Transmembrane and Trans-subunit Regulation of Ectodomain Shedding of Platelet Glycoprotein Ibα*

Ectodomain shedding of transmembrane proteins may be regulated by their cytoplasmic domains. To date, the effecting cytoplasmic domain and the shed extracellular domain have been in the same polypeptide. In this study, shedding of GPIbα, the ligand-binding subunit of the platelet GPIb-IX complex and a marker for platelet senescence and storage lesion, was assessed in Chinese hamster ovary cells with/without functional GPIbα sheddase ADAM17. Mutagenesis of the GPIb-IX complex, which contains GPIbα, GPIbβ, and GPIX subunits, revealed that the intracellular membrane-proximal calmodulin-binding region of GPIbβ is critical for ADAM17-dependent shedding of GPIbα induced by the calmodulin inhibitor, W7. Perturbing the interaction between GPIbα and GPIbβ subunits further lessened the restraint of GPIbβ on GPIbα shedding. However, contrary to the widely accepted model of calmodulin regulation of ectodomain shedding, the R152E/L153E mutation in the GPIbβ cytoplasmic domain disrupted calmodulin binding to GPIbβ but had little effect on GPIbα shedding. Analysis of induction of GPIbα shedding by membrane-permeable GPIbβ-derived peptides implicated the association of GPIbβ with an unidentified intracellular protein in mediating regulation of GPIbα shedding. Overall, these results provide evidence for a novel trans-subunit mechanism for regulating ectodomain shedding.

A large number of transmembrane proteins expressed on the cell surface undergo ectodomain shedding, a process in which the protein is cleaved at a site close to the outer surface of the cell membrane and subsequently its ectodomain released from the cell (1). Because ectodomain shedding plays a vital role in various cellular processes, including cell growth and proliferation, mammalian development, cell migration, and inflammation, its aberration often leads to disease states including cancer (2)(3)(4)(5). Ectodomain shedding is in essence an enzyme-driven proteolytic reaction; therefore, in general there are two ways to regulate it. One way is to modulate the activity of the enzyme that is often referred to as sheddase; the other is to modulate the accessibility of the shedding substrate to the sheddase. Most sheddases belong to the ADAM (a disintegrin and metalloprotease) 2 protease family within the metzincin superfamily (1). Although the structure, function, and regulation of ADAMs have been extensively studied (6 -8), mechanisms modulating the accessibility of shedding substrates have not been well explored.
In addition to extracellular signals such as ligand binding to the shedding substrate (9,10), intracellular signals can also induce shedding of a protein without affecting the catalytic activity of the sheddase. The first reported example of this mechanism is the induction of shedding of L-selectin by calmodulin (CaM) inhibitors (11). CaM co-immunoprecipitates with L-selectin from cell lysates and binds directly to the isolated cytoplasmic domain of L-selectin. Treating L-selectin-expressing cells with trifluoperazine, a CaM inhibitor, causes dissociation of CaM from L-selectin and up-regulates shedding of L-selectin (11,12). CaM inhibitors rapidly trigger shedding of other membrane proteins by a mechanism independent of protein kinase C and calcium influx, both of which can activate ADAMs (13). Furthermore, point mutations in the L-selectin cytoplasmic domain, such as L320E, disrupt CaM association and significantly increase L-selectin shedding (11,12), providing additional evidence for regulation at the level of substrate. Since the report on L-selectin, treatment of CaM inhibitors, such as trifluoperazine and W7 (N-(6-aminohexyl)-5-chloro-1naphthalenesulfonamide), has been found to induce shedding of a number of membrane receptors (13)(14)(15)(16)(17)(18)(19)(20)(21). These membrane receptors contain a cluster of basic residues in the membraneproximal region of their cytoplasmic domain, many of which have been shown to associate with CaM. Thus, it appears that CaM association with the cytoplasmic domain of a membrane protein helps to inhibit its shedding. Dissociation of CaM, either by exogenous inhibitors or disrupting mutations in the CaM-binding site, would up-regulate shedding.
