Molecular Mediators for Raft-dependent Endocytosis of Syndecan-1, a Highly Conserved, Multifunctional Receptor*

Background: Endocytosis via rafts remains incompletely characterized. Results: Raft-dependent endocytosis of syndecan-1 occurs in two phases, each requiring a kinase and a corresponding cytoskeletal partner. Each phase depends on MKKK, a novel, conserved motif within syndecan-1. Conclusion: These findings demonstrate the molecular choreography behind endocytosis of a raft-dependent receptor. Significance: Syndecans mediate uptake of biologically and medically important ligands. Endocytosis via rafts has attracted considerable recent interest, but the molecular mediators remain incompletely characterized. Here, we focused on the syndecan-1 heparan sulfate proteoglycan, a highly conserved, multifunctional receptor that we previously showed to undergo raft-dependent endocytosis upon clustering. Alanine scanning mutagenesis of three to five consecutive cytoplasmic residues at a time revealed that a conserved juxtamembrane motif, MKKK, was the only region required for efficient endocytosis after clustering. Endocytosis of clustered syndecan-1 occurs in two phases, each requiring a kinase and a corresponding cytoskeletal partner. In the initial phase, ligands trigger rapid MKKK-dependent activation of ERK and the localization of syndecan-1 into rafts. Activation of ERK drives the dissociation of syndecan-1 from α-tubulin, a molecule that may act as an anchor for syndecan-1 at the plasma membrane in the basal state. In the second phase, Src family kinases phosphorylate tyrosyl residues within the transmembrane and cytoplasmic regions of syndecan-1, a process that also requires MKKK. Tyrosine phosphorylation of syndecan-1 triggers the robust recruitment of cortactin, which we found to be an essential mediator of efficient actin-dependent endocytosis. These findings represent the first detailed characterization of the molecular events that drive endocytosis of a raft-dependent receptor and identify a novel endocytic motif, MKKK. Moreover, the results provide new tools to study syndecan function and regulation during uptake of its biologically and medically important ligands, such as HIV-1, atherogenic postprandial remnant lipoproteins, and molecules implicated in Alzheimer disease.

There has been considerable recent interest in the molecular participants in raft-mediated endocytosis to compare and con-trast with uptake via coated pits (1)(2)(3)(4). A broad search identified a set of kinases and structural proteins involved in the endocytosis of simian virus 40 (SV40) 2 (5), a ligand that enters via caveolar and noncaveolar rafts for delivery to the endoplasmic reticulum, bypassing canonical endosomes and lysosomes (6,7). Nevertheless, although the features of numerous coated pit receptors that allow recruitment of endocytic machinery have been identified (8 -10), the characteristics of specific raft receptors that trigger endocytosis have not been defined, not even for SV40 receptors (7).
For several reasons, we focused on the syndecan-1 heparan sulfate proteoglycan (HSPG) as an attractive system to investigate molecular determinants of raft-mediated endocytosis. First, we previously reported that the syndecan-1 HSPG directly mediates the internalization and lysosomal delivery of ligands through a novel endocytic pathway via rafts, independent from coated pits (11,12). By several criteria, endocytosis via syndecan was identical with endocytosis via a chimeric receptor, FcR-Synd1, that consists of the ectodomain of the IgG Fc receptor Ia linked to the transmembrane and cytoplasmic domains of human syndecan-1, thereby indicating a dependence on these domains. Efficient endocytosis by this pathway is triggered by clustering of syndecan-1 or the chimera. Syndecan-1 clustering would be expected upon binding a multivalent ligand, such as a virus or a lipoprotein. Clustering of the syndecan transmembrane and cytoplasmic domains causes energyindependent movement into cholesterol-rich and detergentinsoluble membrane rafts within 10 min, a necessary step before endocytosis. Uptake into the cell proceeds with a t1 ⁄ 2 of 1 h and requires an unspecified tyrosine kinase activity and intact actin microfilaments (11,12).
