A functional comparison of mutations in integrin β cytoplasmic domains: effects on the regulation of tyrosine phosphorylation, cell spreading, cell attachment and β1 integrin conformation

Integrins are α/β heterodimeric receptors that mediate cell adhesion by forming a transmembrane linkage between specific extracellular matrix or basement membrane components and the actin cytoskeleton (Hynes, 1992EF17). Integrin engagement activates signaling pathways that regulate the adhesion process itself and adhesion-dependent processes including cell migration, proliferation, survival and differentiation. These signals include the activation of tyrosine kinases, serine-threonine kinases, phosphoinositide kinases and small GTP binding proteins (Clark and Brugge, 1995EF8; Burridge and Chrzanowska-Wodnicka, 1995EF5; Schwartz et al., 1995EF38; Yamada and Miyamoto, 1995EF44; Giancotti and Ruoslahti, 1999EF11). Integrin β cytoplasmic domains play important roles in many aspects of integrin function, including the activation of signaling pathways, the regulation of integrin-ligand interactions, cell attachment and cell spreading (Sastry and Horwitz, 1993EF35; LaFlamme et al., 1997EF21; Hughes and Pfaff, 1998EF16). For example, integrin β cytoplasmic domains are required for the activation of FAK tyrosine phosphorylation, which is an early signaling event during cell adhesion (Guan et al., 1991EF12). Clustering isolated β1 and β3 cytoplasmic domains expressed as interleukin 2 (IL-2) receptor (tac) chimeras on the cell surface is sufficient to activate the tyrosine phosphorylation of FAK (Akiyama et al., 1994EF1; Lukashev et al., 1994EF26; Tahiliani et al., 1997EF41). In addition, these tac-β tail chimeras function as dominant negative inhibitors of specific β1 integrin conformations, as well as β1 integrin function in cell attachment and cell spreading, presumably by titrating β cytoplasmic domain-binding proteins from endogenous integrin β cytoplasmic domains (LaFlamme et al., 1994EF20; Smilenov et al., 1994EF40; Mastrangelo et al., 1999bEF29; Berrier et al., 2000EF4). These results suggest that β cytoplasmic domains regulate these processes by interacting with cellular proteins; however, the protein interactions required for β cytoplasmic domain function in these processes are not known.

Integrin β1 and β3 cytoplasmic domains share a significant degree of amino acid homology, and several proteins such as talin, α-actinin, FAK, paxillin, filamin and the integrin-linked kinase (ILK) can bind to both the β1 and β3 tails (Dedhar and Hannigan, 1996EF10; Hemler, 1998EF13; Liu et al., 2000EF24). One or more of these proteins may interact with β cytoplasmic domains to regulate integrin function. Interestingly, endogenous β1 integrin conformation and function in cell attachment and cell spreading can be inhibited by both tac-β1 and tac-β3 chimeras, suggesting that these processes require at least one protein that can bind to both the β1 and the β3 cytoplasmic domains (LaFlamme et al., 1994EF20; Mastrangelo et al., 1999bEF29; Berrier et al., 2000EF4). Proteins that specifically interact with either the β1 or β3 cytoplasmic domain have also been identified; integrin cytoplasmic domain-associated protein-1 (ICAP-1) and β3-endonexin have been shown to bind specifically to the β1 and β3 cytoplasmic domains, respectively (Chang et al., 1997EF7; Zhang et al., 1999EF47; Shattil et al., 1995EF39). Thus, it is possible that the β1 and β3 cytoplasmic domains may regulate similar aspects of integrin function by interacting with proteins that can associate with both β1 and β3 tails, as well as proteins that associate specifically with only the β1 or β3 cytoplasmic domain.

