SPECIFICITY FOR HOMO-VERSUS HETERO-OLIGOMER FORMATION IN INTEGRIN TRANSMEMBRANE HELICES

Abstract

Transmembrane helices engage in homomeric and heteromeric interactions that play essential roles in folding and assembly of transmembrane proteins. However, features that explain their propensity to interact homomerically or heteromerically and determine the strength of these interactions are poorly understood. Integrins are an ideal model system to address these questions because the transmembrane helices of full-length integrins interact heteromerically when integrins are inactive, but the isolated transmembrane helices are also able to form homo-dimers or homo-oligomers in micelles and bacterial membranes. We sought to determine the features defining specificity for homo versus hetero interactions by conducting a comprehensive comparison of the homomeric and heteromeric interactions of the integrin αIIbβ3 transmembrane helices in biological membranes. Using the TOXCAT assay, we found that residues V700, M701, A703, I704, L705, G708, L709, L712, and L713, located on the same face of the β3 helix, mediate homodimer formation. We then characterized the β3 heterodimer by measuring the ability of β3 helix mutations to cause ligand binding to αIIbβ3. We found that mutating V696, L697, V700, M701, A703. I704, L705, G708, L712, and L713, but not the small-X₃-small motif, S699-X₃-A703, caused constitutive αIIbβ3 activation, as well as persistent αIIbβ3 activation-dependent FAK phosphorylation. Because αIIb and β3 use the same face of their respective transmembrane helices for homomeric and heteromeric interactions, the interacting surface on each has an intrinsic “stickiness” predisposing towards helix-helix interactions in membranes. The residues responsible for heterodimer formation comprise a network of interdigitated sidechains with considerable geometric complementarity; mutations along this interface invariably destabilize heterodimer formation. By contrast, residues responsible for homomeric interactions are dispersed over a wider surface. While most mutations of these residues are destabilizing, some stabilized homo-oligomer formation. We conclude that the αIIbβ3 transmembrane heterodimer shows the hallmark of finely-tuned heterodimeric interaction, while the homomeric interaction is less specific.

Introduction

Lateral interactions between transmembrane (TM) helices play essential roles in folding and assembly of multi-span TM proteins as they exit the translocon. They also mediate assembly and signaling from single-span TM proteins in a manner similar to water-soluble modules such as coiled coils, helix-loop-helix motifs, HAMP and other domains that associate/dissociate or rearrange as part of cellular signaling ^{1; 2}. The soluble transcription factor Jun is a relevant example. Jun exhibits a substantial propensity to form homooligomers, but an even stronger propensity to interact heteromerically with its physiological partner Fos ^{3; 4}. In the case of TM proteins, identifying TM motifs with strong predilection to engage in TM helix associations and dissecting the features that define their preference for homo-versus heterooligomeric assembly will further our understanding of protein recognition and folding.

Integrins provide an ideal system for understanding the propensity for heteromeric versus homomeric helical interactions in membranes. Integrins are heterodimers composed of α and β subunits, each containing a single TM helix ⁵. In resting integrins, the α and β TM helices form a tightly interacting heterodimer. However, when the integrins are activated (i.e., competent for clustering, interacting with extracellular ligands, and modulating intracellular signal transduction cascades), the TM domains physically separate ^{6; 7}. The extracellular domains of activated integrins remain associated, but undergo global conformational changes in response to dissociation of their TM helices ⁶. Thus, the strength of the TM domain heterodimer must be sufficient to maintain integrins in their resting state, but not too strong to prevent TM domain separation, and subsequent integrin activation, in response to cellular stimulation.