The second reason to focus on syndecan-1-mediated endocytosis is that cell-surface HSPGs, including syndecan-1, participate in the catabolism of a number of important ligands, including infectious agents, harmful atherogenic lipoproteins, enzymes, growth factors, platelet secretory products, and proteins implicated in Alzheimer disease (11,(13)(14)(15)(16)(17). Exploring the molecular mechanisms for syndecan-1-mediated endocytosis will substantially advance our understanding of how cells handle these biologically and medically significant ligands.
Third, the transmembrane and cytoplasmic domains of human syndecan-1 are highly conserved, indicating key essential functions. These domains show over 50% aminoacyl identity with the sole Drosophila syndecan, indicating a remarkable degree of preservation during half a billion years of evolution (18). Thus, the endocytic determinants and intracellular partners of syndecan-1 are likely to have broad significance.
Based on sequence alignments, the syndecan-1 cytoplasmic tail has been divided into the first conserved region (C1), the variable domain (V), and the second conserved region (C2) (Fig.  1A) (14,19,20). Each of these cytoplasmic portions has been shown to interact with specific intracellular binding partners and to mediate several cellular functions, although none are directly related to the process of endocytosis. Previous studies indicate that three regions interact with the cytoskeleton, namely the second half of C1 (21), the V-domain (22), and the terminal C2 domain (14,23,24). Moreover, C2 controls syndecan recycling after endocytosis (25).
In this study, we sought to identify determinants within the syndecan-1 cytoplasmic tail, as well as their intracellular partners, that mediate efficient endocytosis upon clustering. Surprisingly, this work implicates none of the known cytoskeletoninteracting domains of syndecan-1 in the endocytosis of multivalent ligands. Instead, we identified a single, conserved juxtamembrane motif, MKKK, in the syndecan-1 cytoplasmic tail that mediates the sequential activation of two kinases, ERK and then Src. Upon activation, the two kinases each control the interaction of syndecan-1 with two key cytoskeletal molecules to mediate efficient endocytosis. Portions of this work were presented at the 2008 and 2011 American Heart Association Scientific Sessions (26,27).

EXPERIMENTAL PROCEDURES
Molecular Methods-Our FcR-Synd1 chimera was previously described (11,12); now it is expressed in the pcDNA3.1 plasmid (Invitrogen). Alanine scanning mutagenesis of the syndecan-1 cytoplasmic tail within FcR-Synd1 was performed with the QuikChange kit (catalog no. 200518, Stratagene-Agilent Technology, Santa Clara, CA), using our unmutated FcR-Synd1 expression plasmid as template and the mutagenesis primers listed in supplemental Table I. All mutants were sequenced to verify the introduction of DNA changes. McArdle 7777 rat hepatoma cells were obtained from the American Type Culture Collection (Manassas, VA; catalog no. CRL-1601) and cultured as described previously (16,28). The unmutated FcR-Synd1 plasmid, all mutant plasmids, and the empty pcDNA3 vector were transfected one at a time into McArdle cells, using the FuGENE 6 reagent (Roche Applied Science). Stably expressing clones were selected with G418, followed by verification of expression by immunoblots of whole-cell homogenates using anti-FcR antibodies. To assess cell-surface display of the chimera and its mutants, we measured cell-surface binding of ligand, i.e. 125 I-labeled nonimmune human IgG (unlabeled IgG purchased from Rockland Immunochemicals, Gilbertsville, PA, catalog no. 009-0102, and then radioiodinated). To avoid receptor internalization or recycling, this particular assay was performed entirely at 4°C.
Cell Culture, Ligand Catabolism, and Protein Analyses-Raft localization and internalization of 125 I-labeled IgG by cells expressing the unmutated parental FcR-Synd1 chimera and the alanine scanning mutants were performed following our published protocols (11,12). In brief, the ligand for FcR-Synd1, 125 I-labeled nonimmune human IgG, was bound at 4°C to the surface of these McArdle cell lines. Unbound material was washed away, and then the cells were incubated for 5-60 min at 37°C, in the absence or presence of our clustering agent (goat F(abЈ) 2 against human IgG Fab; Rockland Immunochemicals, catalog no. 709-1118). Raft localization was assessed by cold Triton insolubility and internalization by resistance to an acid wash that releases surface-bound IgG. To assess the syndecan-1 HSPG itself, we used untransfected McArdle cells, which express this molecule endogenously (16). We used model remnant lipoproteins as ligands for syndecan-1, as described (11,16,29). Model remnant lipoproteins were prepared by adding lipoprotein lipase (LpL), a key protein on remnants recognized by HSPGs, to unlabeled or 125 I-labeled human LDL that we had methylated to an extent that blocks its binding to LDL receptors (mLDL), thereby mimicking apoB 48 (11,29).
