Because epithelial cells express both p55 and p75 TNF␣ receptors (27) and proliferate in response to FGF-2 (28), these findings suggest that downstream signaling components that are essential in the regulation of syndecan-1 expression by TNF␣ and FGF-2 are lacking in epithelial cells. Together, these results indicate that cell surface levels of syndecan-1 are tightly regulated in NMuMG epithelial cells.

TGF␤ Is a General Inducer of Cell Surface Syndecan-1 in Epi-
thelial Cells-To further explore how TGF␤ induces syndecan-1 expression in epithelial cells, we next examined the time course of induction by various concentrations of TGF␤2. Significant induction of cell surface syndecan-1 was observed when NMuMG cells were incubated with Ն1 ng/ml of TGF␤2 for 24 or 48 h, but not 6 h (Fig. 1B). Interestingly, TGF␤2 did not increase steady-state mRNA levels of syndecan-1 in NMuMG cells incubated with TGF␤2 for 9 or 23 h. Further, cycloheximide at 5 and 10 g/ml did not affect the inductive effect of TGF␤2, suggesting that TGF␤ induces syndecan-1 in a posttranslational manner (data not shown).
We also examined if other isoforms of the TGF␤ family similarly induce syndecan-1 expression and found that TGF␤1 and TGF␤3 also up-regulate cell surface syndecan-1 in NMuMG

Syndecan-1 Induction by the TGF␤-PKA Pathway
cells in a dose-dependent manner ( Fig. 2A). Interestingly, cell surface levels of syndecan-4 were not increased in NMuMG cells treated with TGF␤2 for 2 days (Fig. 2B), suggesting that the inductive effects of TGF␤ are specific to syndecan-1. Further, all three TGF␤ isoforms induced syndecan-1 expression in A549 human lung carcinoma and C127 murine mammary gland epithelial cells, albeit differences in the degree of induction by the three TGF␤ isoforms (Fig. 2C). TGF␤ first binds to the type II TGF␤ receptor and then the type I receptor is recruited to form a receptor-ligand complex. The type II receptor phosphorylates and activates the type I receptor, and the phosphorylated type I receptor phosphorylates and activates Smad proteins. Thus, differences in the ability of TGF␤ isoforms to induce syndecan-1 expression in the epithelial cell lines may have been caused by differences in the binding affinities of isoforms or by varied expression of the TGF␤ receptors in the epithelial cell lines. Varied expression of other TGF␤-binding molecules on the cell surface (e.g. betaglycan/type III receptor) may also have contributed to differences in the response of epithelial cell lines tested. Nevertheless, these results indicate that the three isoforms of TGF␤ are potent inducers of cell surface syndecan-1 in epithelial cells.
We next examined if TGF␤ induction affects the structure of glycosaminoglycan (GAG) chains attached on syndecan-1. We analyzed cell surface syndecan-1 from NMuMG cells treated with or without TGF␤2 by Western immunoblotting and found that HS chains attached to syndecan-1 from TGF␤2-stimulated cells are larger than those from untreated cells (data not shown). Rapraeger (29) also reported an increase in the addition of HS and CS chains to syndecan-1 upon treatment with TGF␤. These findings suggest that TGF␤ may also regulate the expression or activity of GAG chain polymerization and modification enzymes. These findings also illustrate a major difference between the induction of cell surface syndecan-1 in epithelial cells and mesenchymal cells because GAG chains on induced syndecan-1 are generally smaller FIGURE 2. TGF␤ is a general inducer of cell surface syndecan-1 in epithelial cells. A, NMuMG cells were incubated with fresh culture medium with or without 1, 5, or 10 ng/ml of TGF␤1 or ␤3 for 2 days at 37°C. Cell surface levels of syndecan-1 were measured by mild trypsinization and dot immunoblotting of the trypsinate. B, NMuMG cells were incubated with or without 10 ng/ml TGF␤2 and cell surface levels of syndecan-4 were measured by FACS. C, C127 and A549 cells were stimulate with or without 10 ng/m of TGF␤1, TGF␤2, or TGF␤3, and cell surface levels of syndecan-1 were quantified by mild trypsinization and dot immunoblotting. Results are shown as mean fold increase over control Ϯ S.E. of triplicate measurements, where control is the level of syndecans on cells incubated with fresh culture medium for the indicated times.