As is the case for the oligomerization domains of many water-soluble heterodimeric transcription factors, integrin TM helices, as isolated peptides in vitro, are able to interact homomerically as well as heteromerically ^{8; 9}. The homomeric association of isolated integrin TM domains has been well documented in micelles and bacterial membranes ^{8; 9} and the first inquiries into the role of integrin TM helices focused on their tendency to form homooligomers ¹⁰. If homooligomerization occurred only after dissociation of the heterodimeric TM pair, then it would drive the equilibrium towards the activated state and provide a mechanism for clustering. Asn-scanning mutagenesis provided support for this hypothesis; mutations that increased homodimerization of the β3 TM helix indeed activated the integrin and electron microscopy of the intact integrin supported this hypothesis ¹⁰. However, subsequent studies found that the αIIbβ3 activation resulting from αIIb and β3 TM helix mutations might instead result from disruption of αIIbβ3 TM helix/helix interaction ¹¹ and most ^{12; 13; 14}, but not all ¹⁵, direct experimental analyses failed to observe TM homooligomerization in integrin activation under physiological conditions. Thus, it appears that homooligomerization is not a dominant or obligatory mechanism in integrin activation and in many cases, it could not be observed in a cellular context. Accordingly, it is interesting to ask what molecular features might discriminate between homo-versus heterooligomerizaton of integrin TM helices. Such a comparative study gets to the heart of the molecular nature of sequence-specific asymmetric helix-helix interactions in the membrane, for which data is currently almost entirely lacking.

Previous studies have established the tendency of the TM domains of various integrins to interact homomerically, as well as heteromerically, in micelles and bacterial membranes ^{8; 9; 16}. The sequence-dependence of heteromeric interactions has been evaluated in bacterial reporter systems that measure the ability of TM helix heterodimers to bring together DNA-binding domains that either induce or repress transcription of specific reporter genes ^{17; 18}. In parallel, sequence-dependent interactions between integrin TM domains have been evaluated by measuring the functional effects of single-site amino acid substitutions in full-length integrins ^{11; 19}. A similar set of residues identified in bacterial reporter system as mediating heteromeric interactions were found to be essential for stabilizing the resting state; in other words, mutations disrupting TM heterodimerization caused constitutive integrin activation.

The interaction of integrin TM helices has been evaluated most extensively for the major platelet integrin αIIbβ3 ²⁰. The sequence motifs involved in αIIbβ3 TM heterodimerization have been identified, leading to a model for the αIIbβ3 TM heterodimer that has been validated by solution NMR and disulfide crosslinking studies ¹⁸. Nonetheless, different results have been obtained for the precise sequence motif in the β3 TM helix required to maintain αIIbβ3 in its resting state. One group, using transiently transfected mammalian cells repeatedly found only one residue in β3, G708, that was functionally important for forming a TM heterodimer with αIIb ^{21; 19}. Others have observed more extended interaction interfaces ^{14; 22}. Further, although the sequence and structural motifs required for homodimerization of the αIIb TM helix have been identified ^{16; 11}, the sequence and structural motifs required for the weaker homooligomerization of the β3 TM helix have not yet been determined ¹⁸. We have prepared a series of well-characterized cell lines stably expressing mutants of the αIIbβ3 TM helices and evaluated their ability to bind the activation-specific and physiologically-relevant ligand fibrinogen, as well as cluster under activating and non-activating conditions. We found that the faces of the β3 TM helix mediating its heteromeric and homomeric interactions are similar and consist of a surface composed largely of hydrophobic residues. Nonetheless, in contrast to the αIIb TM helix, the β3 TM helix has a substantially lower intrinsic affinity for association ¹⁸. Thus, it relies on a highly specific interaction with αIIb to stabilize the αIIbβ3 TM heterodimer.

Results

Identification of the motif mediating the homomeric association of the β3 TM helix

Proteins containing the β3 TM helix form homooligomers when dispersed in phospholipid micelles ⁸. To identify the residues mediating this interaction, we scanned the helix with leucine, alanine, valine, and isoleucine mutations and measured the effect of each mutation on homodimer formation using the TOXCAT assay ²³. In TOXCAT, the homomeric association of a fusion protein consisting of an N-terminal ToxR′ DNA binding domain and a C-terminal MBP domain in the inner membrane of E. coli is driven by the intervening β3 TM, resulting in the transcriptional activation of a chloramphenicol acetyl transferase (CAT) reporter gene. The amount of CAT synthesized is a measure of the strength of TM-mediated dimerization ²³.

