Syntenin-1 Is a New Component of Tetraspanin-Enriched Microdomains: Mechanisms and Consequences of the Interaction of Syntenin-1 with CD63^▿

Cell lines and antibodies.

Cos7, 293T, and HeLa cells were grown in Dulbecco's modified Eagle medium (DMEM; Sigma) supplemented with 10% fetal calf serum. CHO-K1 cells were grown in Ham's F-12 media supplemented with 10% fetal calf serum. The mouse anti-CD81 monoclonal antibody (MAb) (M38) was kindly provided by O. Yoshie. The anti-CD63 (6H1) and anti-CD151 (5C11) MAbs were described previously (4, 5). The anti-CD63 MAbs (1B5) were kindly provided by M. Marsh (University College London, London, United Kingdom). The mouse MAbs to human lamp-1 and lamp-2 were from the Development Studies Hybridoma Bank. The anti-rabbit polyclonal Abs to CD151 were provided by L. Ashman (University of Newcastle, Australia). The monoclonal and polyclonal anti-hemagglutinin (HA) tag Abs (F7 and Y-11) were purchased from Autogen Bioclear. The monoclonal Abs against myc tag sequence (9E10) and green fluorescent protein (3E1) were purchased from the Cancer Research UK antibody production facility. The polyclonal and monoclonal anti-myc tag Abs were from Signaling Technology. Rabbit polyclonal Abs against recombinant glutathione S-transferase (GST)-tagged human syntenin-1 were produced by Eurogentec (Liege, Belgium) and affinity purified using coupled recombinant protein.

Plasmids and transfection.

The cDNAs encoding human syntenin-1 (TACIP-18) and the HA-tagged form of syntenin-2 (TACIP) in pcDNA3.1/Zeo(+) were provided by J. Arribas (Vall d'Hebron Research Institute University Hospital, Barcelona, Spain). The cDNA encoding human CD63 in pZeoSV was described previously (5). The plasmid encoding the GST-tagged form of human syntenin-1 was obtained from T. Cierpicki and Z. Derewenda. The plasmid encoding the Flag-tagged NF-2 was kindly provided by M. Crompton (Royal Holloway University of London, United Kingdom). DNA constructs encoding various mutants of syntenin-1 and CD63 were produced by a standard PCR (sequences of the primers are available upon request). Transfection experiments were carried out using Fugene 6 (Roche) according to the manufacturer's instructions. For transfection with small interfering RNA (siRNA), HeLa cells were detached and electroporated with a single pulse at 250 V and 250 μF using a GenePulser Xcell electroporation system (Bio-Rad) and 2.5 μM of siRNA duplex. The efficacy of knocking down was assessed 48 h later by Western blotting as described below. The following siRNA was found to be effective in the knockdown experiments: 5′-GCUAUAGCAUAGCUGCUUATT-3′. Silencer negative control no. 1 siRNA was purchased from Ambion (catalog no. 4611).

Immunoprecipitation and Western blotting.

The proteins were solubilized into the immunoprecipitation buffer containing 0.8% Brij 98-0.2% Triton X-100-phosphate-buffered saline (PBS) (or 0.5% Brij 98-0.5% Triton X-100-PBS), 2 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, and 10 μg/ml leupeptin for 4 to 16 h at 4°C. The insoluble material was pelleted at 12,000 rpm for 10 min. The cell lysates were then precleared by incubation for 2 h at 4°C with agarose beads conjugated with goat anti-mouse antibodies (mouse immunoglobulin G [mIgG] beads; Sigma). Immune complexes were collected using appropriate MAbs prebound to the mIgG beads and washed four times with the immunoprecipitation buffer. The anti-green fluorescent protein MAb (3E1) was used as a negative control in all immunoprecipitation experiments. The complexes were eluted from the beads with Laemmli sample buffer. Proteins were resolved in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to the nitrocellulose membrane, and developed with the appropriate Ab. Protein bands were visualized using horseradish peroxidase-conjugated secondary antibodies (Sigma) and enhanced chemiluminescence reagent (Amersham Pharmacia Biochem).

Immunoprecipitation of the cell surface pool of CD63.

CHO cells were transfected with the plasmids encoding the HA-tagged form of syntenin-1 and human CD63. Forty-eight hours after the transfection, the cells were washed with ice-cold DMEM and incubated for 1 h at 4°C with purified anti-CD63 MAb (6H1) or with the control MAb. Subsequently, the cells were washed three times with ice-cold DMEM and lysed in the immunoprecipitation buffer containing 0.8% Brij 98-0.2% Triton X-100-PBS and the protease and phosphatase inhibitors. Immunoprecipitation and analysis of the CD63-containing complexes were carried out as described above.