TGF␤ Induces Syndecan-1 through a PKA-dependent Mechanism-Syndecan-1 expression in resident peritoneal macrophages (22) and mesenchymal cells (30) is induced through a mechanism dependent on the cAMP-PKA signaling pathway. Our result showing that forskolin, an adenylate cyclase activator, increases cell surface levels of syndecan-1 on NMuMG cells suggested that the cAMP-PKA pathway is also important in the induction of syndecan-1 by TGF␤ in epithelial cells. TGF␤ has been reported to activate the cAMP-PKA pathway in mouse kidney mesangial cells (31) and the TGF␤-induced Smad3⅐Smad4 complex has been shown to directly activate PKA in mink lung epithelial cells (32). Thus, we tested if TGF␤ induces syndecan-1 expression in epithelial cells through activation of PKA.
We first examined the effects of specific PKA inhibitors, KT5720 and myristoylated PKA peptide inhibitor, on TGF␤induced syndecan-1 expression in NMuMG cells. Confluent NMuMG cells were incubated with TGF␤2 in the absence or presence of various concentrations of KT5720, PKA peptide inhibitor, or the PKC inhibitor bisindolylmaleimide for 2 days. Cell viability measured by the MTT assay was similar among NMuMG cells treated with these kinase inhibitors (data not shown). Both KT5720 and the myristoylated PKA peptide inhibitor, but not bisindolylmaleimide, inhibited syndecan-1 induction by TGF␤2 in a dose-dependent manner (Fig. 3A). We also tested the effects of transfecting a dominant-negative cDNA construct of PKA. The dominant-negative PKA construct harbors point mutations in both cAMP binding sites, preventing its activation by cAMP (26). NMuMG cells were transiently transfected with the dominant-negative PKA construct or the empty vector, incubated with TGF␤2 for 2 days, and then assayed for cell surface levels of syndecan-1. Relative to mock-transfected cells, NMuMG cells transfected with the dominant-negative PKA construct showed a 2-fold decrease in cell surface syndecan-1 (Fig. 3B). Together, these data indicate that TGF␤ induces syndecan-1 in epithelial cells in a PKA-dependent manner.
TGF␤ Increases Ser Phosphorylation of the Syndecan-1 Cytoplasmic Domain-PKA is a Ser/Thr kinase implicated in a wide variety of biological processes (33). The cytoplasmic domain of syndecan-1 contains two Ser residues, of which one is invariant among all syndecans, and one Thr residue. Thus, we next examined whether the cytoplasmic domain of syndecan-1 is Ser-or Thr-phosphorylated upon TGF␤ stimulation. NMuMG cells were incubated with TGF␤2 for various times, treated with mild trypsin to remove the ectodomain, and extracted with detergent. Cell extracts were immunoprecipitated with anti-syndecan-1 cytoplasmic domain antibodies and then immunoblotted with anti-phosphothreonine, anti-phosphoserine, or anti-syndecan-1 cytoplasmic domain antibodies. The anti-syndecan-1 cytoplasmic domain antibody reacts with syndecan-1, but not syndecan-4, cytoplasmic domains (Fig. 4A). At all times post-TGF␤2 stimulation, the immunoprecipitated Sdc1 cytoplasmic domain did not react with anti-phosphothreonine antibodies (data not shown). In contrast, the immunoprecipitated syndecan-1 cytoplasmic domains were detected with anti-phosphoserine antibodies and band intensities were increased by 4-fold at 48-h post-TGF␤2 incubation. (Fig. 4B). Further, the increase in Ser phosphorylation of the syndecan-1 cytoplasmic domain was abrogated when cells were coincubated with TGF␤2 and the PKA inhibitor KT5720 (Fig. 4C). These data indicate that TGF␤ enhances Ser phosphorylation of the syndecan-1 cytoplasmic domain by PKA.