An important determinant for the tendency of a fusion protein to dimerize in TOXCAT is the length of the inserted TM helix ²⁴. We established the optimal length for the inserted β3 TM helix by incrementally deleting one residue from its C-terminal end (Fig. 1A). Importantly, decreasing the length of the β3 TM helix had no effect on fusion protein expression in E. coli (Fig. 1B). Moreover, the ability of each construct to facilitate growth of an MBP-deficient strain of E. coli when maltose was the sole carbon source (MalE complementation) indicated that the MBP domain of each fusion protein was located in the bacterial periplasmic space (Fig. 1C). Nonetheless, despite proper insertion of each fusion protein in the E. coli inner membrane, there were substantial differences in CAT synthesis (Fig. 1D). Because the fusion protein containing β3 residues 693–714 generated the greatest CAT signal, this construct was used in subsequent experiments.

Fig. 1.

β3 TM domain expression in MalE-deficient MM39 E. coli. (a) The β3 TM domain sequences tested in the TOXCAT assay are aligned and identified by their starting and ending residues in the mature protein. TM sequences of WT-GpA and GpA-G83I, positive and negative controls for the TOXCAT assay, are shown at the top. (b) Anti-MBP immunoblots of proteins from E. coli transformed with plasmids containing cDNA for the various length β3 TM domains shown in A. (c) Growth of MalE-deficient MM39 E. coli transformed with plasmids containing β3 TM domain cDNA on M9 minimum medium where maltose is the only carbon source. Only cells expressing MBP in the periplasm survive. E. coli transformed with pMAL-p2 and pMAL-c2 vectors expressing MBP in the periplasm or cytoplasm were included as positive and negative controls, respectively. (d) CAT expression induced by the various length β3 TM domains shown in A, expressed as a percent of the CAT induced by WT GpA (100%). Data shown are the mean and S.D. of triplicate measurements.

Most often, we found that mutations in the β3 TM helix perturbing CAT synthesis were disruptive and decreased CAT synthesis (Fig. 2). Occasionally, CAT synthesis was increased, notably by mutations at L697. Immunoblots of bacterial lysates for MBP revealed that none of the mutations had a substantial effect on expression of chimeric proteins (data not shown).

Fig. 2.

Effect of scanning mutagenesis of the β3 TM domain on CAT expression in TOXCAT. Single-site Leu (red), Ala (pink), Val (gray), or Ile (blue) mutations were made at consecutive positions in the β3 TM domain. Their effect was quantified by CAT-ELISA and compared to the signal from GpA-WT. The horizontal line at 40% corresponds to CAT synthesis by WT β3. Data presented are the mean and S.D. of 4–10 independent experiments

To quantify results shown in Fig. 2, we calculated a Perturbational Index (P_i) as a measure of the overall change in CAT activity at each position in the β3 helix due to mutation, regardless whether the mutant increased or decreased activity (Eq. 1). A plot of P_i for a given residue versus its position along the TM helix revealed peaks at approximately every four residues (peaks were observed at V700, M701; A703, I704, L705; G708, L709; L712, L713) (Fig. 3A). We also found the data well described by a sine wave with a repeat unit of 4.0 ± 0.1 residue, significantly greater than the 3.6 residue/turn period of an ideal α helix. A period greater than 3.6 residues in such an analysis typically indicates that residues important for the interaction lie in a stripe at a small angle relative to the helix axis (Fig. 3B). Thus, a second helix would be expected to interact with this surface with a right-handed crossing angle of approximately 40° (Fig. 3C) ²⁵.

Fig. 3.

A. Perturbation index (P_i) for β3 residues 697–714. TOXCAT results shown in Fig. 2 were used to calculate a P_i as described in Materials and Methods. The red line is the best fit of the data to a sinusoidal function as described in Materials and Methods. B. Ribbon diagram showing that the arrangement of sidechains at positions i, i+4, i+8… gives rise to a ridge of sidechains spiraling in a right-handed manner along an α-helix. C. A second helix (shown in gray) will have a right-handed crossing angle when docked against the first helix.