Because integrin β cytoplasmic domains are small (∼50 amino acids) and are capable of binding to a number of different proteins, it is likely that these proteins have overlapping binding sites. For this reason, we generated a series of substitution mutants to ask whether distinct amino acid residues are involved in the ability of the β cytoplasmic domain to regulate specific aspects of integrin function. Our goal was to identify mutations that can inhibit some aspects of β cytoplasmic domain function and not others, so that in future studies these mutations could be used to identify the protein interactions pertinent to individual aspects of integrin function. Our experimental approach was to express wild-type and mutant β cytoplasmic domains in the context of tac chimeras and to compare the ability of these chimeras to activate tyrosine phosphorylation and to inhibit β1 integrin conformation and function in cell attachment and spreading. With this approach, we demonstrate that the β1 and β3 cytoplasmic domains are sufficient to activate the tyrosine phosphorylation of p130CAS and paxillin in addition to FAK. We also report three categories of β cytoplasmic domain mutants: (1) mutants that behaved like wild-type tac-β1 and tac-β3 in terms of their ability to activate tyrosine phosphorylation, to regulate β1 integrin conformation and to inhibit endogenous integrin β1 function in cell attachment and spreading; (2) mutants that inhibited tac-β1 or tac-β3 function in all of these assays; and (3) mutants that inhibited tac-β1 or tac-β3 function in only a subset of assays. For example, we identified mutations in the WDT motif that inhibited the ability of tac-β1 to regulate β1 integrin conformation and function in cell spreading, but had little effect on cell attachment. We also identified mutations in the membrane-proximal region of the β3 cytoplasmic domain and the C-terminal region of the β1 cytoplasmic domain that greatly diminished the ability of tac-β1 and tac-β3 to inhibit cell spreading, but had little effect in other assays. Interestingly, these latter mutations were previously shown by other laboratories to inhibit the interaction of FAK, paxillin and ICAP-1 with β cytoplasmic domains (Schaller et al., 1995EF37; Chang et al., 1997EF7). Future studies will examine the role of these proteins in mediating β cytoplasmic domain function in cell spreading.

Human foreskin fibroblasts (Vec Technologies) or rat embryonic fibroblasts, REF52s (kindly provided by Chris Turner), were maintained in DMEM supplemented with 10% fetal bovine serum and 2 mM L-glutamine. The A5 and ETC12 CHO cell lines expressing recombinant αIIbβ3 and αIIbβ3▵727, respectively (generously provided by Mark Ginsberg), were maintained in DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine and MEM nonessential amino acids. Transient transfections were performed by electroporation as previously described (Mastrangelo et al., 1999aEF28).

The construction of plasmids encoding chimeric receptors containing the extracellular and transmembrane domains of the tac subunit of IL-2 receptor connected to the wild-type β1, β3, β3B, and β4 cytoplasmic domains has been described previously (LaFlamme et al., 1992; LaFlamme et al., 1994; Homan et al., 1998). The construction of the tac chimeras containing the β3 cytoplasmic domain mutants has also been described (Tahiliani et al., 1997). Tac chimeras containing mutations in the β1 cytoplasmic domain were constructed as outlined below. Site-directed mutagenesis was performed by PCR using wild-type tac-β1 chimera as template DNA, except where indicated. The following mutants were constructed by one-stage PCR: β1-758, 760 (^*/A) using the forward primer 5′-CTCATCTGGAAGCTTTTAATGATAATTGCTGACGCAAGGGAGTTTGCTAAA-3′, and the reverse primer 5′-TTACCTTAGAGCTTTAAATC-3′ (SH17); β1-783,795 (Y/F) using the forward primer 5′-CCATGGAGACGTCCA-3′ (SP3) and the reverse primer 5′-GAGTCACTCGAGTCATTTTCCCTCAAACTTCGGATTGACCACAGTTGTTACGGCACTCTTAAAAATAGGATTTTCACC-3′; β1-787-789 (^*/A), 793 (P/I), using SP3 as the forward primer and 5′-GAGTCACTCGAGGTCATTTTCCCTCATACTTGATATTGACCACAGCTGC-3′ as the reverse primer, and the β1-787-789 (^*/A) mutant (described below) as the template. The following mutants were constructed by two-stage PCR using internal overlapping primers, together with the outside primers SP3 and SH17. The internal overlapping primer pairs were as follows: for β1-783 (Y/A), 5′-GGTGAAAATCCTATTGCTAAGAGTGCCGTAACA-3′ and 5′-TGTTACGGCACTCTTAGCAATAGGATTTTCACC-3′; for β1-763,765 (^*/A), 5′-CATGACAGAAGGGAGGCTGCTGCATTTGAAAAGGAGAAAATG-3′ and 5′-TTTCTCCTTTTCAAATGCAGCAGCCTCCCTTCTGTCATG-3′; for β1-775,776 (^*/A), 5′-AAAATGAATGCCAAAGCGGCCACGGGTGAAAATCCT-3′ and 5′-AGGATTTTCACCCGTGGCCGCTTTGGCATTCATTTTCTC-3′; for β1-775,777 (^*/A), 5′-AAAATGAATGCCAAAGCGGACGCGGGTGAAAATCCTATT-3′ and 5′-AATAGGATTTTCACCCGCGTCCGCTTTGGCATTCATTTTCTC-3′; for β1-787-789 (^*/A), 5′-ATTTATAAGAGTGCCGCTGCAGCTGTGGTCAATCCGAAG-3′ and 5′-CTTCGGATTGACCACAGCTGCAGCGGCACTCTTATAAATAGG-3′; for β1-787 (^*/A), 5′-ATTGACCACAGTTGTTGCGGCACTCTTATAAAT-3′ and 5′-ATTTATAAGAGTGCCGCAACAACTGTGGTCAAT-3′. For β1-787 (^*/A), the outside primers were 5′-GCAGTGGCCGGCTGTG-3′ and SH17. For all constructs, the resulting PCR products were digested with the restriction enzymes HindIII and XhoI and ligated in place of the HindIII/XhoI fragment of the control receptor. Plasmids were sequenced to confirm the integrity of the newly constructed mutants.