The glycoprotein (GP) Ib-IX complex is a platelet receptor for von Willebrand factor and other ligands that play important roles in hemostasis, thrombosis, and inflammation (22). Under physiological conditions ADAM17 cleaves GPIb␣, the main ligand-binding subunit of the GPIb-IX complex (Fig. 1A), at a site between membrane-proximal residues Gly 464 and Val 465 (23,24). The soluble shed extracellular domain of GPIb␣ is a potential biomarker for platelet storage lesion and senescence (23,25,26). Although shedding of GPIb␣ can be induced by CaM inhibitors (24), GPIb␣ does not conform to previous observations in that its cytoplasmic domain does not contain a CaM-binding sequence. Instead, the cytoplasmic domain of GPIb␤, another subunit in the GPIb-IX complex that is not shed, contains a membrane-proximal positively charged sequence that interacts with CaM (27). In the GPIb-IX complex GPIb␣ is associated with two GPIb␤ subunits via extracellular membrane-proximal disulfide bonds and noncovalent transmembrane domain interactions (28,29), we investigated in this study whether CaM association with the cytoplasmic domain of GPIb␤ regulates ectodomain shedding of GPIb␣ in a trans-subunit manner.

EXPERIMENTAL PROCEDURES
Materials-The vector pDX was used in earlier studies on expression of the GPIb-IX complex in Chinese hamster ovary (CHO) cells (30). The CHO K1 cell line was obtained from ATCC (Manassas, VA). The CHO M2 cell line that lacks functional ADAM17 has been described (31). CaM inhibitor W7 and the broadspectrum hydroxamic acid-based metalloprotease inhibitor GM6001 were purchased from Calbiochem (La Jolla, CA) and dissolved in 5% dimethyl sulfoxide (DMSO) before use.
To obtain the Ib␣ IX construct, in which the GPIb␣ cytoplasmic domain was replaced with the GPIX counterpart, a DNA sequence encoding GPIX residues Ala 155 -Asp 161 and a stop codon was appended to that encoding the GPIb␣ transmembrane domain by consecutive PCR. The resulting gene fragment was then ligated as a BspEI/XbaI fragment into the pDX-Ib␣ vector as described (33). Glu-scanning mutagenesis in the GPIb␤ cytoplasmic domain was carried out using QuikChange (Stratagene, La Jolla, CA). The mutant gene fragment was inserted as a NotI/XmaI fragment into the pDX-HA-Ib␤ vector (34)  gen) as described previously (33,34). To ensure accurate comparison of GPIb␣ surface expression levels among various transfected cells, key parameters including the amount of vector DNA and transfection efficiency were kept constant (33). CHO cells transfected with sham vectors (for 0% expression level) and with the wild-type GPIb-IX complex (for 100% expression level) were always included in each transfection experiment. At 48 h after transfection, the cells were used for shedding assays as described below. The CHO cell line stably expressing the wild-type GPIb-IX complex was obtained as described (32). Flow Cytometry-The surface expression level of GPIb␣ was analyzed by flow cytometry (32,33). Briefly, CHO cells resuspended in phosphate-buffered saline (PBS) were incubated with 2 g/ml of anti-GPIb␣ monoclonal antibody WM23 for 1 h at room temperature, washed twice with PBS containing 5% bovine serum albumin (PBS/BSA), and stained with fluorescein isothiocyanate-conjugated rabbit anti-mouse IgG (Invitrogen) for 1 h at room temperature. After a final wash with PBS/BSA to remove unbound antibody, the cells were fixed and examined in a Beckman-Coulter Epics XL flow cytometer. To quantitate the GPIb␣ surface expression level, the mean fluorescence intensity of the entire cell population (10,000 cells) was normalized to the value of cells expressing the wild-type complex set as 100%.