Effect of mutations on the heteromeric interactions of the β3 TM helix

TM domain separation causes αIIbβ3 activation which can be detected by measuring fibrinogen binding to the activated integrin. Previously, we have found that mutations perturbing the homomeric interaction of the αIIb TM helix in TOXCAT also caused fibrinogen binding when introduced into the full-length integrin ¹¹. Accordingly, to identify motifs mediating heteromeric interactions of the β3 TM helix, we introduced disruptive and non-disruptive β3 TM mutations identified in TOXCAT into full-length β3 and measured their effect on fibrinogen binding to αIIbβ3 stably expressed in CHO cells. Disruptive mutations, i.e., those decreasing CAT synthesis in TOXCAT, induced αIIbβ3 activation (Fig. 2, Fig. 4A). For example, replacing I704 with Leu, Ala, or Val produced a 6.8-fold increase in constitutive fibrinogen binding relative to WT, whereas mutating S699 had little effect on CAT synthesis or fibrinogen binding. Overall, we found that mutating β3 residues V696, L697, V700, M701, A703. I704, L705, G708, L712, and L713 caused constitutive αIIbβ3 activation, but to varying degrees. We then fit the fibrinogen binding results to a continuous sine function. The period of the sine wave (4.0 ± 0.2 residue) was identical to the sine wave fit to the TOXCAT results, as was its phase (Fig. 4B). Thus, β3 residues forming the interface of the αIIbβ3 TM heterodimer are predominantly hydrophobic amino acids with large side chains and lie along the same face of the β3 TM helix that mediates its homomeric interactions.

Fig. 4.

Fibrinogen binding to CHO cells expressing αIIbβ3 containing β3 TM domain mutations. A. Alexa 488-fibrinogen binding to αIIbβ3 was measured by flow cytometry in the absence (constitutive binding) or presence (stimulated binding) of 5 mM DTT. Fibrinogen binding measurements were used to calculate a binding index for each mutation as described in Materials and Methods. Data shown are the mean and SE of 3–10 measurements. B. Comparison of the fibrinogen binding data shown in A (solid lines) and a mean perturbation index calculated from mutational analyses of β3 TM helix binding to αIIb and β1 TM helix binding to α2 and α5 reported in reference ¹⁸ (dashed lines) with a sinusoidal function with a 4.0 residue periodicity fit to the fibrinogen binding data (red line).

The β3 TM mutations causing constitutive fibrinogen binding to αIIbβ3 are also found to cluster in the interface of the recently described NMR structures of the αIIbβ3 TM domain heterodimer solved in bicelles ²² or organic solvent ²⁶. The involved residues include V696, L697, V700, M701, A703, I704, L705, G708, and L712. We also found L713 to be important for dimerization, although it is not present in the heterodimer interface in these structures. However, this region of the β3 TM helix represents a location at which the two NMR structures differ. Disulfide cross-linking between engineered Cys residues has also been used to probe the geometry of the αIIbβ3 TM interface ¹⁹. Reasonably strong cross-linking to αIIb residues was observed not only for β3 L712, present in the interface of the NMR structures, but also for L713. Specifically, β3 L713C and L712C crosslink with similar efficiency to αIIb residues L980C, L983C and V984C, and L913C. These data support our observation that β3 L713 contributes to interactions with αIIb in at least one conformation of resting αIIbβ3.

We have also compared the effect of β3 mutations on fibrinogen binding to full-length αIIbβ3 with the effect of single-site β3 and β1 TM helix mutations on the heteromeric interaction of the helices on their complementary α subunit TM helices using a dominant negative (DN) bacterial reporter assay ¹⁸. As shown in Fig. 4B, the mean P_i profile for the DN assays is very similar to the effect of β3 mutations on fibrinogen binding to αIIbβ3. The phase and period of the curves are identical within experimental error and highlight the importance of residues spaced at 4-residue intervals, including V696, V700, I704, G708, and L712. Nonetheless, there are significant differences in the effects of neighboring residues: M701A and L705A show large effects on fibrinogen binding, but smaller effects in the DN assay; the opposite is the case for V700A. Interestingly, mutations of β1 residue V717, the residue corresponding to M701 in β3, have a large effect on the heteromeric interaction of β1 with the α5 TM helix, but a smaller effect on its heteromeric interaction with the α2 TM helix. Similarly, mutation of β3 residue 703 and the corresponding β1 residue had variable effects in the bacterial assay, depending on its binding partner.

It is also instructive to compare the effects of mutations in the homomeric TOXCAT assay and the heteromeric DN assay. The effects of mutations in the homomeric assay are generally smaller than in the heteromeric assay, as determined by the magnitude of the P_i. Although the assays are not identically configured, this finding suggests that the heteromeric interactions are more specific, in the sense that they are much more sensitive to mutation. Moreover, some residues, such as β3 A711, that markedly decrease CAT synthesis in TOXCAT had little effect on intact αIIbβ3, as assessed by either the DN bacterial or mammalian CHO cell assay. Thus, the heteromeric and homomeric interfaces are similar, but not identical.

