The Transmembrane Domains of β and IX Subunits Mediate the Localization of the Platelet Glycoprotein Ib-IX Complex to the Glycosphingolipid-enriched Membrane Domain*

Background: Localization of the GP Ib-IX complex to the lipid domain is mediated by the β and IX subunits. Results: Mutations in β/IX TMDs inhibit GP Ib-IX localization to the lipid domain. Conclusion: Localization of the GP Ib-IX complex to the lipid domain is mediated by β/IX TMDs. Significance: The β/IX TMDs may be a novel therapeutic target. We have previously reported that the structural elements of the GP Ib-IX complex required for its localization to glycosphingolipid-enriched membranes (GEMs) reside in the Ibβ and IX subunits. To identify them, we generated a series of cell lines expressing mutant GP Ibβ and GP IX where 1) the cytoplasmic tails (CTs) of either or both GP Ibβ and IX are truncated, and 2) the transmembrane domains (TMDs) of GP Ibβ and GP IX were swapped with the TMD of a non-GEMs associating molecule, human transferrin receptor. Sucrose density fractionation analysis showed that the removal of either or both of the CTs from GP Ibβ and GP IX does not alter GP Ibα-GEMs association when compared with the wild type. In contrast, swapping of the TMDs of either GP Ibβ or GP IX with that of transferrin receptor results in a significant loss (∼50%) of GP Ibα from the low density GEMs fractions, with the largest effect seen in the dual TMD-replaced cells (>80% loss) when compared with the wild type cells (100% of GP Ibα present in the GEMs fractions). Under high shear flow, the TMD-swapped cells adhere poorly to a von Willebrand factor-immobilized surface to a much lesser extent than the previously reported disulfide linkage dysfunctional GP Ibα-expressing cells. Thus, our data demonstrate that the bundle of GP Ibβ and GP IX TMDs instead of their individual CTs is the structural element that mediates the β/IX complex localization to the membrane GEMs, which through the α/β disulfide linkage brings GP Ibα into the GEMs.

We have previously reported that the structural elements of the GP Ib-IX complex required for its localization to glycosphingolipid-enriched membranes (GEMs) reside in the Ib␤ and IX subunits. To identify them, we generated a series of cell lines expressing mutant GP Ib␤ and GP IX where 1) the cytoplasmic tails (CTs) of either or both GP Ib␤ and IX are truncated, and 2) the transmembrane domains (TMDs) of GP Ib␤ and GP IX were swapped with the TMD of a non-GEMs associating molecule, human transferrin receptor. Sucrose density fractionation analysis showed that the removal of either or both of the CTs from GP Ib␤ and GP IX does not alter GP Ib␣-GEMs association when compared with the wild type. In contrast, swapping of the TMDs of either GP Ib␤ or GP IX with that of transferrin receptor results in a significant loss (ϳ50%) of GP Ib␣ from the low density GEMs fractions, with the largest effect seen in the dual TMD-replaced cells (>80% loss) when compared with the wild type cells (100% of GP Ib␣ present in the GEMs fractions). Under high shear flow, the TMD-swapped cells adhere poorly to a von Willebrand factor-immobilized surface to a much lesser extent than the previously reported disulfide linkage dysfunctional GP Ib␣-expressing cells. Thus, our data demonstrate that the bundle of GP Ib␤ and GP IX TMDs instead of their individual CTs is the structural element that mediates the ␤/IX complex localization to the membrane GEMs, which through the ␣/␤ disulfide linkage brings GP Ib␣ into the GEMs.