PKA Binds and Phosphorylates the Cytoplasmic Domain of Syndecan-1-These results led us to examine whether PKA directly binds and Ser phosphorylates the cytoplasmic domain of syndecan-1. NMuMG cells were stimulated with TGF␤2, cell extracts were immunoprecipitated with 281-2 anti-syndecan-1 or Ky8.2 anti-syndecan-4 antibodies, and immunoprecipitated materials were probed with either anti-PKA regulatory subunit I␣ (RI␣) or anti-PKA RII␣ antibodies. As shown, the 281-2 antisyndecan-1 antibody co-immunoprecipitated both PKA RI␣ and RII␣ (Fig. 5A) subunits, and co-immunoprecipitation was inhibited in the presence of excess recombinant syndecan-1 cytoplasmic domain (Fig. 5B). In contrast, Ky8.2 anti-syndecan-4 antibodies did not co-immunoprecipitate the PKA regulatory subunits (Fig. 5A). These data indicate that PKA specifically binds to the cytoplasmic domain of syndecan-1 in vivo.
We next assessed if the syndecan-1 cytoplasmic domain is a substrate for PKA in an in vitro kinase assay. Purified recombinant syndecan-1 cytoplasmic domain, syndecan-4 cytoplasmic domain, or syndecan-1 ectodomain were incubated with the recombinant PKA catalytic subunit and [␥-32 P]ATP in the absence or presence of PKA inhibitor peptides. Syndecan-1 cytoplasmic domains, but not syndecan-1 ectodomains, were phosphorylated by PKA and this was inhibited by the PKA inhibitor peptide (Fig. 5C). Surprisingly, the syndecan-4 cytoplasmic domain was also phosphorylated by PKA. These results suggest that PKA phosphorylates the invariant Ser 286 in the syndecan-1 cytoplasmic domain.
Phosphorylation of the Invariant Ser 286 Is Essential for Induction of Cell Surface Syndecan-1 by TGF␤-To determine if the invariant Ser 286 is indeed phosphorylated by PKA and if this is important in the induction of syndecan-1 by TGF␤, we examined the effects of point-mutating Ser 286 . Human A431 cells were transduced with adenoviruses harboring wild type or S286G point-mutated mouse syndecan-1 constructs, stimulated with 2.5 or 10 ng/ml TGF␤2 for 2 days, and cell surface levels of mouse syndecan-1 were measured with the 281-2 antimouse syndecan-1 antibody that does not cross-react with human syndecan-1. As shown, surface levels of wild-type mouse syndecan-1 expressed in A431 cells were increased by 3-fold when treated with 10 ng/ml TGF␤2 relative to those treated with culture medium (Fig. 6A). Cell surface expression of endogenous human syndecan-1 was similarly induced by TGF␤2 in A431 cells transduced with wild-type or S286G mouse syndecan-1, indicating that expression of mouse syndecan-1 did not have unanticipated secondary effects. In contrast, cell surface levels of the S286G mouse syndecan-1 were not induced by TGF␤2 (Fig. 6A), indicating that Ser 286 is essential for TGF␤ induction. Furthermore, wild-type mouse syndecan-1 expressed in A431 cells was Ser-phosphorylated in response to TGF␤2 in a PKA-dependent manner, but not S286G mouse syndecan-1 (Fig. 6B). Interestingly, PKA bound to both wild type and S286G mouse syndecan-1 in vivo (Fig.  6C), indicating that Ser 286 is dispensable for PKA binding to the syndecan-1 cytoplasmic domain. Taken together, these data indicate that PKA phosphorylation of the invariant Ser 286 in the syndecan-1 cytoplasmic domain is essential for TGF␤ induction of cell surface syndecan-1.