Ectodomain Shedding of GPIb␣ from Transfected CHO Cells-To measure W7-induced shedding of GPIb␣, transfected CHO cells (approximately 5 ϫ 10 5 cells) were washed with PBS containing Ca 2ϩ and Mg 2ϩ and treated with either 150 M W7 or 5% DMSO at 37°C for 3 h. Proteins contained in the PBS supernatant were precipitated with 5% trichloroacetic acid on ice for 1 h. Cells were either assayed for surface expression of GPIb␣ by flow cytometry, or lysed in lysis buffer (50 mM Tris⅐HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 5 mM N-ethylmaleimide, 1% Triton X-100) containing a 1:50 dilution of a protease inhibitor mixture for mammalian cells (Sigma). The supernatant from the cell lysate was obtained by centrifugation at 15,600 ϫ g for 5 min at room temperature. To measure constitutive shedding of GPIb␣ from transiently transfected cells, the culture medium was collected 2 days after transfection, and glycocalicin (GC) released into the medium was immunoprecipitated with WM23 and protein G-agarose beads (Invitrogen).
In both experiments, glycocalicin precipitated from the medium or full-length GPIb␣ in cell lysates were resolved in an 8% Tris glycine/SDS gel under reducing or non-reducing conditions, transferred to PVDF membranes, and blotted with WM23. As a loading control, blots were stained with anti-actin antibody (Sigma). To assess expression of HA-tagged GPIb␤   and GPIX, cell lysates were resolved in 12% Tris glycine/SDS gels, transferred to PVDF membranes, and blotted with anti-HA antibody (Sigma) and polyclonal anti-GPIX antibody (Santa Cruz Biotechnology, Santa Cruz, CA), respectively. Antibody binding was detected using peroxidase-conjugated secondary antibodies and chemiluminescence substrates (PerkinElmer Life Sciences). Finally, the membrane was exposed to Kodak BioMax X-AR film or directly visualized and quantitated using a GeneGnome HR imaging system (Syngene BioImaging, Frederick, MD). The exposure time was adjusted to avoid saturation and to ensure exposure in the linear range. The relative intensity of the blotted bands was quantitated with background correction, and normalized against the intensity of the corresponding actin band.
The GC:GPIb␣ ratio for the wild-type and mutant GPIb-IX complex, an indicator of the extent of GPIb␣ shedding, is calculated as the ratio of the level of GC measured by Western blot to the surface expression level of GPIb␣ quantitated by flow cytometry. The ratios were then normalized to that of the wildtype complex in the same experiment set as 1.0.
CaM Pulldown Assay-Cell lysates were incubated with 20 g of GST-CaM (27) at 4°C for 1 h, and then incubated with 50 l of glutathione-Sepharose 4B beads (GE Healthcare) at room temperature for 1 h. Beads were washed with Tris-buffered saline and bound HA-tagged GPIb␤ was eluted with SDS sample buffer, resolved using a 5-20% Tris glycine/ SDS gel under reducing conditions, transferred to PVDF membrane, and blotted with anti-HA antibody. Palmitoylated Peptide-induced Shedding-N-terminal palmitoylated peptides, palmitoyl-RLRRL-RARAR (named pal-WT), palmitoyl-RLEELRARAR (pal-151E/152E), and palmitoyl-RLREERARAR (pal-152E/153E), were synthesized on Fmoc-Ala-Wang resin using standard Fmoc/tBu chemistry and the HBTU/HOBt protocol (36). Peptides were purified by reverse-phase HPLC and confirmed by mass spectrometry. CHO cells stably expressing the GPIb-IX complex (32) were washed with PBS containing Ca 2ϩ and Mg 2ϩ , incubated in the same buffer with each peptide at the indicated concentrations or 5% DMSO as solvent control at 37°C for 2 h. The extent of GPIb␣ shedding induced by the peptide treatment, including GC release into the PBS buffer and decrease of the GPIb␣ surface expression level, was measured as described above.