Purified 7G7B6 mouse mAb was used to coat magnetic beads conjugated with goat anti-mouse IgG (Polysciences). To activate signaling, chimeric receptors were clustered on the cell surface by incubating transfected cells with the antibody-coated magnetic beads as previously described (Akiyama et al., 1994EF1; Tahiliani et al., 1997EF41; Mastrangelo et al., 1999aEF28). Positively expressing cells were magnetically sorted and then lysed on ice for 15 minutes in mRIPA containing 50 mM Tris (pH 7.4), 1% NP-40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 μg/ml leupeptin, 2 μg/ml aprotinin, 1 mM sodium vandate and 1 mM sodium fluoride. Protein concentrations were determined using a MicroBCA assay (Pierce). Lysates were either analyzed directly by western blotting or used for immunoprecipitations. Paxillin and p130CAS were immunoprecipitated using monoclonal antibodies (mAbs) to paxillin (clone 349, Transduction Laboratories) or to p130CAS (clone 21, Transduction Laboratories). Cell lysates or immunoprecipitates were separated by 10% SDS-PAGE and analyzed on western blots with the mAb 4G10 to phosphotyrosine (Upstate Biotechnology) and visualized by enhanced chemiluminescence (Amersham). Blots were reprobed for p130CAS or paxillin as previously described (Tahiliani et al., 1997EF41). To quantitatively compare the ability of the different tac-β tail mutants to trigger tyrosine phosphorylation, the tyrosine phosphorylation signal, including signals in the MW range of FAK and p130CAS, was normalized to the amount of protein present as assayed by reprobing the blot for p130CAS or paxillin using a PSDI scanner from Molecular Dynamics and ImageQuant software.

The surface expression of the different tac chimeras was analyzed and compared by flow cytometry using mAbs specific for the tac subunit of the IL-2 receptor. Similar surface expression of the various mutant tac chimeras and the wild-type tac-β1 and tac-β3 chimera was observed by flow cytometry in several independent experiments. However, for signaling experiments, transfected cells were also routinely examined for the expression of the tac chimeras by immunofluorescence microscopy using a FITC-conjugated mAb to the tac subunit of the IL-2 receptor (Accurate Chemical and Scientific Corporation and Becton Dickinson) to ensure similar expression levels of the different tac chimeras in individual experiments.