The function of the GP Ib-IX complex can be regulated by various mechanisms, including conformational changes in each individual subunit (1-7), extracellular intersubunit interaction (6, 8 -11), oligomerization to increase the avidity and affinity to its ligand (12), intracellular association with signaling molecules (e.g. 14-3-3 (13), P I3-kinase (14), and calmodulin (15)), and co-localization with other receptors on the platelet surface (e.g. FcR␥ (16) and Fc␥RIIA (17,18)). One of the features of cell membranes is the existence of specialized glycosphingolipidenriched membranes (GEMs), 3 to which not only a number of receptors, in platelets, such as GP Ib-IX complex (19), Fc␥RIIA (19), P 2 X 1 (20), and P 2 Y 12 (21), but also various kinases, phosphatases, and adaptor proteins (22,23) are localized. Therefore, one would expect that upon ligand-receptor interaction, the receptor-associated GEMs can act as a platform or carrier to support a transmembrane signal, as well as the interplay of different signals elicited by various GEM-associated receptor-ligand interactions, helping them bind in an efficient, controlled, and synergistic manner. Furthermore, a number of investigations have shown that the single bond between the GP Ib-IX complex and vWf cannot be maintained under high shear force imposed on either platelets or the GP Ib-IX-expressing CHOs. In addition, vWf is a large multimeric molecule possessing multiple sites that can interact with GP Ib␣; therefore, GP Ib-IX complexes could be clustering on vWf, which allows for resistance to the shear force through the anchorage of the GP Ib-IX complex to the platelet cytoskeleton (3)(4)(5). Along this line, a GEM-based localization to concentrate larger numbers of GP Ib␣ molecules should facilitate such multivalent binding.
It is known that all of the GP Ib-IX subunits (GP Ib␣, GP Ib␤, and GP IX) are type I transmembrane proteins that reside in the GEMs. Lack of or dysfunction in this association caused by a disruption of the GEMs structure (e.g. M␤CD treatment (19)) or introduction of a loss of association mutations to GP Ib␣ (24) eliminates or inhibits the GP Ib-IX complex interaction with vWf, tyrosine phosphorylation, and GP Ib-IX complex-initiated thrombus formation. 4 At this point, we only know that GP Ib␤ and GP IX possess unidentified structural elements required for the association of the GP Ib-IX complex with GEMs that is independent of the palmitoylation state of GP Ib␤ and GP IX. Although the transmembrane domains (TMDs) of these two molecules seem like good candidates because of their location near the membrane lipids (25)(26)(27)(28), there is no experimental evidence to support this notion. In addition, previous reports have shown that the association of a transmembrane receptor with GEMs is not predisposed by its TMD, but rather, it is mediated by the specific targeting signals residing in the individual protein itself (29). For instance, conformational changes in the extracellular domain can affect the GEMs targeting of the T-cell adhesion protein CD2 (98); a 60-residue membrane-proximal sequence is reported to mediate the EGF receptor to the GEMs (99); and a membrane-proximal signal in the cytoplasmic tail (CT) of the T-cell co-receptor CD4 is required for association with the GEMs (85). In this study, because even minor changes in the extracellular portions of GP Ib␤ and GP IX may cause nonexpression of the entire GP Ib-IX complex (6,30), we chose to start our investigation with the roles of the CTs and TMDs of GP Ib␤ and GP IX. We sequentially mutated them by either truncation or swapping with the TMD of Trf, a known non-GEMs associating transmembrane protein (28) to examine whether any of them play a role in mediating the localization of the GP Ib-IX complex to the GEMs.