RESULTS
The human GPIb-IX complex is primarily expressed in anucleate platelets that are not amenable to mutagenesis studies. CHO cells, which do not express endogenous GPIb-IX, have been used in studies of ectodomain shedding (37,38). As in platelets, GPIb␣ is found in the plasma membrane of transfected CHO K1 cells, and its efficient expression requires cotransfection of GPIb␤ and GPIX (30,34). To assess shedding of GPIb␣ in CHO cells, we monitored the amount of soluble extracellular domain of GPIb␣, also known as GC (39), released into the culture medium by immunoblotting and measured concurrently the level of GPIb␣ remaining on the cell surface by flow cytometry. As in platelets (24), GPIb␣ is constitutively shed from stably or transiently transfected CHO cells expressing the GPIb-IX complex (referred in this study as Ib␣/Ib␤/IX cells). Shedding can be further up-regulated by W7 in a dose-dependent manner (Fig. 1, B and C,  supplemental Fig. S1). In addition, both basal and W7-induced GPIb␣ shedding from CHO cells is abrogated by GM6001, a broad-spectrum metalloprotease inhibitor. Finally, the complete lack of free GC in the culture medium of transfected CHO M2 cells that lack functional ADAM17 (31), despite expression of a high level of GPIb␣, clearly indicates that shedding of GPIb␣ in transfected CHO cells is catalyzed by ADAM17, as also true for platelets (Fig. 1D) OCTOBER

Identification of Intracellular Elements in GPIb␤ That
Regulate Shedding of GPIb␣-Because W7 treatment induces ectodomain shedding by presumably dissociating CaM from the shedding substrate, we reasoned that the element responding to W7-induced shedding of GPIb␣ should reside in cytoplasmic domains of the GPIb-IX complex. Moreover, genetic alteration of the cytoplasmic domains should abolish W7-in-duced shedding of GPIb␣ and/or enhance constitutive shedding of GPIb␣. Initially, whole domain replacement was carried out to locate the cytoplasmic sequence critical for W7-induced shedding of GPIb␣. As illustrated in Fig. 2A, the cytoplasmic domain of GPIb␣ was replaced by the GPIX counterpart, not known to associate with any intracellular proteins or play a role in GPIb-IX assembly (Ib␣ IX ). The cytoplasmic domain of GPIX was replaced by a polyalanine sequence (IX pA ). Because the membrane-proximal region of the GPIb␤ cytoplasmic domain is absolutely required for surface expression of the GPIb-IX complex (34), only the membrane-distal region that contains residues 161-181 was truncated in this study (Ib␤ 160⌬ ). To characterize the effect of these changes on GPIb␣ shedding, each mutant construct was cotransfected with the other two wild-type subunits into CHO K1 cells (e.g. Ib␣ IX with wild-type GPIb␤ and GPIX, designated as Ib␣ IX /Ib␤/IX in Fig. 2B). The transfected cells were treated with 150 M W7 or solvent (5% DMSO) for 2 h in PBS containing Ca 2ϩ and Mg 2ϩ at 37°C, after which the extent of W7-induced GPIb␣ shedding was monitored.
Consistent with previous observations (34,40,41), some, but not all, alterations in cytoplasmic domains caused a moderate decrease in expression of the GPIb-IX complex in transiently transfected cells (Fig. 2B). Because the level of GC released into the culture medium depends on both shedding efficiency and the abundance of GPIb␣ as shedding substrate, the difference in GPIb␣ surface expression level precluded the direct comparison of GC levels between cell lines as an accurate assessment of the mutational effect on GPIb␣ shedding efficiency. Instead, a relative ratio of the level of shedding product versus the level of remaining shedding substrate on the cell surface, the GC: GPIb␣ ratio, was normalized against the corresponding ratio observed in cells expressing the wild-type GPIb-IX complex (Ib␣/Ib␤/IX), and used to denote the extent of GPIb␣ shedding. The inclusion of only the GPIb␣ surface expression level, but not its overall expression level, in measurement of GPIb␣ shedding makes irrelevant the possibility that a mutated GPIb␤ cytoplasmic domain may alter the distribution of GPIb␣ between the cell surface and internal compartments. Comparison of GC:GPIb␣ ratios for cells transiently expressing wild-type and mutant complexes (Fig. 2B) indicated that although replacement of GPIb␣ or GPIX cytoplasmic domains, or truncation of the membrane-distal region of the GPIb␤ cytoplasmic domain, affected surface expression of GPIb␣, none of the mutations abolished W7-induced shedding of GPIb␣ nor did they significantly alter its extent.