The ability of tac chimeras containing integrin β cytoplasmic domains to inhibit the expression of the 9EG7 epitope was analyzed by two-color flow cytometry as previously described (Mastrangelo et al., 1999aEF28; Mastrangelo et al., 1999bEF29). Briefly, human fibroblasts transiently transfected with the various tac chimeras were harvested and then allowed to recover at 37°C for 15 minutes in serum-free DMEM. Each sample was stained with rat mAb, 9EG7 (Pharmingen) and then with a fluorescein-conjugated mouse anti-rat (Pharmingen) and a phycoerythrin (PE)-conjugated mouse mAb specific for the tac subunit of the IL-2 receptor (Pharmingen). The levels of expression of the 9EG7 epitope and the various tac chimeras were analyzed on individual cells using a FACScan flow cytometer (Becton Dickinson). To normalize 9EG7 expression to total β1 integrin expression, an aliquot of each sample was doubly stained with mAb K20 to the β1 subunit and a mAb to the tac subunit of the IL-2 receptor (Mastrangelo et al., 1999aEF28). The ratio of the fluorescence of 9EG7 to K20 was used to calculate the relative expression of the 9EG7 epitope ((mean fluorescence intensity of 9EG7/mean fluorescence intensity of K20) ×100). However, it is important to note that the expression of the K20 epitope on normal fibroblasts was not significantly changed by the expression of the various tac-β tail constructs (data not shown). The inhibition of expression of the 9EG7 epitope was then calculated by comparing the relative expression of the 9EG7 epitope on tac-β tail expressing cells to its expression on cells expressing the control receptor.

The ability of tac-β1 chimeras to inhibit cell attachment was analyzed as previously described (Mastrangelo et al., 1999a; Mastrangelo et al., 1999b). Human fibroblasts were transiently transfected with the various tac chimeras, harvested and allowed to recover as described above, and then 6×10⁵ cells from each transfection were allowed to attach for 30 minutes at 37°C in serum-free DMEM to a well of a six-well tissue culture plate previously coated with human fibronectin (10 μg/ml). The plates were then rotated on an orbital shaker for 30 seconds at 150 rpm and the medium containing the unattached cells was removed. Each well was gently washed twice with 0.8 ml of PBS, which was added to the unattached samples. The attached cells were then removed with trypsin/EDTA. The expression level of the various tac chimeras was then analyzed on the unattached cells, the attached cells and a sample of the starting population of cells by flow cytometry using a mAb specific for the tac subunit. To quantitatively recover and analyze unattached cells, 3×10⁵ untransfected fibroblasts were added to each sample containing the unattached cells. The expression of the tac chimeras on the attached and unattached samples was then compared to their expression on the starting population.

Transiently transfected fibroblasts were harvested and allowed to recover for 20 minutes at 37°C. Approximately 2×10⁵ cells in 2 mls of serum-free medium were plated onto glass coverslips coated with 10 μg/ml fibronectin placed in six-well tissue culture dishes. The cells were allowed to spread for 60 minutes at 37°C. Spreading assays using A5 and ETC12 CHO cell lines were performed similarly; however, cells were allowed to recover for 60 minutes and then A5 cells were plated on 15 μg/ml fibrinogen for 30 minutes and ETC12 cells were plated for 2 hours on 10 μg/ml fibronectin. The cells were then fixed and stained with a FITC-conjugated mAb to the tac subunit (Becton Dickinson). Microscopy was performed using an Olympus Provos microscope equipped with phase contrast and epifluorescence and attached to a SPOT, low light camera interfaced with a PC computer. Cell area and fluorescence intensity were measured using Image Pro-Plus software (Media Cybernetics). Round cells that had not begun to spread were found to have cell areas of less than 600 μm² (Berrier et al., 2000).