Experimental Procedures
Swapping the TMDs of GP Ib␤ and GP IX with That of Trf and Deletion of the CTs of GP Ib␤ and GP IX-All constructs containing Ib␤ and IX have been previously described (24). The TMDs and the palmitoylation sites of both GP Ib␤ ( 123 WGAL-AAQLALLGLGLLHALLLVLLLC 148-palm , 26 amino acids, ␤ TC ) and GP IX ( 136 VALVAVAALGLALLAGLLC 154-palm , 19 amino acids, IX TC ) were replaced with the inverted and depalmitoylated human Trf receptor 1 TMD ( 88 YGLYGIMFGILFF-VIVAITGYA 67 IS 65 , 24 amino acids (accession no. P02786). The underlined/bold alanine is mutated from cysteine (28). Based on the Trf TMD DNA sequence, the following four overlapping PCR primers were designed to replace the TMDs of GP Ib␤ and GP IX. The underlined sequences are the Trf TMD coding or complementary sequences: sequence 1, ATAGCCC-AAGTAGCCAATCATAAATCCAATCAAGAAAAAGACG-ATCACAGCAATAGTCCCATAGGCGATACTGCAGAGC-GGGCCGGGAG (GP Ib␤ overlapping forward primer); sequence 2, AGTATCGCCTATGGGACTATTGCTGTGATCG-TCTTTTTCTTGATTGGATTTATGATTGGCTACTTGG-GCTATCGCCTGCGGAGG (GP Ib␤ overlapping reverse primer); sequence 3, ATAGCCCAAGTAGCCAATCATAAA-TCCAATCAAGAAAAAGACGATCACAGCAATAGTCCC-ATAGGCGATACTGTCCCACAAGACCC (IX overlapping forward primer); and sequence 4, AGTATCGCCTATGGGA-CTATTGCTGTGATCGTCTTTTTCTTGATTGGATTTA-TGATTGGCTACTTGGGCTATGCCACCACAGAGGC (IX overlapping reverse primer). Whenever overlapping PCR was performed, there were always two rounds of amplification involved. The first reaction was comprised of two PCRs with 1) a GP Ib␤ or GP IX cDNA forward primer (5Ј prime with start codon and restriction site) and the overlapping reverse primer, or 2) the overlapping forward primer and a GP Ib␤ or GP IX cDNA reverse primer (3Ј prime with stop codon and restriction site). The resulting two products were purified by agarose gel electrophoresis, mixed in an equal molar ratio, annealed, and extended with a high fidelity polymerase (Turbo Pfu; Stratagene) and then used as templates for the second round of reactions where the cDNA forward and reverse primers of the GP Ib␤ or GP IX flanked with restriction sites were used to amplify and clone the Trf-TMD-swapped Ib␤ and IX cDNAs into the PDX vector (31). To delete both of the CTs and the intramembrane proximal palmitoylation sites of GP Ib␤ (␤ delϩC ) and GP IX (IX delϩc ), Turbo Pfu-driven PCRs were performed with specific primers complementary to the transmembrane sequences where two stop codons were introduced to the original palmitoylation cysteine sites to ensure translational stop. All of the TMD replacements and CT deletions mentioned were verified through sequencing.
Generation of CHO Cell Lines Expressing Wild Type and Mutant GP Ib-IX Complex-CHO cells expressing wild type and mutant GP Ib-IX complexes were generated and maintained as previously described (5,24,31). GP Ib␣-positive cells were first incubated with GP Ib␣-specific mouse monoclonal antibody SZ2 (Beckman Coulter) and then sorted by antimouse IgG conjugated magnetic beads (Invitrogen) ( Table 1). The expression levels of GP Ib␣ in these cells were determined by flow cytometry with a phycoerythrin-labeled anti-GP Ib␣ antibody (Beckman Coulter), and their mean fluorescence intensities were normalized then normalized to the wild type cells. Standard errors were calculated from three independent experiments.
Sucrose Density Gradient Centrifugation-A detailed procedure has been described previously (24). In brief, cells expressing the GP Ib-IX complex were lysed with 1% Brij 35 and then loaded to the top of a sucrose density gradient. After centrifugation at 34,000 rpm in a swinging bucket centrifuge for 18 h at 4°C, 12 equal fractions were taken from the top of the gradient and analyzed using an SDS-PAGE gel. GP Ibs were then detected by Western blotting with each respective antibody (24). The GEM-associating molecules were all found within the first four low density fractions.

Cell line Characteristics
Swapping with Trf TMD ␣ WT ␤ TC IX WT ␣ WT ␤ TC IX TC rpm for 1 min and lysed with 1% Brij 35 for 1 h on ice. After centrifugation at 15,000 ϫ g for 10 min, the supernatants were precleared with 50 l of protein A/G beads (GE Healthcare Life Sciences) for 30 min at 4°C. After removing the protein A/G beads by centrifugation, the samples were incubated overnight with 1 g of anti-GP IX monoclonal SZ1 antibody (Beckman Coulter) (32) and then immunoprecipitated with protein A/G beads for 1 h at 4°C. The beads were then pelleted by centrifugation and washed three times, and the bound protein was eluted by boiling in 2ϫ sample buffer containing 2% ␤-mercaptoethanol. Immunoprecipitated proteins were resolved by 4ϳ12% Novex Tris-glycine gels (Life Technologies) under reducing conditions and transferred to PVDF for Western blotting and detection of GP Ib␣ and GP Ib␤.