Effect of β3 TM helix mutations on FAK phosphorylation

Substantial differences have been reported in the ability of αIIb and β3 TM domain mutations to activate αIIbβ3 ^{19; 21}. To provide independent confirmation of the results shown in Fig. 4, we tested the ability of various β3 TM domain mutations to cause focal adhesion kinase (FAK) phosphorylation. Ligand binding to activated αIIbβ3 is followed by the formation of αIIbβ3 clusters that initiate downstream signaling events, including phosphorylation of FAK on tyrosine residues ²⁷. Previously, we found that causing αIIbβ3 activation by mutating the αIIb and β3 TM helices was accompanied by persistent FAK phosphorylation, implying that activated αIIbβ3 was clustered ^{10; 11}. Here, we measured FAK phosphorylation when CHO cells expressing various β3 TM domain mutants were adherent to fibrinogen or re-suspended in buffer. As shown in Fig. 5, there was concordance between the ability of β3 domain mutations to induce fibrinogen binding to αIIbβ3 and cause FAK phosphorylation. Thus, when transfected CHO cells were adherent to fibrinogen, FAK was phosphorylated in cells expressing WT αIIbβ3 and various activating and non-activating β3 TM helix mutations. However, when CHO cells were re-suspended in buffer, FAK was no longer phosphorylated in cells expressing WT αIIbβ3 and the non-activating mutation A711L, phosphorylation was present, but decreased, in cells expressing L705A and L712A, and was present and unchanged in cells expressing A703L, I704L, G708A, and L713A. Further, in results not shown in the Fig. 5, there was no persistent FAK phosphorylation in re-suspended cells when S699 was mutated to Leu and the persistent FAK phosphorylation caused by L697A and L698L was comparable to that of L705A. Because FAK phosphorylation results from clustering of activated integrins, these results provide confirmatory evidence that mutations along the interface of the β3 TM helix with αIIb induce αIIbβ3 activation.

Fig. 5.

FAK phosphorylation in CHO cells expressing WT αIIbβ3 and αIIbβ3 containing β3 TM mutations. Phosphotyrosine (pTyr) immunoblots of FAK from lysates of CHO cells maintained in suspension (top row) or adherent to immobilized fibrinogen (third row). Total FAK in the immunoprecipitates, detected by an anti-FAK antibody, is shown in the second and fourth rows.

Visualization of β3 TM helix oligomerization interfaces

To visualize the interfaces on the β3 TM helix responsible for its homomeric and heteromeric interactions, the functional consequences of the various β3 TM helix mutations were divided into equal quartiles corresponding to marked, intermediate, minimal, and no effect and were displayed as such on an α-helix. The residues responsible for homomeric and heteromeric association of the helix, i.e., residues present in the marked and intermediate quartiles, were located on the same face of the helix (Fig. 6). Residues mediating heteromeric association were clustered on the helical face, whereas residues mediating homomeric association were more widely distributed, suggesting less specific interaction and possibly even a dynamic ensemble of associated states in which different conformers have different sensitivity to mutation. We also observed notable differences between the interfaces. For example, mutating G708 had a marked effect on fibrinogen binding, but was only minimally disruptive in TOXCAT. Conversely, mutating residues at the C-terminus of the helix had substantial disruptive effects in TOXCAT, but did not induce constitutive fibrinogen binding. It is also noteworthy that while the β3 helix contains a small residue-X₃-small residue motif (S699-X₃-A703), this motif is not located in either the hetero- or homooligomer interface, but on the opposite side of the helix where it would be available to possibly mediate other intermolecular interactions.

Fig. 6.

Models of the β3 TM helix showing the location of residues participating in homomeric and heteromeric interactions. Models were constructed using PyMol (DeLano Scientific) and are based on the NMR structure reported by Lau et al ³⁵. Residues along the β3 helix are depicted as colored spheres whose size reflects the magnitude of the disruptive effect of mutations on either homomeric and heteromeric β3 helix interactions. Red, yellow, green, and blue spheres indicate marked, intermediate, minimal, and no effect, respectively. Gray spheres indicate positions in the helix that were not studied because mutants at these position expressed poorly in CHO cells.