Cellular extractions, immunoprecipitations, co-immunoprecipitations, immunoblots, and siRNA knockdowns were performed following our published protocols (16,28). Antibodies against target molecules (total target as well as forms with site-specific phosphorylations) are listed in supplemental Table  II. The nontarget siRNA pool, designed and microarray tested by the manufacturer for minimal targeting of rat genes, and the siRNA pool against rat cortactin were from Dharmacon (Lafayette, CO; catalog nos. D-001910-10-20 and L-080141-01-0010, respectively).
Statistical Analyses-Quantitative data from our experiments were analyzed using SigmaStat version 3 (SPSS Inc., Chicago). Normally distributed data are reported as the mean Ϯ S.E., n ϭ 3 per group per experiment. For comparisons between a single experimental group and a control, Student's unpaired two-tailed t test was used. For comparisons involving several groups simultaneously, analysis of variance (ANOVA) was initially used, followed by pairwise comparisons of each experimental group versus the control group by the Dunnett qЈ statistic.

RESULTS
Alanine Scanning Mutagenesis Identifies a Single Highly Conserved Juxtamembrane Motif, MKKK, in the Syndecan-1 Cytoplasmic Tail as Essential for Efficient Endocytosis after Clustering-We embarked on a comprehensive survey of the syndecan-1 cytoplasmic tail using alanine scanning mutagenesis. Our mutations, displayed in Fig. 1A, tampered with only 3-5 consecutive aminoacyl residues at a time. To avoid interference from the native unmutated syndecan-1 HSPG, particularly during clustering, the mutations were introduced into the FcR-Synd1 chimera, and hence we used nonimmune human IgG as the ligand. We covered the entire FIGURE 1. Alanine scanning mutagenesis identifies a single highly conserved juxtamembrane motif, MKKK, in the syndecan-1 cytoplasmic tail as essential for efficient endocytosis after clustering. A, native unmutated (UM) and mutated syndecan-1 sequences. Top sequence shows the final transmembrane (TM) aminoacyl residue and the entire cytoplasmic tail of syndecan-1. These regions form the C terminus of the UM FcR-Synd1 chimera. The first conserved (C1), variable (V), and second conserved (C2) domains of the syndecan-1 cytoplasmic tail are indicated. Lower sequences show the C-terminal aminoacyl residues of each of our alanine scanning mutants of the FcR-Synd1 chimera. Mutated residues are indicated in purple. We designated each mutant construct by the short sequence that was replaced by an equal number of alanines (e.g. MKKK34A and DEGSY35A). The UM and mutant constructs were expressed, one at a time, in McArdle 7777 hepatoma cells, followed by analysis of ligand catabolism. B and C, raft localization and internalization triggered by clustering. The ligand for FcR-Synd1, 125 I-labeled nonimmune human IgG, was bound at 4°C to the surface of the McArdle cell lines described in A. Unbound material was washed away, and then the cells were incubated for 1 h at 37°C in the absence or presence of our clustering agent (goat F(abЈ) 2 against human IgG Fab). Raft localization was assessed by cold Triton insolubility and internalization by resistance to an acid wash that releases surface-bound IgG. syndecan cytoplasmic tail within the FcR-Synd1 chimera, except for the first cytoplasmic residue, Arg (R), which we found to be required for export to the cell surface. The unmutated (UM) parent and mutant constructs were each expressed in McArdle 7777 hepatoma cells, followed by analysis of the catabolism of labeled IgG.