Flow Chamber Assay-A detailed procedure has been described previously (31). In brief, CHO cells were incubated on immobilized vWf (20 g/ml) for 1 min in a parallel plate flow chamber and then perfused with TBS, 0.5% BSA (Sigma) at flow rates that generated wall shear stresses of 2.5, 10, or 20 dyn/cm 2 . The experiments were recorded in real time for 1 min, by a high speed digital camera (model Quantix; Photometrics, Tucson, AZ) connected to an inverted stage microscope (Eclipse TE300; Nikon, Garden City, NY). The rolling velocity was defined as the distance a cell traveled during a defined period (m/s). Each rolling experiment was performed three times, and the error bars were calculated from the mean rolling velocities of 100 cells in five different view fields.

CT Truncation and Depalmitoylation of Either or Both of GP Ib␤ and GP IX Do Not Interfere with GP Ib␣ Localization to the
GEMs-To investigate whether the CTs of GP Ib␤ and GP IX play any roles in the localization of GP Ib␣ to the membrane GEMs, we generated stable CHO cell lines where wild type GP Ib␣ was co-transfected with either CT-truncated and depalmitoylated GP Ib␤ (␤ delϩc ), GP IX (IX delϩc ), or both (␤ delϩc IX delϩc ). To rule out the possibility that the transfection reagents change the nature of the membrane in the cells, we made stable cell lines instead of using a transient expression approach. As shown in Fig. 1 (A and B), when compared with   Fig. 1B). Upon extraction with 1% Brij 35, nearly identical distribution patterns of the GP Ib subunits from the sucrose gradient fractionation were seen in both the mutant and wild type cells (24). All the GP Ib molecules are present in the low density GEM fractions (fractions No. 1-3), as identified by blotting the known GEM-associating caveolin (Fig. 1C) (24). To test whether the mutations affect the formation of the GP Ib-IX complex, we lysed the cells with 1% Brij 35 and incubated the lysates with SZ1, an anti-GP IX specific antibody (32,33). Upon immunoprecipitation, we observed that all the GP Ib subunits can be detected in the precipitates of our ␣␤IX-expressing cells (Fig. 1D). In contrast, only GP Ib␤ can be detected in the ␤IXexpressing cells, validating previous findings that GP Ib␤ and GP IX are tightly associated (10,34). Taken together, our data not only clearly demonstrate that the structural segment possessing the GEM-targeting signal does not reside in the CTs of GP Ib␤ and GP IX but also further supports our previous finding that the association of the GP Ib-IX complexes with the GEMs is not predisposed by the palmitoylation state of either GP Ib␤ or GP IX.

Swapping the TMD of Either or Both of GP Ib␤ and GP IX with That of Trf Inhibits GP Ib␣ Localization to the GEMs-To
investigate whether the TMDs and/or which TMD of GP Ib␤ and GP IX plays a role in partitioning the GP Ib-IX complex into the GEMs, we replaced the TMDs of GP Ib␤ and GP IX with that of Trf, a known non-GEMs associating transmembrane protein. Previous studies have shown that upon replacement of the TMD with Trf TMD, IgE receptor function was abolished because of a segregation of the host molecule from the GEMs (28). Because the Trf is a type II transmembrane protein, whereas GP Ib␤ and GP IX are type I transmembrane proteins, we inserted an inverted Trf TMD sequence into both, such that the orientation of the Trf TMD across the plasma membrane is the same as its normal orientation in the Trf receptor. Moreover, it was shown that Cys 68 can be palmitoylated, another piece of evidence demonstrating that palmitoylation itself does not predispose the host protein to associate with the GEMs. To prevent obscuring the interpretation, we also mutate this cysteine to an alanine (Cys 68 to Ala 68 ). After sorting, we found that the individual swapping of the TMD of either GP Ib␤ (␣␤ TC IX WT ) or GP IX (␣␤ WT IX TC ) greatly reduced the expression