Clustering isolated integrin β cytoplasmic domains triggers the tyrosine phosphorylation of p130CAS and paxillin. Previous studies have demonstrated that clustering tac chimeras containing either the wild-type β1 or β3 cytoplasmic domain is sufficient to trigger the tyrosine phosphorylation of FAK (Akiyama et al., 1994EF1; Tahiliani et al., 1997EF41). To determine whether integrin β cytoplasmic domains are also sufficient to trigger the tyrosine phosphorylation of p130CAS and paxillin, tac chimeras containing the β1, β3, β3B or β4 cytoplasmic domain (Fig. 1) were clustered on the surface of normal human fibroblasts while in suspension. When cell lysates were prepared and analyzed for phosphotyrosine, we found that both the wild-type β1 and β3 cytoplasmic domains were sufficient to trigger the tyrosine phosphorylation of proteins that had the approximate molecular weights of 130 and 120 kDa (Fig. 1B). In comparison, the control receptor lacking a cytoplasmic domain, or chimeric receptors containing either the β3B or β4 cytoplasmic domain were incapable of triggering these same events (Fig. 1B). Previous studies had already confirmed the identity of the 120 kDa band as FAK (Akiyama et al., 1994EF1; Tahiliani et al., 1997EF41). The identity of the tyrosine-phosphorylated protein at 130 kDa as p130CAS was confirmed by immunoprecipitation after clustering tac-β1 (Fig. 1C). Similar results were obtained by clustering tac-β3 (see below). Because phosphorylated proteins with the mobility of phosphorylated paxillin (68-76 kDa) were easily observed in lysates generated from REF52 cells, we initially used these cells to analyze paxillin phosphorylation. As shown in Fig. 1D, phosphorylated paxillin was easily observed in lysates and immunoprecipitates after clustering tac-β1 on the surface of REF52 cells. In later experiments, the tyrosine phosphorylation of paxillin was also observed following clustering of either tac-β1 or tac-β3 in normal human fibroblasts (see below; Fig. 4D). Thus, clustering integrin β cytoplasmic domains is sufficient to trigger the tyrosine phosphorylation of p130CAS and paxillin, in addition to FAK.

We targeted mutations to motifs conserved in the β1 and β3 cytoplasmic domains (Fig. 2), and then compared the ability of these mutant β tails to trigger tyrosine phosphorylation. The β1-758,760 (^*/A) and β1-763,765 (^*/A) mutants targeted either the histidine (H) and arginine (R) or the phenylalanine (F) and lysine (K) residues, respectively, in the conserved membrane-proximal HDR(RorK)EFAK motif. We found that the β1-758,760 (^*/A) mutant triggered tyrosine phosphorylation to similar levels as the wild-type β1 cytoplasmic domain (Fig. 3A). Although there was variability between experiments, on average the β1-763,765 (^*/A) mutant triggered slightly reduced levels of tyrosine phosphorylation compared to wild-type β1 (Fig. 3D). The β1-775,776 (^*/A) and the β1-775,777 (^*/A) mutants with alanine substitutions in the conserved WDT motif consistently showed a reduced ability to activate tyrosine phosphorylation, with only FAK showing significant phosphorylation in cell lysates (Fig. 3A). Changing the membrane-proximal NPXY motif to NPXA in the β1-783 (Y/A) mutant abrogated the ability of the β1 cytoplasmic domain to activate tyrosine phosphorylation. The level of tyrosine phosphorylation observed was comparable to that in lysates from cells expressing the control receptor (Fig. 3A). By contrast, the substitution of phenylalanine for the tyrosine residues in both the membrane-proximal and the membranedistal NPXY motifs in the β1-783,795 (Y/F) mutant consistently resulted in only slightly reduced tyrosine phosphorylation (Fig. 3B). Thus, tyrosine phosphorylation of the β1 cytoplasmic domain itself is not required to activate these signaling events. Additionally, the β1-787-789 (^*/A) mutant containing AAA in place of the VTT motif consistently resulted in a loss of tyrosine phosphorylation, similar to the β1-783 (Y/A) mutant (Fig. 3A). Thus, it seems that both a membrane-proximal NPXY motif and VTT motif are necessary for the β1 cytoplasmic domain to activate tyrosine phosphorylation.