All alanine scanning mutants moved into rafts upon ligand clustering (Fig. 1B). Replacement of a highly conserved juxtamembrane motif, MKKK, with four alanines was the only mutation that blocked efficient endocytosis after clustering (Fig. 1C). As a control, Fig. 1D shows that the MKKK34A mutant is expressed well and traffics to the cell surface, at similar levels to the unmutated construct. Thus, despite prior reports that other syndecan cytoplasmic domains link to the cytoskeleton (14, 21-24), we found the juxtamembrane MKKK motif to be uniquely critical for this raft-and actin-dependent endocytic pathway.

MKKK Motif Mediates Basal Association with ␣-Tubulin, Rapid ERK Activation upon Ligand Binding, and Then ERK-dependent Dissociation from ␣-Tubulin, a Required
Step for Efficient Endocytosis-To determine how the MKKK motif mediates endocytosis, we examined its role in several previously reported syndecan functions, even though none of those func-tions had been known to affect ligand internalization. We began with ERK activation, because it can occur after ligand binding (30) and was reported to be an early event during syndecan-1mediated internalization (Ͻ10 min) (31). Previous studies did not indicate which site(s) on ERK became phosphorylated upon syndecan engagement. Here, we used specific antibodies that distinguish between phosphorylations at Thr-202 and Tyr-204 (Thr(P)-202 and Tyr(P)-204, abbreviated in Fig. 2 as pT202 and pY204) (28). Binding of nonimmune, monomeric IgG to FcR-Synd1, without clustering, triggered phosphorylation at Thr-202 ( Fig. 2A), whereas clustering was required for the formation of dually phosphorylated ERK (Fig. 2, A and B). Similarly, addition of methylated LDL (mLDL) to untransfected hepatocytes provoked Thr-202 phosphorylation of ERK (Fig. 2C), whereas the further addition of lipoprotein lipase, which converts mLDL into a multivalent ligand for syndecan-1, was required for the formation of dually phosphorylated Thr(P)-202/Tyr(P)-204-ERK (Fig. 2, C and D). In all cases, ERK phosphorylation was evident by 5 min after addition of ligand. Of note, replacement of the MKKK motif in FcR-Synd1 with four consecutive alanines (MKKK34A) completely abolished any ERK phosphorylation by either monomeric or multivalent ligands (Fig. 2, E and F). McArdle hepatocytes expressing the unmutated FcR-Synd1 chimera were preincubated at 4°C without or with unlabeled IgG as indicated, unbound material was washed away, and then the cells were incubated at 37°C for the indicated times, in the absence or presence of our clustering agent (ϩC). Displayed are immunoblots of cellular homogenates for phosphorylations of ERK at Thr-202 (pT202-ERK, A) and Tyr-204 (pY204-ERK, B). Immunoblots for total ERK (t-ERK, meaning phosphorylated plus unphosphorylated forms) are also shown for each sample. C and D, phosphorylations of ERK stimulated by ligands for syndecan-1. Untransfected McArdle hepatocytes, which express endogenous syndecan-1, were incubated at 37°C, without or with mLDL, without or with lipoprotein lipase (ϩLpL), for the indicated times. Displayed are immunoblots for phosphorylated ERK (C, pT202; D, pY204) and for total ERK. E and F, MKKK motif mediates ERK activation. McArdle hepatocytes expressing the FcR-Synd1 (MKKK34A) mutant were treated as in A. Displayed are immunoblots for phosphorylated (E, pT202; F, pY204) and total ERK. The data in this figure are representative of a total of three independent immunoblotting experiments.
A previous report indicated that an affinity column made from a Sepharose-linked peptide corresponding to the cytoplasmic tail of syndecan-3 (neural syndecan) was able to bind ␣-tubulin, Src family kinases, and cortactin from crude brain homogenates (32). Binding was attributed to the C1 domain of the syndecan-3 peptide (32). Here, to examine a possible association between ␣-tubulin and the syndecan-1 cytoplasmic tail in our cultured cells, we used co-immunoprecipitations. In the absence of bound ligand, both FcR-Synd1 and native syndecan-1 strongly associated with ␣-tubulin (Fig. 3, A and B, 0 min). Basal association with ␣-tubulin was entirely dependent on the MKKK motif, thereby pinpointing this function to those four amino acids within the C1 domain (Fig. 3C). Surprisingly, addition of ligand provoked the dissociation of FcR-Synd1 and syndecan-1 from ␣-tubulin within 30 -45 min (Fig. 3, A and B).