Biotinylated peptides, and pull-down, and enzyme-linked immunosorbent assays (ELISA).

N-terminally biotinylated peptides corresponding to the C-termini of tetraspanins were synthesized by AltaBioscience (Birmingham, United Kingdom). The following tetraspanin peptides were used: CD63, VKSIRSGYEVM; CD151, YRSLKLEHY; SAS, RNQKDPRANPSAFL; CD9, AIRRNREMV; and CD82, VHSEDYSKVPKY. For the pull-down experiments, 293T cells, ectopically expressing HA-tagged syntenins, were lysed overnight in the immunoprecipitation buffer (0.5% Brij 98-0.5% Triton X-100-PBS) containing protease and phosphatase inhibitors. Biotinylated peptides were conjugated to avidin-agarose (Sigma) for 2 h at 4°C in the washing buffer (0.5% Brij 98-0.5% Triton X-100-PBS) and subsequently incubated with the cell lysates for 5 to 16 h at 4°C. After three washes, pulled-down proteins were eluted from agarose into 1× Laemmli buffer, boiled, and resolved in SDS-PAGE. The proteins were then transferred to the nitrocellulose membrane and probed with the anti-HA tag rabbit polyclonal antibodies. The GST-tagged recombinant syntenin-1 proteins for ELISA were produced in the BL21 strain of Escherichia coli by using a standard protocol. The proteins (0.5 μg) were immobilized on 96-well plates overnight in 50 μl of coating buffer (0.1 M Na₂HPO₄, adjusted to pH 9 with 0.1 M NaH₂PO₄) at 4°C. Plates were then washed gently twice in PBS-Tween (0.05%). The wells were then incubated in blocking buffer (2% bovine serum albumin in PBS-Tween [0.05%]) for 1 h at 37°C and subsequently washed once with PBS-Tween (0.05%). Biotinylated peptides at a range of concentrations (made up in 50 μl of blocking buffer) were added to the wells and incubated for a minimum of 2 h at 37°C. After the incubation period, the plates were washed four times in PBS-Tween (0.05%), followed by addition of streptavidin-horseradish peroxidase (1:2,000 in blocking buffer) and subsequent incubation for 1 h at room temperature. Wells were washed eight times in PBS-Tween (0.05%) and 100 μl of 3,3′,5,5′ tetramethylbenzidine (Tebu-bio Laboratories) was added. The reaction was allowed to proceed for 20 min and then stopped by the addition of 100 μl 1 M HCl. Levels of binding were analyzed by measuring absorbance at A₄₅₀ using a plate reader.

NMR spectroscopy and C-terminal peptide ligand titration.

For the heteronuclear magnetic resonance spectroscopy (NMR) analysis of syntenin-1 and CD63 interaction, uniformly ¹⁵N-labeled syntenin-1 PDZ tandem domain (113 to 273), referred to as “syntenin-1 PDZ12,” was expressed as a GST fusion in Escherichia coli BL21 (DE3) at 25°C in M9 minimal medium using ¹⁵NH₄Cl as the sole nitrogen source. The purification of GST-fused protein was performed as reported previously (24). The NMR samples contained 200 μM protein, 150 mM NaCl, 500 μM TCEP [Tris(2-carboxyethyl)phosphine], and 50 μM AEBSF in 50 mM Tris buffer (pH 7.5). All NMR spectra were recorded at 30°C on a Varian Inova 800-MHz spectrometer equipped with a room temperature 5-mm ¹H/¹³C/¹⁵N z-axis pulse field gradient probe, with data processed using the Azara package and analyzed using Ansig (28). Earlier chemical shift assignments (9) were confirmed by tracing sequential nuclear Overhauser effect connectivities across the protein backbone from a three-dimensional heteronuclear single quantum correlation (HSQC)-nuclear Overhauser effect spectroscopy spectrum recorded with a mixing time of 150 ms using a 400 μM sample of ¹⁵N-labeled syntenin-1 PDZ12 protein. Synthesized C-terminal CD63 peptide (RVKSIRSGYEVM) was purchased from Sigma-Genosys and contained an extra Arg residue at the N terminus, to improve peptide solubility. Titrations of syntenin-1 PDZ12 proteins with C-terminal CD63 peptide were conducted by recording a series of ¹H,¹⁵N HSQC spectra of ¹⁵N-labeled protein (200 μM), with increasing molar concentrations of the peptide ligand, up to a protein-to-peptide ratio of 1:8. Some amide proton resonances were obscured at pH 7.5 due to rapid exchange with solvent, and hence the peptide titration was also performed at pH 6.5, with chemical shifts being assigned to the nearest neighbors. A nonlinear regression fit of the chemical changes as a function of ligand concentration was used to calculate the dissociation constant for the syntenin-1 PDZ12-CD63C complex (2). The combined backbone ¹H and ¹⁵N chemical shift changes, Δδ, were calculated according to equation 1, where ΔN is the ¹⁵N chemical shift difference in Hz, and ΔH is the ¹H chemical shift difference in Hz, between free and peptide (eightfold excess) titrated protein:

(1)

Immunofluorescence staining.

Cells were grown on glass coverslips in complete media for 24 to 36 h. Spread cells were fixed with 2% paraformaldehyde-PBS for 10 to 15 min. The staining with primary and fluorochrome-conjugated secondary Abs was carried out as previously described (6). The staining was analyzed using a Nikon Eclipse E600 microscope. Images were acquired using a Leica DC200 digital camera and subsequently processed using the DC200 image processing program.

Antibody uptake/pulse-chase.

HeLa cells grown on the glass coverslips were transfected with the plasmids encoding various HA-tagged syntenin-1 constructs. After 48 h posttransfection, cells were incubated with 10 μg/ml MAb 6H1 in serum-free DMEM (supplemented with 10 mM HEPES) for 1 h at 4°C. Cells were rinsed three times with ice-cold DMEM (zero time point) and then transferred to prewarmed complete growth media at 37°C for different lengths of time (30 min to 4 h). At each time point, a set of coverslips was taken out of the incubator and cells were immediately fixed for 20 min with 2% paraformaldehyde at room temperature. Cells were subsequently permeabilized in 0.1% Triton X-100. Cells were then rinsed in PBS and blocked for 1 h with blocking buffer (20% heat-inactivated normal goat serum-2% human serum-PBS). After blocking, cells were incubated with Alexa-conjugated goat anti-mouse Ab (1:1,000 in blocking buffer) for 1 h at room temperature. Finally, cells were rinsed in PBS and water. Coverslips were subsequently processed for analysis as described above. The internalization was scored as blocked/impaired when MAb clearly outlined cell perimeter. At least three independent experiments were performed for an individual construct with at least 70 to 80 syntenin-1-positive cells scored in each of the experiments.

Syntenin-1 is associated with tetraspanin-enriched microdomains: interaction of syntenin-1 with CD63.

It has previously been shown that tetraspanins are abundant on exosomes, 50- to 90-nm vesicles secreted by various cell types (12, 33, 39). We hypothesized that some of the cytoplasmic proteins that are recruited to exosomes can interact with tetraspanins directly. Indeed, we found that syntenin-1 can be coimmunoprecipitated with various tetraspanins (e.g., CD9, CD63, and CD81) from the lysates of MDA-MB-231 cells (Fig. 1A). The association with syntenin-1 was specific, since tetraspanins did not interact with annexin II or fatty acid synthase, two proteins that are also abundant on exosomes (results are not shown). To establish whether the interactions between the tetraspanins and syntenin-1 are direct, we initially performed immunoprecipitation experiments under conditions that disrupt the association of tetraspanins with most of their transmembrane partners (i.e., in the presence of 0.5% Brij 98-0.5% Triton X-100). In these experiments, only CD63 remained associated with syntenin-1 (Fig. 1B). These data suggest that CD63 directly interacts with syntenin-1, while the association of syntenin-1 with CD9 and CD81 may be mediated by other proteins.

Syntenin-1 is known to interact with various transmembrane proteins via their C-terminal cytoplasmic domains (36). A series of experiments were carried out to examine whether the C-terminal cytoplasmic portion of CD63 is involved in the interaction with syntenin-1. Firstly, we found that the immobilized CD63 C-terminal peptide (CD63-C), but not the corresponding regions of some other tetraspanins (e.g., CD151, CD9, SAS, and CD82), can retain syntenin-1 from cell lysates (Fig. 2A). The specificity of this interaction was further confirmed by the fact that CD63-C did not pull down syntenin-2, a closely related PDZ domain-containing protein (Fig. 2A). Secondly, the CD63-C peptide directly binds to recombinant GST-syntenin-1 in ELISA experiments (Fig. 2B). In a series of control ELISA experiments, we found that the CD63-C peptide did not bind to recombinant GST, nor did the CD9-C or CD151-C peptides interact with GST-syntenin-1 (Fig. 2B). Finally, we found that the deletion of the last two amino acids in CD63 completely abolished the association of the full-length protein with syntenin-1 (Fig. 2C). Thus, we concluded that the interaction between CD63 and syntenin-1 is direct and mediated by the C-terminal cytoplasmic tail of the tetraspanin.