of GP Ib␣ to a level ϳ60% of that in the wild type cells (Fig. 2, A and B). This suggests that the TMDs of both GP Ib␤ and GP IX contribute to the expression of GP Ib␣. When both of the TMDs were replaced (␣␤ TC IX TC ), the GP Ib␣ expression level was ϳ16% of that in the wild type cells (Fig. 2B), suggesting that the TMD bundle formed by ␤/IX harbors the most appropriate conformation and surface nature for the GP Ib-IX complex to be efficiently expressed on the cell surface. Furthermore, when we subjected our cell lines to sucrose density fractionation analysis, we found that individual replacements of the TMD of GP Ib␤ and GP IX caused more than 50% of the GP Ib␣ to translocate to the non-GEM fractions (48% in ␣␤ WT IX TC and 42% in ␣␤ TC IX WT cells; Fig. 2D), when compared with the wild type cells (100% in ␣␤ WT IX WT ; Fig. 1C). Interestingly, swapping the GP IX TMD had less of an impact than that of the GP Ib␤ TMD. Moreover, dual replacement of both TMDs (␣␤ TC IX TC ) partitioned greater than 80% of the GP Ib␣ into non-GEM fractions (ϳ18% left in the GEM fractions; Fig. 2D). Finally, the GP Ib-IX complex is still intact because all of the GP Ib subunits appeared in similar sucrose density fractions (Fig. 2C), and the mutant GP Ib␤ and GP IX still associate with GP Ib␣ upon immunoprecipitation with anti-GP IX antibody SZ1 (Fig. 2E). Thus, for the first time our data demonstrate that 1) the TMDs of both GP Ib␤ and GP IX contribute to the association of the GP Ib-IX complex with GEMs. and 2) the TMD bundle of GP Ib␤ and GP IX is a critical structural segment in the GP Ib-IX complex for GEM association.
Loss of GEM Association Inhibits GP Ib-IX Complex Interaction with vWf under Flow-We have reported that the disulfide linkage dysfunctional GP Ib␣ expresses at a level ϳ85% of the wild type GP Ib␣, of which ϳ40% still associates with the CHO GEMs. This reduction greatly increases the rolling velocity of these cells at high shear stresses when compared with wild type cells (ϳ2-fold and ϳ3-fold increase at 10 and 20 dyn/cm 2 , respectively) (24). To our surprise, however, when we perfused our CHO cells expressing individually Trf-TMD replaced GP Ib␤ or GP IX, they rolled at a velocity 4 -5-fold faster than the wild type cells (Fig. 3). As shown in Fig. 2 (B and D), ␤/IX individual swapped cells express GP Ib␣ at an average level that is ϳ60% of the wild type GP Ib␣, and there were ϳ50% of the GP Ib␣ present in the GEMs of these CHO cells when compared with the wild type cells (100%; Fig. 1C). Therefore, we estimated the ratio of the GP Ib␣ copy number in the GEMs of CHO cells expressing either disulfide linkage deficient GP Ib␣ or ␤/IX TMD-swapped GP Ib␣ to be 85% ϫ 40%:60% ϫ 50% ϭ 34:30. Moreover, consistent with previous reports, the non-GP Ib␣expressing CHO cells (␤/IX) do not roll on the vWf-coated surfaces (35), and the previously reported GP Ib␣ mutation (␣ W129F ␤ WT IX WT ) causes no change in the rolling velocity (31), indicating that the faster rolling of the ␤/IX individual TMDswapped cells is specifically due to an impaired vWf binding ability. Thus, our data suggest that at high shear stress, loss of the GEM association and alteration of the transmembrane domains of GP Ib␤ and GP IX may synergistically impair the vWf interaction with GP Ib␣ on these individual TMDswapped cell surfaces. Unfortunately, when we examined our GP Ib␤/GP IX TMD dual swapped mutant CHO cells (␣␤ TC IX TC ), we could find fast rolling cells on the immobilized vWf surface only at a shear stress of 2.5 dyn/cm 2 . Because the expression level of GP Ib␣ in these cells is ϳ16% of that in the wild type cells (Fig. 2B), we speculate that the loss of vWf binding could not be solely attributed to the reduction of the GEM association upon dual swapping of the ␤/IX TMDs. Rather, it might be due to the decreased density of GP Ib␣ on the cell surface (35).