These results indicate three new functions for the MKKK motif of the syndecan cytoplasmic tail, namely Thr-202 phosphorylation of ERK (Fig. 2, A, C and E), regulated dissociation from ␣-tubulin (Fig. 3, A-C), and Tyr-204 phosphorylation of ERK (Fig. 2, B, D and F). The first two are triggered by ligand binding and the third by clustering. To determine whether these effects are causally connected, we used the ERK inhibitor U0126. This compound completely blocked the ability of any ligand, either monomeric or multivalent, to provoke the dissociation of FcR-Synd1 or syndecan-1 from ␣-tubulin (Fig. 3, D and E). In most of our experiments, ERK inhibition actually enhanced the association between the syndecan cytoplasmic tail and ␣-tubulin after addition of ligand (Fig. 3, D and E). To test a role for these processes in syndecan-mediated endocytosis, we found that ERK inhibition significantly interfered with internalization of clustered 125 I-labeled IgG by FcR-Synd1 and of LpL-enriched 125 I-labeled mLDL by syndecan-1 (Fig. 3, F and  G). Thus, activation of ERK is an essential step for efficient syndecan-mediated endocytosis of multivalent ligands, possibly through its role in freeing the MKKK domain from ␣-tubulin.

Upon Clustering, the MKKK Motif Gradually Triggers Src-dependent Tyrosine Phosphorylation of the Syndecan-1 Transmembrane and Cytoplasmic Domains, Which in Turn Causes the Robust Recruitment of Cortactin and Efficient Endocytosis-
We next focused on tyrosine phosphorylations. We previously reported an essential role for an unspecified tyrosine kinase activity in syndecan-mediated endocytosis (11). The syndecan-1 cytoplasmic tail contains four absolutely conserved tyrosyl residues (Fig. 1) (14,19,20), and as just noted, the C1 domain of syndecan-3 can bind Src family kinases (32). Here, we found that ligand binding, but only with clustering, provoked tyrosine phosphorylation of the syndecan transmembrane and cytoplasmic domains of the FcR-Synd1 chimera (Fig.  4A). Moreover, this effect was entirely abolished by replacement of the MKKK motif with four alanines (Fig. 4A). Interest-ingly, robust tyrosine phosphorylation of FcR-Synd1 and of syndecan-1 after clustering required 45-60 min (Fig. 4, B and C), i.e. it resembles the time course that we had reported for syndecan-1-mediated endocytosis (11,12). To characterize the enzymes responsible for tyrosine phosphorylation of FcR-Synd1 and syndecan-1 after clustering, we found that the process was blocked by two structurally distinct Src family kinase inhibitors, namely Src inhibitor-1 and PP2 (Fig. 4, D and E). Moreover, inhibition of Src family kinases impaired syndecan-1-mediated endocytosis (Fig. 4F).
We examined potential interactions with cortactin, which is a particularly attractive molecule in this circumstance, because it has been shown to bind filamentous actin and to participate in membrane-cytoskeletal interactions in several circumstances, such as clathrin-dependent and -independent endocytosis (33). Unlike the situation with ␣-tubulin, however, basal association of FcR-Synd1 and native syndecan-1 with cortactin was low or undetectable by co-immunoprecipitation (Fig. 5, A  and B, 0 min). Ligand binding, but only with clustering, stimulated the recruitment of cortactin to FcR-Synd1 and syndecan-1 (Fig. 5, A and B, lower immunoblots). Cortactin recruit- ment to FcR-Synd1 and syndecan-1 after clustering also required 45-60 min. Recruitment of cortactin after clustering was entirely dependent on the MKKK motif (Fig. 5C).