Thus, it appears that the membrane-proximal region of the GPIb␤ cytoplasmic domain, which contains the CaM-binding site and is the only cytoplasmic sequence in  the GPIb-IX complex that had not been changed in our study, may be critical for mediating W7-induced shedding of GPIb␣. Because replacing the membrane-proximal region of the GPIb␤ cytoplasmic domain with a polyalanine sequence abolished surface expression of GPIb␣ (34), thus making it impossible to monitor its shedding, site-specific mutagenesis was performed to probe the role of this region in the regulation of GPIb␣ shedding. More specifically, GPIb␤ residues 149 -154 were mutated to Glu, one or two residues at a time. The mutant GPIb␤ was transiently cotransfected with wild-type GPIb␣ and GPIX in CHO K1 cells, to enable evaluation of constitutive shedding of GPIb␣ (Fig. 3). In all cases, GPIb␣ shedding was mediated by ADAM17, because no GC was generated from transfected CHO M2 cells lacking functional ADAM17 and surface GPIb␣ levels were relatively higher than those in transfected CHO K1 cells (supplemental Fig. S2). Some mutations, such as Ib␤ R149E in which GPIb␤ residue Arg 149 was mutated to Glu, caused a significant decrease in surface expression of GPIb␣ (Fig. 3B). Nonetheless, in most cells sufficient GPIb␣ was expressed to allow characterization of GPIb␣ shedding. All single Glu mutant complexes exhibited GC:GPIb␣ ratios indistinguishable from wild-type. In contrast, double Glu mutant complexes exhibited different GC:GPIb␣ ratios. Although the GC:GPIb␣ ratio of cells expressing the GPIb␤-R152E/L153E mutant complex (Ib␣/Ib␤ R152E/L153E /IX) was the same as that of wild-type, ratios of cells expressing Ib␣/Ib␤ R149E/L150E /IX, Ib␣/Ib␤ L150E/R151E /IX, or Ib␣/Ib␤ R151E/R152E /IX complexes were significantly higher than wild-type (Fig. 3B). These cells had little GPIb␣ on the surface, likely due to a combination of decreased GPIb␣ expression and increased shedding activity. Together, these results show that the membrane-proximal region of the GPIb␤ cytoplasmic domain is important for regulation of GPIb␣ shedding.
Mechanism of GPIb␤ Regulation of GPIb␣ Shedding-We next explored the mechanism underlying regulation of GPIb␣ shedding by the membrane-proximal region of the GPIb␤ cytoplasmic domain. In the wild-type GPIb-IX complex, GPIb␣ connects with two GPIb␤ subunits via disulfide bonds to form the GPIb complex (28). Misassembly of the complex, due to disruption of native inter-subunit interactions, often results in formation of so-called non-GPIb complexes in which GPIb␣ mostly forms non-native disulfide bonds with GPIb␤ (35). GPIb formation, as an indicator for GPIb-IX assembly, can be distinguished from non-GPIb complexes by SDS-PAGE under nonreducing conditions followed by immunoblot for GPIb␣ (Fig.  4A). Similar to earlier findings (34), many Glu mutations in this study disrupted or perturbed correct assembly of the GPIb-IX complex as evidenced by the lack of, or decrease in native GPIb formation and concurrent formation of non-GPIb complexes (Fig. 4). Consistently, coimmunoprecipitation studies showed that GPIb␣ formed a complex with each mutant GPIb␤ tested (Fig. 5). The extent of complex misassembly correlated well with the mutation-induced decrease in GPIb␣ expression, but it did not correlate with mutational effects on GPIb␣ shedding. For example, the R151E mutation hampered formation of native GPIb␣-GPIb␤ disulfide bonds to a higher degree than R152E, and caused a more pronounced decrease in GPIb␣ expression. However, both mutations had the same GC:GPIb␣ ratio as wild-type and little effect on GPIb␣ shedding (Fig. 3B). Moreover, other mutations known to cause misassembly and decrease surface expression of the GPIb-IX complex, such as the H139L mutation in the GPIb␤ transmembrane domain, or a polyleucine-replaced transmembrane domain in either GPIb␤ or GPIb␣ (33), did not increase the level of GPIb␣ shedding (Fig. 6). Even when GPIb␣ was expressed alone