The results for the β3 mutants were similar to those for the β1 mutants. The β3-723,726 (^*/A) mutant, which targets the aspartic acid (D) and glutamic acid (E) residues in the conserved membrane-proximal HDR(RorK)EFAK motif, signaled as efficiently as the wild-type β3 cytoplasmic domain (Fig. 3E). Thus, the membrane-proximal region does not appear to be required for either the β1 or the β3 cytoplasmic domain to activate tyrosine phosphorylation. An alanine substitution for the tyrosine in the membrane-proximal NPXY motif in the β3-747 (Y/A) mutant abolished the ability of the β3 cytoplasmic tail to signal (Fig. 3E), whereas the more conservative substitution of phenylalanine for tyrosine in the β3-747 (Y/F) mutant did not inhibit tyrosine phosphorylation (Fig. 3E). These results are similar to those obtained for the analogous β1 mutants (Figs 2, 3). Interestingly, the β3-751-753 (^*/A) mutant containing AAA for the TST motif consistently triggered tyrosine phosphorylation, although at reduced levels compared to the wild-type β3 tail (Fig. 3E). This is different from the analogous β1-787-789 (^*/A) mutant, which lost the ability to trigger tyrosine phosphorylation. Thus, the TST motif in the β3 tail plays a more modulatory role, whereas the analogous VTT motif in the β1 tail appears to be required for the activation of tyrosine phosphorylation.

A side-by-side comparison of the signaling phenotypes of analogous β1-787-789 (^*/A) and β3-751-753 (^*/A) mutants confirmed that the β1-787-789 (^*/A) mutant was unable to activate tyrosine phosphorylation, whereas the β3-751-753 (^*/A) mutant triggered the tyrosine phosphorylation, albeit not as efficiently as the wild-type β cytoplasmic domains (Fig. 4). The differences were not time-dependent as this same pattern of tyrosine phosphorylation was observed following either 40 minutes or 90 minutes of clustering (Fig. 4A). Additionally, the surface expression of the β1-787-789 (^*/A) mutant was similar to that of the β3-751-753 (^*/A) mutant and the levels of cell-surface expression of these mutants were comparable to the levels of expression of the tac chimeras containing wild-type β cytoplasmic domains (Fig. 4B). Immunoprecipitation of p130CAS (Fig. 4C) and paxillin (Fig. 4D) also demonstrated that their phosphorylation was triggered by the β3-751-753 (^*/A) mutant, but not the β1-787-789 (^*/A) mutant (Fig. 4C,D).

The VTT motif in the β1 cytoplasmic domain is flanked by two NPXY motifs (NPIY and NPKY), whereas the TST motif in the β3 cytoplasmic domain is flanked by a NPXY motif and a NXXY motif (NPLY and NITY). Therefore, the conformation of the β1 and β3 tails may differ in the regions of the VTT and TST motifs, and this difference may be responsible for the different phenotypes of the β3-751-753 (^*/A) and β1-787-789 (^*/A) mutants. To test this hypothesis, we constructed the β1-787-789 (^*/A), 793(P/I) mutant, which contains a substitution of an isoleucine for the proline in the NPKY motif of β1. However, we found that the β1-787-789 (^*/A), 793 (P/I) mutant did not trigger tyrosine phosphorylation (Fig. 3C), suggesting that the difference between these β1 and β3 mutants is not simply the presence of a proline residue in the membrane-distal NPXY motif of the β1 tail.

In addition, as the VTT motif in the β1 cytoplasmic domain overlaps a region previously shown to be important in the binding of ICAP-1 to the β1 tail (Chang et al., 1997EF7), we tested whether a more conservative mutation, previously shown to inhibit the interaction of ICAP-1 with the β1 tail (Chang et al., 1997EF7), would inhibit signaling initiated by clustering the β1 cytoplasmic domain. We constructed the β1-787(V/A) mutant, which contains an alanine substitution for valine 787. However, we found that the β1-787 (V/A) mutant was able to activate tyrosine phosphorylation (Fig. 3C) to levels only slightly reduced compared to wild-type tac-β1. Although this result suggests that ICAP-1 binding to the β1 tail is not required to activate tyrosine phosphorylation, we have not yet tested whether this mutation affects the binding of ICAP-1 to tac-β1.