Thus, tyrosine phosphorylation of syndecan-1, cortactin recruitment, and endocytosis occur with essentially the same time course after clustering. To establish a causal chain, we constructed a new mutant, in which we replaced all four syndecan tyrosines in FcR-Synd1 with alanines. This construct, FcR-Synd1 (4Y34A), still underwent raft localization upon clustering (data not shown) and loss of tubulin upon ligand binding (Fig. 5D), but it failed to undergo tyrosine phosphorylation (supplemental Fig. I), recruitment of cortactin (Fig. 5E), or endocytosis after clustering (Fig. 5F). Thus, tyrosine phosphorylation, even in the continued presence of the MKKK motif, is required for robust cortactin recruitment and efficient endocytosis. In contrast, inhibition of ERK still allowed cortactin recruitment (supplemental Fig. II, A and B; compare with Fig. 3,  D-G).
To independently verify this sequence of events using the unmutated FcR-Synd1 chimera and the syndecan-1 HSPG, we examined each step in the endocytic pathway in the absence versus presence of the Src family kinase inhibitor PP2. Inhibition of Src family kinases had no effect on ERK phosphorylation (supplemental Fig. III, A-D), raft localization (data not shown), or loss of ␣-tubulin from FcR-Synd1 or syndecan-1 after addition of multivalent ligands (supplemental Fig. III, E and F; com-pare with Fig. 5D). Consistent with the ability of PP2 to block tyrosine phosphorylation of FcR-Synd1 and syndecan-1 after clustering (Fig. 4, D and E), Src family kinase inhibition also stopped the recruitment of cortactin, as assessed by co-immunoprecipitations (supplemental Fig. III, G and H; compare with Fig. 5E). As noted above and in Fig. 4F, inhibition of Src family kinases interfered with endocytosis of multivalent ligands, i.e. similar to the effect of the 4Y34A mutation (Fig. 5F).

Cortactin Is Required for Rapid Syndecan-mediated Activation of ERK upon Ligand Binding, ERK-dependent Dissociation from ␣-Tubulin, and Efficient Endocytosis of FcR-Synd1 and Syndecan-1 after Clustering, but Cortactin Is Not Needed for
Tyrosine Phosphorylation of Syndecan-1-To examine the role of cortactin, we used siRNA to knock down this protein in cultured hepatocytes (Fig. 6A), followed by examination of each step that we have found in the syndecan endocytic pathway. As expected from the low levels of association from 5-15 min (Fig.  5, A and B), cortactin knockdown did not affect raft localization of FcR-Synd1 after clustering (data not shown). Remarkably, we found that cortactin knockdown abolished the ability of FcR-Synd1 or syndecan-1 to trigger Thr-202 or Tyr-204 phosphorylation of ERK after ligand binding (Fig. 6, B-E). Consistent with this finding, cortactin knockdown also stopped the loss of ␣-tubulin from FcR-Synd1 and syndecan-1 after ligand binding (Fig. 6, F and G). Because ERK activation and ␣-tubulin loss are rapid effects, these results suggest an early interaction of the syndecan-1 cytoplasmic tail with cortactin that is not detected by co-immunoprecipitation or else an interaction of ERK with cortactin that is required for syndecan-1-mediated phosphorylations at Thr-202 and Tyr-204. Tyrosine phosphorylation after clustering of FcR-Synd1 and syndecan-1 still occurred after cortactin knockdown (Fig. 7, A and B). These data, in combination with the results in Fig. 5E and supplemental Fig. III, G and H, indicate that tyrosine phosphorylation causes cortactin recruitment, not the other way around. Finally, partial knockdown of cortactin inhibited efficient endocytosis of clustered 125 I-IgG by FcR-Synd1 and of LpL-enriched 125 I-mLDL by syndecan-1 (Fig. 7, C and D).

DISCUSSION
Our results provide the first detailed picture of sequential molecular events required for raft-dependent endocytosis. We examined the syndecan-1 HSPG after clustering induced by multivalent ligands. Its endocytosis can be divided into two phases, based on the kinase and the corresponding cytoskeletal partner that are involved. In the initial phase, ligands trigger rapid activation of ERK and localization of syndecan-1 into rafts (Fig. 8A). Syndecan-mediated activation of ERK depends on the MKKK motif and on the presence of cortactin. Raft localization requires clustering and cholesterol-rich microdomains, with no dependence on MKKK, ERK, Src family kinases, cortactin, or active cellular metabolism. Activation of ERK drives the dissociation of syndecan-1 from ␣-tubulin. Both ␣-tubulin and cortactin bind via the MKKK motif of syndecan-1, and loss of ␣-tubulin is required for efficient endocytosis, but persistent association with ␣-tubulin does not block cortactin recruitment. Instead, these results suggest that ␣-tubulin may act as an anchor for syndecan-1 at the plasma membrane in the basal state.