in transfected CHO cells, in the absence of GPIb␤ and GPIX, it was shed at a rate similar to that of the wild-type GPIb-IX com- plex (supplemental Fig. S3). Thus it is unlikely that these Glu mutations increased constitutive GPIb␣ shedding by inducing complex misassembly or inducing dissociation of GPIb␣ from GPIb␤. The membrane-proximal region of the GPIb␤ cytoplasmic domain and the shedding cleavage site of GPIb␣ are connected physically through GPIb␣-GPIb␤ disulfide bonds and their transmembrane domains. Therefore, we reasoned that the association between GPIb␣ and GPIb␤ through covalent disulfide bonds and noncovalent transmembrane domain interactions (28,29,35) should be important in mediating GPIb␤ regulation of GPIb␣ shedding. Because GPIb␣ containing the C484S/C485S double mutation (GPIb␣ C484S/C485S ) does not form disulfide bonds with GPIb␤ but still forms a wild-type-like complex with GPIb␤ and GPIX through altered interactions between transmembrane domains (32), we characterized the effects of GPIb␤ cytoplasmic mutations on ectodomain shedding of GPIb␣ C484S/C485S (Fig. 7). Although the C484S/C485S double mutation did not affect constitutive shedding of GPIb␣, most of the Glu-containing GPIb␤ mutations significantly increased the GC:GPIb␣ ratio in cells expressing GPIb␣ C484S/C485S compared with wild-type GPIb␣ ( Fig. 7 and supplemental Fig. S4). The increase was most notable for many single-Glu mutants and Ib␤ R152E/L153E . These results indicate that the GPIb␤ cytoplasmic domain may regulate GPIb␣ shedding through interactions between the two subunits. Both the membrane-proximal region of the GPIb␤ cytoplasmic domain and the interactions between GPIb␣ and GPIb␤ may be required for regulation of GPIb␣ shedding.
The membrane-proximal region of the GPIb␤ cytoplasmic domain contains a CaM-binding site (27). To test whether Glu mutations in this region affect association of CaM with GPIb␤, N-terminal HAtagged wild-type or three representative mutant GPIb␤s (HA-Ib␤) were each cotransfected with wildtype GPIb␣ and GPIX into CHO K1 cells, and HA-Ib␤ was pulled down by GST-fused CaM and glutathione-coated beads. Although wildtype HA-Ib␤ associated with GST-CaM as expected, a negligible amount of mutant HA-Ib␤ construct was pulled down (Fig. 8), indicating a lack of CaM association. This did not correlate with the differential effects of the three mutations on constitutive shedding of GPIb␣ (Fig. 3B), suggesting that mutation-induced dissociation of CaM from the GPIb␤ cytoplasmic domain may not be sufficient to induce shedding of GPIb␣.
To test the idea that a GPIb␤ cytoplasmic domain-mediated protein binding event is involved in the regulation of GPIb␣ shedding, three palmitoylated peptides based on the wild-type (pal-WT) or mutant sequences (pal-151E/152E, pal-152E/ 153E) of the membrane-proximal region of GPIb␤ cytoplasmic domain were synthesized. N-terminal palmitoylation facilitates peptide diffusion through the cell membrane into the cytoplasm (42). CHO cells stably expressing the GPIb-IX complex were treated separately with these membrane-permeable peptides, and their ability to induce GPIb␣ shedding was characterized (Fig. 9A). Although treatment with pal-WT, or pal-152E/153E, induced release of GC into the culture medium and a corresponding decrease of GPIb␣ on the cell surface in a dosedependent manner, treatment with pal-151E/152E had little effect. Treatment with palmitoylated peptides did not affect GPIb formation as shown by Western blot under non-reducing conditions (Fig. 9B), thus excluding the possibility that the  treatment may induce GPIb␣ shedding by dissociating GPIb␣ from GPIb␤. Instead, the relative effects of these synthetic peptides on GPIb␣ shedding matched perfectly those of the corresponding Glu mutations in transfected cells, suggesting that these peptides may affect shedding of GPIb␣ by inducing dissociation of intracellular proteins from the endogenous GPIb␤ cytoplasmic domain (Fig. 10).