The contribution of PDZ domains in the interaction of syntenin-1 with CD63.

Previous studies have shown that the presence of both PDZ domains is critical for the interaction of syntenin-1 with its various transmembrane partners (36). Indeed, the Gly¹²⁶→Asp and Gly²¹⁰→Glu mutations in the interacting loops of PDZ1 and PDZ2, respectively, completely abolished these interactions (27). The interaction of the mutants with CD63-C peptide was examined by ELISA and in coimmunoprecipitation experiments with the native CD63. As shown in Fig. 3A, the mutations in both PDZ domains markedly diminished interaction of CD63-C peptide with syntenin-1. Interestingly, the mutation in PDZ1 had a more dramatic effect on this interaction. These data were in agreement with the results of the coimmunoprecipitation experiments, in which we found that mutations in either of the PDZ domains dramatically decreased the amount of syntenin-1 associated with native CD63 (Fig. 3B).

These experiments suggested either that both PDZ domains can bind CD63-C (although with a preference for PDZ1) or that the tetraspanin tail is bound by one PDZ domain, with mutation of the other having an allosteric effect on the interaction. To examine the interaction in more detail, we carried out NMR titration analysis of ¹⁵N-labeled syntenin-1 PDZ12 with unlabeled CD63 C-terminal peptide. The most significant chemical shift changes were induced in the PDZ1 domain upon CD63-C binding (Fig. 4A), with only minimal changes occurring in the PDZ2 domain. Even when the peptide was added to an eightfold excess, it was insufficient to saturate the binding of CD63-C to syntenin-1 PDZ12, indicating a weak affinity for the peptide. The chemical shift perturbations were used to map the binding sites, revealing a contact surface predominantly in the PDZ1 domain binding pocket (Fig. 4B). Chemical shifts of residues Gly¹²⁶ and Ile¹²⁵ in the carboxylate binding loop of PDZ1 domain were perturbed even with stoichiometric amounts of CD63-C peptide (Fig. 4A, insets). Chemical shift perturbations were also observed for residues Arg¹¹⁹, Asp¹²⁰, Val¹⁴¹, Ala¹⁴³, Arg¹⁷⁶, Leu¹²⁷, Arg¹²⁸, and Arg¹⁷⁹ in the vicinity of the carboxylate binding loop of PDZ1 (Fig. 4B). In contrast, only minor chemical shift changes were found in the PDZ2 domain, being limited to residues Gly²¹⁰ and His²⁰⁸ in the carboxylate binding loop (Fig. 4B). Data from the NMR titration experiment were also used to estimate the dissociation constants for the binding of CD63-C to PDZ domains of syntenin-1. As inferred from the HSQC spectra, CD63-C peptide binds to PDZ1 domain with a K_d of 3.9 ± 1.2 mM (results not shown). In contrast, the affinity of the PDZ2 domain was too weak to measure precisely. Thus, the NMR results suggest that CD63 can bind both PDZ domains of syntenin-1 with clear preference for PDZ1, thus corroborating our results from ELISA and coimmunoprecipitation (Fig. 3A).

CD63 is associated with syntenin-1 on the plasma membrane: the role of the C terminus of syntenin-1.

Syntenin-1 is found in various membranous compartments in cells, including early and recycling endosomes, and also on the plasma membrane (14, 45). Double immunofluorescence experiments showed that in permeabilized HeLa cells, ectopically expressed syntenin-1 is associated with vesicular organelles, some of which were also positive for CD63 (Fig. 5A, upper panels). Although CD63 is expressed on the surface of HeLa cells (as detected by flow cytometry), no surface staining was observed in these experiments. However, when we preincubated cells with the anti-CD63 MAb at 4°C prior to fixation and permeabilization, both CD63 and syntenin-1 were readily detectable and colocalized at the plasma membrane (Fig. 5A, lower panels). This apparent “discrepancy” could be explained by a relatively high sensitivity of the surface pools of CD63 and syntenin-1 to detergent extraction after fixation with paraformaldehyde.