Discussion

In this paper, we have addressed the features of TM helix sequences that encode for interaction specificity versus affinity. Previous studies of water-soluble coiled-coil domains, such as those of Fos/Jun, found that buried hydrophobic residues at positions “a” and “d” of their heptad repeats provided a strong but non-specific driving force for oligomerization that did not readily distinguish between homomeric and heteromeric association ¹. For example, Fos and Jun preferentially associated heteromerically using electrostatic interactions to strongly destabilize homomeric association and weakly stabilize heteromeric association. Similar effects have been observed for a large number of other members of the leucine zipper family of transcriptional factors which use a combination of sidechain packing geometry, hydrogen-bonding, and electrostatics to provide specificity ²⁸. Here, we asked whether similar effects might be observed in TM helices, although the nature of the interactions would be different.

The homomeric and heteromeric interactions of the αIIb TM domain have been characterized previously ^{11; 16} and the results extended to other integrin α subunits ¹⁸. The αIIb TM domain contains a strongly interactive G-X₃-G motif that mediates homomeric and heteromeric interactions. Nonetheless, αIIb TM-containing heterodimers and αIIb TM homodimers are geometrically distinct and respond differently to mutations. When the αIIb TM domain interacts heteromerically with β3, its G-X₃-G motif packs onto a groove formed by larger residues on the β3 helix such that all mutations in αIIb along this interface disrupt the heterodimer ¹⁸. The G-X₃-G motif in αIIb also mediates a glycophorin A-like homomeric interaction ¹⁶. But in this case, mutations extend the motif to make it more glycine zipper-like ²⁹ and enhance, rather than diminish, its homomeric affinity ¹⁶. Thus, the divergent effects of mutations on homomeric and heteromeric interactions indicate that the native αIIb TM sequence is optimized to interact heteromerically with β3. However, whether homomeric interactions of αIIb, or β3 for that matter, participate in αIIbβ3 function in vivo remains an open question.

Here we provide the first comprehensive comparison of homomeric versus heteromeric interactions of the β3 TM helix. Previous studies using several different bacterial reporter systems found that the homomeric and heteromeric interactions of the isolated β3 TM helix are substantially weaker than those of the αIIb helix ^{17; 18}. Nonetheless, heteromeric association of β3 with αIIb is sufficiently strong to maintain αIIbβ3 in its resting conformation until circulating platelets encounter vascular damage. Although like αIIb, the β3 TM helix contains a small-X₃-small motif, S699-X₃-A703, this motif appears less prone to association than the canonical G-X₃-G motif. This is born out in our studies showing that S699-X₃-A703 is not involved in either homomeric or heteromeric β3 TM helix interactions. Investigation of heteromeric β3 TM helix interactions by mutagenesis of the full-length integrin revealed only activating (disruptive) mutations. By contrast, the TOXCAT assay revealed that mutations can either enhance or disrupt homomeric interactions. Taken together, these observations suggest that the αIIb and β3 TM sequences have evolved to be a “matched set” and that mutations to either interaction interface destabilize their association.

From a global perspective, the same face of the β3 helix is used for homomeric and heteromeric TM helix interactions, as is evident from the conserved phase of the sine waves in Fig. 3A and 4B. However, there are substantial differences in the essential residues for each type of interaction (Fig. 6). Near the N-terminus of the β3 TM helix, the sidechains of L697, M701, and I704 pack against the G-X₃-G motif of αIIb. Furthermore, G708 forms a geometrically complementary interaction with large residues in the αIIb helix; hence, these β3 residues are found to be particularly important for heterodimer formation (Fig. 6). On the other hand, G708 is substantially less important for interaction in the homo-oligomer as assessed by Ala or Leu mutations, although replacement of G708 with Asn stabilizes trimer formation in vitro ¹⁰. This finding is consistent with the observation that Asn residues can stabilize oligomerization in membranes through hydrogen-bond formation ^{30; 31}. Also, residues C-terminal to G708 are more important for homo-oligomerization rather than heterodimer formation.