DISCUSSION
Our results presented here define how TGF␤ signals through PKA to induce cell surface expression of syndecan-1 in epithelial cells. TGF␤ induced syndecan-1 in a time-and dose-dependent, and post-translational manner. All three TGF␤ isoforms increased cell surface expression of syndecan-1 in several mouse and human epithelial cell lines, suggesting that TGF␤ is a general inducer of epithelial syndecan-1. Induction of syndecan-1 by TGF␤ was inhibited by transfection of a dominantnegative construct of PKA and by specific inhibitors of PKA. Further, our data showed that: (i) Ser phosphorylation of the cytoplasmic domain of syndecan-1 is increased by TGF␤ treatment and inhibited by a PKA inhibitor; (ii) PKA binds and phosphorylates the syndecan-1 cytoplasmic domain;  -sdc1CPD). B, NMuMG cells were incubated with TGF␤2 for the indicated times and then treated with trypsin to remove the ectodomain. Whole cell lysates were immunoprecipitated with anti-syndecan-1 cytoplasmic domain antibodies. The immunoprecipitated syndecan-1 cytoplasmic domain was Western-immunoblotted with anti-phosphoserine or anti-syndecan-1 cytoplasmic domain antibodies. C, NMuMG cells were incubated with culture medium, TGF␤2, KT5720, or TGF␤2 and KT5720 for 2 days, and Ser phosphorylation of the immunoprecipitated syndecan-1 cytoplasmic domain was determined. and (iii) mutating the invariant Ser 286 in the cytoplasmic domain abrogates both TGF␤-induced cell surface expression of syndecan-1 and phosphorylation of the syndecan-1 cytoplasmic domain by PKA. On the basis of these findings, we propose a novel regulatory mechanism in which TGF␤ signals through PKA phosphorylation of the syndecan-1 cytoplasmic domain to induce expression of syndecan-1 at the epithelial cell surface.
TGF␤, a prototype of the TGF␤ superfamily (34), regulates a diverse array of cellular processes, such as adhesion, proliferation and differentiation, that play key roles in physiological mechanisms including development, tissue repair, inflammation and tumorigenesis (35,36). TGF␤ functions primarily through its signaling receptor complex that activates Smad proteins (37). However, our results indicate that TGF␤ induces cell surface syndecan-1 in a PKA-dependent manner. How TGF␤ communicates with PKA is not clear, but at least in mesangial cells, TGF␤ has been shown to activate PKA (31). Moreover, the Smad3⅐Smad4 complex induced by TGF␤ has been shown to directly activate PKA (32). These observations suggest that the TGF␤-Smad pathway directly activates PKA to induce syndecan-1 expression at the cell surface. Alternatively, the TGF␤-Smad pathway may induce expression of an intermediate signaling molecule that activates PKA to up-regulate cell surface syndecan-1. Support for this latter mechanism comes from our observation that induction of syndecan-1 by TGF␤ requires at least a 1-day incubation and maximal induction is observed when epithelial cells are incubated with TGF␤ for 2 days. This putative mechanism is also supported by the finding that cell surface syndecan-1 is still induced when epithelial cells are pulsed with TGF␤ for only 3 h and then incubated without TGF␤ for 2 days (data not shown). Connective tissue growth factor (CTGF) is a potential candidate for the intermediate signaling molecule because it is a one of the soluble factors induced directly by TGF␤ (38). However, CTGF is produced mainly by fibroblast, not by epithelial cells (39), and CTGF is not known to activate PKA. Further studies are needed to clarify how TGF␤ activates PKA when inducing syndecan-1 in epithelial cells.
The cytoplasmic domain of syndecans is highly homologous and contains several signature motifs, including one invariant Ser, three invariant Tyr, and a Glu-Phe-Tyr-Ala PDZ binding domain at the C terminus. Tyr phosphorylation of syndecan-1 through -4 by several PTKs (40 -42) and Ser phosphorylation

Syndecan-1 Induction by the TGF␤-PKA Pathway
of syndecan-2 (43,44) and -4 (45, 46) by PKC have been reported, but to our knowledge, this is the first demonstration of Ser phosphorylation of the syndecan-1 cytoplasmic domain by PKA.
The cytoplasmic domain of syndecan-1 has two Ser residues, of which one is invariant, and one Thr residue. Sequences flanking the two Ser residues do not conform to any of the three consensus phosphorylation sites for PKA (Arg-Arg-X-Ser/Thr, Arg/Lys-X-X-Ser/Thr, and Arg/Lys-X-Ser/Thr) (47), but the sequence surrounding the Thr residue does (Lys 302 -Pro-Thr 304 ). However, several biological substrates of PKA do not contain these consensus sequences and several molecules that contain consensus sequences are not phosphorylated by PKA, indicating that identification of potential phosphorylation sites based solely on the substrate primary structure is implausible and that direct experimental verification is essential. Consistent with these facts, our study showed that the invariant Ser 286 , but not Thr, residue is phosphorylated by PKA when epithelial cells are incubated with TGF␤. The functional significance of the invariant Ser was further underscored by the finding that replacement of the invariant Ser 286 with Gly in the syndecan-1 cytoplasmic domain abrogates its induction at the cell surface by TGF␤. These data suggest an intriguing mechanism where TGF␤-induced PKA phosphorylation of the invariant Ser 286 in the syndecan-1 cytoplasmic domain mediates the induction of cell surface syndecan-1.