Next, we examined whether CD63 and syntenin-1 are associated with each other at the cell surface. As illustrated in Fig. 5B (upper panel), only a minor proportion (<10%) of the total amount of CD63 can be immunoprecipitated by anti-CD63 MAb prebound to the cells. In spite of this, ∼60% of the total pool of cellular syntenin-1 is associated with the surface-exposed CD63 (Fig. 5B, lower panel). These data suggest that the CD63-syntenin-1 complex is abundant at the cell surface.

It has previously been reported that deletion of the first 101 amino acids shifts equilibrium of the cellular distribution of syntenin-1 towards the plasma membrane (44). Hence, one might expect that deletion of the sequence N-terminal to PDZ1 would enhance the interaction between syntenin-1 and CD63. However, the immunoprecipitation experiments showed that the enhanced recruitment of the syntenin-1ΔN mutant (with 100 N-terminal amino acids deleted) to the plasma membrane did not result in the stoichiometric increase of its association with CD63 (Fig. 6B). On the other hand, we observed that the deletion of the last 23 amino acids following PDZ2 (syntenin-1ΔC mutant) obliterated its association with CD63 (Fig. 6B). Further analysis revealed that the region between the residues 281 and 293 is important for the stability of the CD63-syntenin-1 complex (Fig. 6C). Notably, the pattern of distribution of the syntenin-1ΔC mutant in cells was markedly different compared to that of syntenin-1 or syntenin-1ΔN, with less of the protein associated with the internal and plasma membranes (Fig. 6A). To examine whether the negative effect of the ΔC deletion on this association is due to the redistribution of the syntenin-1ΔC mutant in cells, we carried out pull-down experiments using CD63-C peptide. As illustrated in Fig. 6D, the syntenin-1ΔC mutant could not be retained by the immobilized peptide. Finally, ELISA experiments also showed that the direct binding of CD63-C peptide to purified syntenin-1ΔC is significantly weaker than that of the wild-type protein (Fig. 6E). These data strongly suggest that the C terminus is directly involved in stabilizing the association of syntenin-1 with CD63.

Overexpression of merlin does not affect interaction of syntenin-1 with CD63.

Previous studies have suggested that NF-2/merlin/schwannomin can also interact with syntenin-1 via the PDZ1 domain (21, 24). Thus, it was conceivable that elevated expression of merlin may substitute for CD63 in the complex with syntenin-1. A possible regulatory role of merlin was examined in transient transfection experiments. Figure 7 shows that increased amounts of merlin in Cos7 cells had no effect on the association of syntenin-1 with CD63. These results indicated that in cells the CD63-syntenin-1 and NF-2-syntenin-1 complexes coexist independently of one another.

Overexpression of syntenin-1 decreases the rate of internalization of CD63.

It is thought that CD63 undergoes rapid constitutive internalization from the plasma membrane by mechanisms involving the AP-2 complex (22). The μ2 chain of AP-2 directly interacts with the tyrosine-based sorting motif in the C-terminal cytoplasmic domain of CD63 (G-Y-E-V-M) (34) and directs the protein to clathrin-dependent endocytosis (22). We hypothesized that syntenin-1, which binds to the same region of CD63, may interfere with the assembly of the CD63-AP-2 complex and thereby inhibit endocytosis of the protein from the plasma membrane. Indeed, the rate of MAb-induced internalization of CD63 was reduced in cells expressing high levels of the exogenously introduced wild-type syntenin-1 (Fig. 8A and D). Specifically, 30 min after the temperature shift, the intracellular staining of the anti-CD63 MAb was observed in only ∼16% of the syntenin-1-expressing cells compared to 100% in the nontransfected cells. As shown in Fig. 8, the differences were also apparent at the later time points: ∼37% and ∼60% of the anti-CD63 MAbs were internalized after cells were incubated at 37°C for one and four hours, respectively. The inhibitory effect of syntenin-1 on CD63 expression was even more pronounced in cells expressing the N-terminal deletion mutant of this protein. We did not observe any internalization of the anti-CD63 MAb even after 4 h of incubation at 37°C (Fig. 8B and D). In control experiments, we found that neither wild-type syntenin-1 nor the syntenin-1ΔN mutant changed the kinetics of internalization of labeled transferrin (Fig. 8C). Taken together, these results show that syntenin-1 plays an important role in the regulation of constitutive endocytosis of CD63.