Discussion
Even though we have known that the GP Ib␤/GP IX complex possesses the targeting signal for the localization of the com-plex to the membrane GEMs, the detailed structural necessity of them remains unknown (24). In this study, we aimed to identify these structural elements. The reason we began our investigation from the CTs and the TMDs of GP Ib␤ and GP IX is because both types of domain have been shown to mediate GEM association in other proteins (25)(26)(27)(28). In the case of the GP Ib-IX complex, three features may give the complex the ability to associate with the GEMs: 1) all of the GP Ib-IX sub- units are transmembrane proteins with TMDs of different length that can interact with membrane lipids; 2) the intramembrane proximal amino acids in GP Ib␤ are positively charged, which possibly can bind to negatively charged GEMspecific lipids; and 3) both GP Ib␤ and GP IX are palmitate modified proteins, a possibility that we have previously ruled out. In our sequential mutagenesis, we first altered the CTs and then the TMDs of GP Ib␤ and GP IX. We found that only the changes to the TMDs altered the association with the GEMs, with the largest effect seen in the dual TMD-replaced cells. This observation is particularly interesting because it suggests that the ␤/IX TMD bundle, as a whole, could be the structural segment for GP Ib␣ partitioning into the GEMs. Our data implicate a new feature of the regulation of the GP Ib-IX function, i.e. the orderly formation of ␤/IX TMD bundles that produces the most appropriate surface for interaction with the GEM-specific lipids, thereby mediating the GP Ib-IX complex localization to the GEMs. It is interesting to note that CT-removed and depalmitoylated GP Ib␤ only slightly reduced GP Ib␣ expression when wild type GP IX is co-transfected (␣␤ delϩc IX WT ). In comparison, when both GP Ib␤ and GP IX have their CT domains removed, and they are depalmitoylated, the expression of GP Ib␣ is even higher than the wild type cells (␣␤ delϩc IX delϩc ). These observations are surprising, because they are largely opposite to what has been described previously where the cells transfected with CT-removed but palmitoylation intact GP Ib␤ (␤ del ) were unable to express GP Ib␣ on their surface (36). The only difference between these two CT-removed GP Ib␤ proteins is our mutant GP Ib␤ is also depalmitoylated. It is well known that palmitate modifications play a role in the sorting and trafficking of host proteins (37). Even though no function has been attributed to the acylation of GP Ib␤, previous evidence (36) and our data implicate that both the palmitoylation and the intramembrane proximal basic amino acids ( 148-palm CRLRRLRARARA-RAAAR 164 ) of GP Ib␤ are needed to determine the intracellular sorting route of wild type GP Ib␤ for successful expression, which then allows GP Ib␣ to reach the cell surface. The insertion of an acyl group into and the binding of the positively charged amino acids to the inner leaflet of the phospholipid bilayer of the intracellular organelles might be the mechanism. The acyl group and the short CT in the GP IX subunit may also contribute to this, because without it, the GP Ib␣ expression was also slightly reduced in ␣␤ WT IX delϩc cells. On the other hand, mutant GP Ib␤ in ␣␤ delϩc IX WT cells can restore to a level ϳ81% of the wild type cells (Fig. 1B) when compared with the CT-removed but palmitoylation intact GP Ib␤ (␤ del ) (36). These data suggest that without the governing factors (CT and palmitate modification), GP Ib␤ might employ a different sorting route to achieve a relatively successful expression of itself and therein GP Ib␣. The choice of such different sorting route may also be facilitated by simultaneous removal of the acyl chain and CT from the GP IX subunit, thereby achieving a higher GP Ib␣ expression in ␣␤ delϩc IX delϩc cells than in the wild type cells. Further investigations on the sorting and secretory pathways in these cells for the GP Ib-IX complex expression are needed to clarify these possibilities.