In the second phase, Src family kinases phosphorylate tyrosyl residues in the transmembrane and cytoplasmic region of syndecan-1, in a process that also requires the MKKK motif. We infer that these events are facilitated by the movement of clustered syndecan-1 into rafts (12), where Src family kinases are present (34). Rafts also provide an environment shielded from transmembrane tyrosine phosphatases (35). Tyrosine phosphorylation of syndecan-1 triggers the robust recruitment of cortactin, which we found to be an essential mediator of efficient actin-dependent endocytosis (Fig. 8B).
Several aspects of this endocytic pathway could allow for physiologic regulation. First, a rate-limiting step appears to be Src-mediated tyrosine phosphorylation of syndecan-1, which almost immediately causes robust cortactin recruitment and endocytosis (Figs. 4F and 5 and supplemental Fig. III, G and H). Thus, processes that enhance tyrosine phosphorylation or inhibit tyrosine phosphatases might accelerate syndecan-1mediated endocytosis. An attractive candidate is insulin, which is released in the postprandial state, activates the tyrosine kinase activity of the insulin receptor, activates ERK, and stimulates NOX4 to disable protein-tyrosine phosphatases (28,36,37). Other growth factors produce similar effects (38), and cytokines, additional enzymes, chaperones, and other ancillary molecules may also participate (5, 39 -42). Second, the need for clustering to trigger specific steps in syndecan-1-mediated endocytosis could allow the system to sense ligand size, valence, binding affinity, and potentially other characteristics. Consistent with this idea, our previous side-by-side comparison showed that a larger ligand, VLDL, underwent more rapid catabolism than did a smaller ligand, LpL-enriched mLDL (see page 1963 of Ref. 16). Moreover, a series of artificial ligands that recognize different sulfation epitopes on the heparan sulfate side chains of syndecans bound similarly to the cell surface, but only a strong ligand for 2-O-sulfated groups drove endocytosis, presumably by triggering clustering and membrane invagination (43). Third, we previously reported that regulated secretion of molecules, particularly sulfatase-2 (SULF2), that enzymatically modify or sterically hinder the binding domains on the heparan sulfate side chains of syndecan-1 can alter ligand catabolism in vitro and in vivo (16,20,44,45). Fourth, co-operation with other raft-associated receptors, such as SR-BI or clustered glypicans, could also affect ligand trafficking (20,43).
This detailed characterization of ligand handling by syndecan-1 also sheds light on disease. For example, syndecans, including syndecan-1, have been shown to facilitate infection of human macrophages with HIV-1 (46,47). Presumably, the virus acts as a multivalent ligand, driving syndecan into rafts, which other studies have shown to be the preferred site for fusion of the HIV-1 envelope with the target cell membrane, to allow infectious entry (48 -50). Syndecans appear ideally suited for hijacking by HIV-1, because the ϳ1-h delay between their rapid movement into rafts and subsequent endocytosis would allow time for envelope fusion and hence infection to occur, before  Fig. 6, B and C. Displayed are immunoprecipitations with anti-FcR antibodies (IP: FcR-Synd1), followed by immunoblots with antibodies against phosphotyrosine residues (IB: pY) or against the FcR ectodomain (IB : FcR-Synd1). B, tyrosine phosphorylation of syndecan-1 after clustering does not require cortactin. McArdle hepatocytes were treated as in Fig. 6, D and E. Displayed are immunoprecipitations with anti-syndecan-1 antibodies, followed by immunoblots for phosphotyrosines or syndecan-1. C, efficient endocytosis of FcR-Synd1 after clustering requires cortactin. McArdle hepatocytes expressing the unmutated FcR-Synd1 chimera were transfected with nontarget (black columns) or cortactin (tan columns) siRNAs. Partial knockdown of cortactin protein was verified in parallel wells (inset). After siRNA transfection, raft localization and internalization of 125 