In conclusion, this work provides a comprehensive mutational analysis of homomeric versus heteromeric TM helix interactions in biological membranes. The αIIb TM sequence drives strong homomeric and heteromeric interactions, mediated by its G-X₃-G motif ^{17; 18}. But whether it interacts homomerically versus heteromerically appears to be defined by the heteromeric pre-association of the αIIbβ3 extracellular domains, as well as weaker interactions between its cytoplasmic tails. The β3 TM sequence, with a lower intrinsic affinity for association, relies primarily on highly specific interactions with αIIb to stabilize the heteromeric αIIbβ3 complex. The use of the less “sticky” β3 partner leads to a heterodimer complex that has reduced affinity when compared to idealized homodimers such as that of glycophorin A. It is likely that this feature is essential for switchable systems such as integrins that rely on dynamic TM helix separation as part of their activation mechanism.

Materials and Methods

Vectors and Strains

Expression vectors pccKAN, pccgpA-wt, and pccgpA-G83I and E. coli strain MM39 ²³, were kindly provided by Dr. Donald M. Engelman. The EcoRV restriction site between the ToxR TM region and the malB gene in pccKAN was changed to a BamHI site. Fragments encoding the β3 TM domain, flanked by NheI and BamHI sites, were amplified from full-length β3 cDNA and ligated in frame into the vector. Point mutations in the β3 TM domain were generated using a QuikChange mutagenesis kit (Stratagene). Sequences of the wild-type (WT) and mutant β3 TMs were confirmed by DNA sequencing.

Expression of a Chimeric Protein in E. coli

E. coli strain MM39 was transformed with plasmids containing cDNA for the WT β3 TM domain or various β3 TM domain mutants. A single colony was used to inoculate 5 ml of LB broth containing 100 μg/mL ampicillin and grown to OD₆₀₀ of 0.6 at 37 °C with vigorous shaking. One mL of the culture was chilled on ice, pelleted, and resuspended in 100 μL of LDS sample buffer for analysis by SDS-PAGE (Invitrogen). Eighteen μL of the bacterial lysate in LDS buffer was then loaded onto 10% NuPAGE Bis-Tris pre-cast gels in 3-[N-morpholino]propanesulfonic acid buffer (Invitrogen). The separated proteins were transferred to nitrocellulose and immunoblotted with an anti-MBP monoclonal antibody (Sigma).

MalE Complementation Test

Glucose in M9 minimal medium plates was replaced by 0.4% maltose as the sole carbon source ²³. Transformed MM39 cells were streaked onto these plates containing ampicillin and incubated for two days at 37 °C.

TOXCAT Assay

The TOXCAT assay was performed as reported previously ²³. Briefly, one ml of bacterial culture was grown to an OD₆₀₀ of 0.6, chilled on ice, and pelleted at 4 °C. The pelleted cells were resuspended and washed in 0.6 mL TBS buffer (20 mM Tris, pH 8, containing 100 mM NaCl and 2 mM EDTA). Upon resuspension in 0.5 mL TBS on ice, the cells were treated with 100 μg/ml lysozyme for 20 minutes. Triton X-100 (0.4%), 10 mM dithiothreitol, and bacterial protease inhibitor cocktail (Sigma) were then added to the mixture. After a 30 min incubation, the mixture was sonicated briefly and stored overnight at 4 °C. On the following day, supernatants were assayed for chloramphicol acetyl transferase (CAT) synthesis using a CAT-ELISA kit (Roche). Assay calibration was carried out using standards provided by the manufacturer. In each experiment, the strongly dimerizing TM domain of glycophorin A (GpA)-WT, and the weakly-dimerizing TM domain of the GpA mutant G83I were included as positive and negative controls, respectively. CAT-ELISA results were expressed as a percentage of CAT synthesis induced by GpA-WT in the same experiment. Chimeric protein expression was quantified from immunoblots using a Personal Densitometer SI (Molecular Dynamics) and used to normalize CAT expression by the various constructs.