One other interesting observation is that both of our individual TMD-swapped cells (␣␤ TC IX WT and ␣␤ WT IX TC ) achieved appreciable and comparable GP Ib␣ expression on their cell surfaces (Fig. 2, A and B). In contrast, previous reports have shown that transient expression of mutant GP Ib␤ and GP IX whose TMDs are replaced either by polyleucine (pL, ␤ pL and IX pL ) or polyleucine-alanine (pLA, ␤ pLA and IX pLA ) residues causes nonexpression of GP Ib␣ only in mutant GP Ib␤ cells instead of mutant GP IX cells (38). Therefore, it is possible that 1) the pL or pLA type of TMD may by itself interfere with the expression of GP Ib␤ and therein GP Ib␣, and/or 2) the Trf TMD in the swapped GP Ib␤ has an improved capability of forming stable bonds with the TMDs of GP Ib␣ and GP IX to maintain a relatively higher expression of GP Ib␣ on the cell surface. On the other hand, in contrast to the CT removal and depalmitoylation, TMD swapping also altered the GP Ib␣-GEM association. Therefore, it is possible that the impaired GEM association may also contribute to the reduction in GP Ib␣ expression in the TMD-swapped cells.
We further provide evidence showing that the reduction in GP Ib␣-GEM association by the altered TMDs of GP Ib␤ and GP IX can inhibit GP Ib␣ interaction with vWf at high shear. Unexpectedly, even though the TMD-swapped cells (␣␤ TC IX WT and ␣␤ WT IX TC ) have comparable levels of GP Ib␣ found in the GEMs to the disulfide linkage dysfunctional GP Ib␣-expressing cells (␣ SS ␤ WT IX WT ), we found that these cells have significantly reduced capability of resisting the high shear force (4 -5 times faster than the wild type cells) when compared with the disulfide linkage dysfunctional GP Ib␣-expressing cells (␣ SS ␤ WT IX WT , 2-3 times faster than the wild type cells) (24). Thus, our data suggested that the faster rolling of the TMDswapped cells may be caused by both the reduction of GEM association and the alteration of the ␤/IX TMDs. For the latter, because it has been shown that 1) the interaction between GP Ib␤ and GP IX is primarily mediated by their TMDs (10) and 2) the GP Ib␣ TMD can also interact with GP Ib␤ and GP IX forming a four-helix bundle (6), it is possible that the inter-TMD interactions among ␣, ␤, and IX in our TMD-swapped cells are weakened. Upon binding of the GEM-associating GP Ib-IX complex to the immobilized vWf, an impaired clustering of the GP Ib-IX complex inside the GEMs caused by inefficient inter-TMD interactions may also affect the multivalency of the vWf-GP Ib␣ bonding, thereby contributing to the poor resistance to high shear force and faster movement of the ␤/IX TMD-swapped cells under flow. Furthermore, we did not find dual TMD-swapped cells adhere to the vWf-immobilized surface at shear stress of 10 dyn/cm 2 or higher. Nevertheless, because we were unable to increase the GP Ib␣ expression to a level comparable with that in either individual TMD-swapped cells, we do not intend to attribute the loss of vWf binding solely to the dramatic decrease in the GP Ib␣-GEM association in these cells. Based on previous report regarding the effect of GP Ib␣ density on cell rolling velocity, instead, we speculate that the loss of vWf binding might be primarily due to the low level of GP Ib␣ expression on the cell surface. Taken together, our study provides the first evidence that the TMDs of GP Ib␤ and GP IX are critical for the GP Ib-IX complex to associate with the GEMs (membrane location). In addition, our data also suggest that ␤/IX TMDs contribute to the oligomerization of the GP Ib-IX complex (clustering) upon vWf binding, a notion that needs further investigation to ascertain whether that is the case. Nevertheless, considering the significantly increased rolling velocities of the TMD-swapped mutant cells, we speculate that targeting the TMDs of GP Ib␤ and GP IX may have the potential to become a therapeutic approach to specifically and potently interfere with the shear-induced GP Ib-IX-vWf interaction.
Author Contributions-G. X., D. S., Z. Z., and Y. P. designed the study and wrote the manuscript. G. X., D. S., and Z. Z. designed, performed and analyzed the experiments shown in the table and figures. Y. R. helped with the cell culture. T. S. S. helped with manuscript editing. J. A. L. provided critical reagents. All authors have approved the final version of the manuscript.