The PKA holoenzyme is a heterotetramer comprised of two regulatory (R) subunits that maintain two catalytic (C) subunits in an inactive state. Binding of cAMP to tandem sites in each R subunit dissociates and activates the PKA C subunits. PKA activity is also regulated by protein kinase A-anchoring proteins (AKAPs), which are signalorganizing molecules that compartmentalize various kinases that are regulated by second messengers (48). Interestingly, several AKAPs, such as ezrin (49) and CASK/LIN-2 (50), bind to the C-terminal PDZ binding domain in the syndecan cytoplasmic domain. Moreover, our results show that regulatory subunits of PKA are constitutively bound to the syndecan-1 cytoplasmic domain. These data suggest the possibility that syndecan-1 anchors and compartmentalizes PKA either directly or indirectly through AKAPs. This mechanism would allow syndecan-1 to localize signaling molecules that regulate its cell surface expression, such as PKA, in the vicinity of its cytoplasmic domain.
How phosphorylation of the syndecan-1 cytoplasmic domain by PKA increases cell surface syndecan-1 expression remains to FIGURE 6. Phosphorylation of the invariant Ser 286 by PKA mediates the induction of cell surface syndecan-1 by TGF␤. A, human A431 cells were transduced with adenoviruses harboring wild type or S286G mouse syndecan-1 cDNA constructs and incubated with 0, 2.5, or 10 ng/ml TGF␤2 for 2 days. Cell surface levels of transduced mouse syndecan-1 and endogenous human syndecan-1 were measured by mild trypsinization and dot immunoblotting using 281-2 anti-mouse syndecan-1 or B-B4 anti-human syndecan-1 antibodies. Each data point is the mean Ϯ S.E. of triplicate determinations. B, A431 cells transduced with wild-type or S286G mouse syndecan-1 were incubated with or without TGF␤2 (10 ng/ml) and KT5720 (2 M) for 2 days. Syndecan-1 was immunoprecipitated from cell lysates using 281-2 anti-mouse syndecan-1 antibodies coupled to Ultralink Biosupport medium. The precipitated mouse syndecan-1 was treated with 20 milliunits/ml heparinase III and 30 milliunits/ml CS ABC lyase, and syndecan-1 core proteins were analyzed by Western immunoblotting with anti-phosphoserine or rabbit polyclonal anti-syndecan-1 ectodomain antibodies. C, A431 cells expressing wild type or S286G mouse syndecan-1 were incubated with TGF␤2 as indicated and cell lysates were immunoprecipitated with 281-2 anti-syndecan-1 antibodies. The immunoprecipitated materials were analyzed by Western immunoblotting using antibodies directed against the PKA regulatory subunit I␣ (anti-PKA RI␣) or II␣ (anti-PKA RII␣). be determined. However, PKA has been shown to enhance exocytosis (51,52) and inhibit endocytosis (53,54) of cell surface proteins and increase their cell surface levels. Increased exocytosis of intracellular syndecan-1, whether from a newly synthesized or stored pool, could potentially explain how cell surface syndecan-1 levels are increased post-translationally by PKA. Alternatively, reduced endocytosis of cell surface syndecan-1 could increase syndecan-1 levels in a post-translational manner. In fact, our preliminary data show that the turnover rate of wild type syndecan-1 is decreased by TGF␤ treatment, but not with syndecan-1 harboring the S286G mutation. If true, these data would indicate that PKA participates in a previously unsuspected mechanism that regulates the level of cell surface syndecan-1 by modulating the interaction of the syndecan-1 cytoplasmic domain with the endocytic apparatus. These results add a new dimension to the fine-tuning of syndecan-1 functions.