Analysis of TOXCAT Results

A perturbation index P_i for β3 TM mutations was calculated using the formalism of Treutlein et al. ³² and was based on exhaustive mutagenesis of the GpA dimer interface by Lemmon et al ³³. The P_i reflects the mean fold-change in CAT activity measured for each β3 TM domain mutation relative to GpA. The fold change is calculated using the equation:

$$P_i=10exp\left[\frac{1}{n}\sum _{j=1}^n\mid log(x_j/x_{WT})\mid \right]$$ (Eq. 1)

where n is the number of mutations at a given position i, and x_j is the TOXCAT activity at position i for mutation j. The helical phases for β3 TM mutations were then determined by fitting P_i as a function of position to a sine function:

$${(P_i)}_x=A·sin\left(\frac{2\pi (x+\phi )}{n}\right)+B$$ (Eq. 2)

in which x is the residue number, φ is the phase (in residues), A is the amplitude, n is the period in residues, and B is a constant ¹⁶.

Stable Expression of αIIbβ3 Mutants in CHO Cells

αIIb and β3 cDNAs were subcloned into the expression plasmids pcDNA3.1(+)-Neo and pcDNA3.1(+)-Zeo, respectively, prior to co-transfection into CHO cells using FUGENE 6 (Roche). Transfected cells were grown in selection medium containing G418 and Zeocin for three weeks and then sorted twice by fluorescence-activated cell sorting (FACS) for cells expressing high levels of αIIbβ3 ¹¹.

Fibrinogen Binding to CHO Cells Expressing αIIbβ3

Fibrinogen binding to αIIbβ3 expressed on the CHO cell surface was measured as described previously ¹¹. Briefly, CHO cells (2×10⁶ cells/ml) were incubated with the β3-specific monoclonal antibody (mAb) SSA6 ³⁴ on ice for 30 minutes. Labeled cells were washed and incubated for 30 min at 37 °C with phycoerythrin-conjugated anti-mouse IgG (Molecular Probes), 200 μg/ml fibrinogen conjugated with Alexa 488 (Molecular Probes), freshly-made 5 mM dithiothreitol (DTT) ± 5 mM EDTA. The cells were washed, fixed with 0.37% formalin in PBS, and examined by 2-color FACS analysis. Fibrinogen specifically bound to αIIbβ3 was defined as fibrinogen binding inhibited by EDTA ¹¹.

Specific fibrinogen binding data were used to calculate a fibrinogen binding index (F_i), using the equation:

$${\text{F}}_{\text{i}}=({\text{FB}}_{\text{c}}\text{mut}/{\text{FB}}_{\text{DTT}}\text{mut})/{\text{FB}}_{\text{c}}\text{wt}/{\text{FB}}_{\text{DTT}}\text{wt}$$ (Eq. 3)

where FB_cmut represents constitutive fibrinogen binding to αIIbβ3; FB_DTTmut, fibrinogen binding to mutant αIIbβ3 induced by 5 mM DTT; FB_cwt, constitutive fibrinogen binding to wild-type αIIbβ3; and FB_DTTwt, fibrinogen binding to wild-type αIIbβ3 induced by 5 mM DTT.

Focal adhesion kinase (FAK) phosphorylation

Phosphorylation of FAK in CHO cells stably expressing αIIbβ3 and either adherent to fibrinogen-coated tissue culture plates or placed in suspension in centrifuge tubes was measured as described previously ^{9; 16}. Briefly, adherent cells or cells resuspended in 20 mM HEPES buffer, pH 7.4, containing 137 mM NaCl, 2.7 mM MgCl₂, 5.6 mM glucose, and 3.3 mM NaH₂PO₄ were lysed with RIPA buffer (10 mM Tris_HCl, pH 7.2, containing 158 mM NaCl, 2 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, as well as 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, and protease inhibitor cocktail (Sigma). FAK was immunoprecipitated from 500 μg of cellular lysate and 300 μg of immunoprecitated protein was subjected to SDS-PAGE in a 3–8% NuPAGE Tris-Acetate gel (Invitrogen). FAK phosphorylation was detected by immunoblotting using the mouse anti-phosphotyrosine mAb 4G10 (Upstate Biotech). The presence of comparable amounts of FAK in each lysate was confirmed by immunoblotting 100 μg of lysate protein with the anti-FAK monoclonal antibody 4.47 (Millipore).

Abbreviations used

TM

transmembrane

CHO

Chinese hamster ovary

MBP

maltose binding protein

CAT

chloramphenicol acetyl transferase

DTT

dithiothreitol

P_i

perturbation index

F_i

fibrinogen binding index

FAK

focal adhesion kinase