I-labeled IgG triggered by clustering were assessed after 1 h at 37°C, as in Fig. 1. Displayed are clustering-dependent raft localization and internalization of ligand, normalized to control values from cells expressing the unmutated chimera (mean Ϯ S.E., n ϭ 3; non-normalized control values were 510 Ϯ 18.16 and 482 Ϯ 13.6 ng/mg, respectively). D, efficient syndecan-1-mediated endocytosis after clustering requires cortactin. McArdle hepatocytes were transfected with nontarget (black columns) or cortactin (tan columns) siRNAs. After siRNA transfection, cells were incubated for 1 h at 37°C with 125 I-labeled mLDL, without or with LpL. Displayed are LpL-dependent surface (Surf) binding and internalization of ligand, normalized to control values from cells receiving nontarget siRNA (mean Ϯ S.E., n ϭ 3; non-normalized control values were 301.27 Ϯ 11.26 and 542.07 Ϯ 19.14 ng/mg, respectively). **, p Ͻ 0.01 (two-tailed Student's t test, C and D). The data are representative of a total of three independent IP/IB and ligand catabolism experiments. syndecans can deliver the viral ligand into lysosomes for hydrolysis and sterilization. We infer that the preferential binding of the HIV-1 envelope glycoprotein gp120 to 6-O-sulfated groups over 2-O-sulfated groups on syndecans (47) may further delay endocytosis (43), thereby favoring infectious entry via fusion with cellular rafts. Moreover, ligand binding to syndecan-1 activates ERK, and ERK activation has been shown to facilitate HIV-1 infectivity and the assembly and release of new virions (51,52).
We (11)(12)(13) and others (53) have demonstrated a crucial role for syndecan-1 in the safe hepatic disposal of atherogenic postprandial remnant lipoproteins, thereby resolving a long-standing controversy over the identity of the receptors responsible for this important pathway. Syndecan-1 is abundantly expressed on the sinusoidal surface of hepatic parenchymal cells (54,55), along microvilli facing the space of Disse (53), where remnant lipoproteins are cleared (11,13,20). Knowledge of the syndecan-1 endocytic pathway has already allowed us to identify specific molecular defects that impair its function in type 1 and type 2 diabetes mellitus and may thereby contribute to dyslipoproteinemia and increased cardiovascular risk in those conditions (16,20,45,56). Our current results indicate additional points of control (Fig. 8) and may facilitate future studies on how the healthy liver handles the lipid load from remnant lipoproteins that it internalizes via syndecan-1. In diabetic dyslipoproteinemia and other conditions, continuously high levels of ligands for syndecan-1, such as remnant lipoproteins or sulfatase-2, may cause aberrant activation or desensitization of ERK or Src family kinases, with downstream effects. Syndecans may also mediate uptake of cholesterol-rich aggregated lipoproteins by macrophages within the arterial wall, leading to the formation of foam cells, a hallmark of atherosclerosis (57).
Differences among raft-dependent receptors emphasize the need for precise molecular characterizations of each. As noted above, SV40 enters via rafts for delivery to the endoplasmic reticulum, whereas syndecan-1 enters via rafts and carries its ligands to lysosomes for destruction (11,12). Although Src was one of six major kinases identified in SV40 endocytosis (58), the generalizability of findings from that system to any other raftdependent receptor will need to be tested. Of note, the MKKK motif appears, with one conservative substitution, in the juxtamembrane region of the Drosophila syndecan cytoplasmic tail (MRKK) (18). The same single conservative substitution is present in human syndecan-2, which also mediates endocytosis under some circumstances (11), suggesting considerable biologic importance of this novel motif.
Overall, our findings provide new insights into raft-mediated endocytosis by a highly conserved receptor, syndecan-1, and they will facilitate additional studies into its roles in ligand binding, downstream signaling, endocytosis, and other trafficking in normal and diseased states.