DISCUSSION
Biological signals are transmitted through cell surface receptors across the cell membrane in both directions. The shedding cleavage site is located near the junction of extracellular and transmembrane domains of a transmembrane protein. Thus modulation of ectodomain shedding by its cytoplasmic domain can be considered as an inside-out signaling process. Previous reports have shown that the effecting cytoplasmic domain and the shed extracellular domain are located in the same polypeptide. The unique subunit organization of the GPIb-IX complex prompted us in the present study to investigate whether a cytoplasmic domain can regulate ectodomain shedding of an associated subunit. Using mutational analysis of GPIb-IX in transfected CHO cells, we demonstrated that the cytoplasmic domains of GPIb␣ or GPIX, or the membrane-distal region of the GPIb␤ cytoplasmic domain, are not required for either constitutive or W7-induced shedding of GPIb␣ (Fig. 2). Furthermore, site-specific mutations in the membrane-proximal region of the GPIb␤ cytoplasmic domain significantly increased constitutive shedding of GPIb␣ (Fig. 3). Finally, perturbing the noncovalent interaction between GPIb␤ and GPIb␣ further lessened the restraint of the GPIb␤ cytoplasmic domain on shedding of GPIb␣ (Fig. 7). Together, these results demonstrate that the membrane-proximal region of the GPIb␤ cytoplasmic domain helps to maintain stable surface expression and inhibits shedding of GPIb␣. To the best of our knowledge, this is the first report of regulation of protein ectodomain shedding by the cytoplasmic domain of another polypeptide. This novel transsubunit aspect of GPIb␣ shedding regulation will have significant implications for general shedding regulatory mechanisms.
How do cytoplasmic domains regulate ectodomain shedding on the other side of the cell membrane? The underlying molecular and structural basis remains elusive, but the association of the cytoplasmic domain with intracellular proteins clearly plays a role. CaM is known to be associated with the cytoplasmic domain of a number of shed proteins (11,(13)(14)(15)(16)(17)(18)(19)(20)(21). However, our study showed that the R152E/L153E double mutation in the GPIb␤ cytoplasmic domain disrupts GPIb␤ association with CaM but exerts little effect on GPIb␣ shedding (Fig. 3, 8). This would argue against a critical role for CaM association in directly mediating regulation of GPIb␣ shedding by the GPIb␤ cytoplasmic domain. Our results suggest that an intracellular binding partner other than CaM may be associated with the membrane-proximal region of the GPIb␤ cytoplasmic domain. Disruption of the association, either by site-specific mutations or membrane-permeable synthetic peptides based on this GPIb␤ sequence may lead to an increase in the accessibility of GPIb␣ to the sheddase (Fig. 10). Although the identity of this intracellular protein remains to be uncovered, its ability to associate with the wild-type GPIb␤ cytoplasmic domain or the R152E/L153E mutant, but not the R151E/R152E or R149E/ L150E mutants, may be helpful in its identification.
Induction of ectodomain shedding by membrane-permeable CaM inhibitors is well documented, implicating CaM as a regulatory factor in ectodomain shedding (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). However, if CaM association with the GPIb␤ cytoplasmic domain is not directly involved in regulating shedding, how could the effect of CaM inhibitors on GPIb␣ shedding be explained? One possibility is that CaM inhibitors may have other effects in addition to dissociating CaM from GPIb␤ in the cell. For instance, W7 was recently shown to insert into the plasma membrane and change the electrostatic surface potential of the membrane (43), which may influence a number of cellular processes and lead to stimulation of ectodomain shedding. Another possibility is that CaM may regulate other intracellular interactions involving the GPIb␤ cytoplasmic domain. For example, it was recently shown that CaM and moesin bind to the same intracellular membrane-proximal region of L-selectin, and that both proteins may be involved in L-selectin shedding (44,45). Additional investigation will be required to explore these possibilities. In this regard, our findings on trans-subunit regulation of GPIb␣ shedding by GPIb␤ are significant because they imply intracellular regulators need not interact directly with the shed protein, but may act via an associated transmembrane protein.