Previous studies had demonstrated that high levels of tac-β1 or tac-β3 can inhibit specific β1 integrin conformations, as well as β1 function in cell attachment (Mastrangelo et al., 1999bEF29). To determine whether distinct regions of the integrin β1 tail regulate these processes, we compared the ability of wild-type and mutant tac-β1 chimeras to regulate β1 integrin conformation and to inhibit β1 function in cell attachment. To monitor the conformation of the β1 subunit, we assayed the expression of the 9EG7 epitope (Lenter et al., 1993EF22). Previous studies indicated that the expression of the 9EG7 epitope on the β1 subunit can be regulated by divalent cations (Bazzoni et al., 1998EF2) and ligand binding and depends on the amino acid sequence of the integrin β cytoplasmic domain (Belkin et al., 1997EF3; Sakai et al., 1998EF34; Wennerberg et al., 1998EF42). In addition, the expression of the 9EG7 epitope correlates with α5β1 function in ligand binding and in some instances with integrin function in cell attachment (Mastrangelo et al., 1999bEF29; Belkin et al., 1997EF3; Bazzoni et al., 1998EF2; Sakai et al., 1998EF34; Wennerberg et al., 1998EF42).

When we compared the expression of the 9EG7 epitope in cells transiently expressing either the wild-type or mutant tac-β1 chimeras by two-color flow cytometry, we found that tac-β1 chimeras containing either the β1-758,760 (^*/A), the β1-763,765 (^*/A), the β1-787 (V/A), or the β1-783, 795 (Y/F) mutant inhibited the expression of the 9EG7 epitope similar to wild-type tac-β1 (Fig. 5A). By contrast, tac-β1 chimeras containing the either the β1-775,776 (^*/A), the β1-775,777 (^*/A), the β1-783 (Y/A), β1-787-789 (^*/A) or the β1-787-789 (^*/A), 793 (P/I) mutant showed diminished abilities to inhibit the 9EG7 expression (Fig. 5A). These results suggest that the NPXY, WDT and VTT motifs in the β1 tail are important for protein interactions with β cytoplasmic domains that regulate β1 integrin conformation. Interestingly, the same mutations affected the ability of the β cytoplasmic domain to activate tyrosine phosphorylation; however, mutations in the WDT motif had an intermediate effect on the activation of tyrosine phosphorylation (Fig. 2).

To compare the ability of wild-type and mutant tac-β1 chimeras to inhibit cell attachment on fibronectin, normal human fibroblasts transiently expressing the various tac-β1 chimeras were allowed to adhere to fibronectin for 30 minutes at 37°C. The levels of expression of the chimeric receptors on the starting, attached and unattached populations of cells were compared by flow cytometry. Inhibition of cell attachment results in an increase in the mean fluorescence intensity of chimeric receptor expression in the unattached population of cells. We found that tac-β1 chimeras containing either the β1-758,760 (^*/A), β1-763,765 (^*/A), β1-775, 776 (^*/A), β1-783,795 (Y/F) or the β1-787 (V/A) mutant inhibited cell attachment similar to wild-type tac-β1 (Fig. 5B). The tac-β1 chimeras containing the β1-775, 777 (^*/A) mutant also inhibited cell attachment, but not as dramatically. The tac-β1 chimeras containing either the β1-783 (Y/A), β1-787-789 (^*/A), or the β1-787-789 (^*/A), 793 (P/I) mutant did not inhibit cell attachment to fibronectin. These results suggest that the NPXY and the VTT motifs of the β1 tail are required for the regulation of β1 integrin conformation and for β1 cytoplasmic domain function in cell attachment (Fig. 2). The analogous motifs in the β3 tail had similar phenotypes in these assays (Mastrangelo et al., 1999b; see summary in Fig. 2). Interestingly, mutations in the WDT motif of the β1 tail that inhibited the ability of tac-β1 to regulate β1 integrin conformation had little effect on the ability of tac-β1 to inhibit cell attachment. This was especially true of the β1-775,776 (^*/A) mutant. These results suggest that distinct protein interactions are involved in the ability of the β cytoplasmic domain to regulate β1 integrin conformation and β1 function in cell attachment.

Previous studies had indicated that moderate levels of expression of either tac-β1 or tac-β3 allow cell attachment, but inhibit cell spreading (LaFlamme et al., 1994; Mastrangelo et al., 1999b; Berrier et al., 2000). To compare the ability of the wild-type and mutant tac-β1 and tac-β3 chimeras to inhibit cell spreading, we measured the extent of cell spreading as a function of the expression level of the chimeric receptor (Berrier et al., 2000). For these experiments, normal fibroblasts transiently expressing tac-β tail chimeras were plated onto fibronectin-coated glass coverslips and allowed to spread for one hour. Adherent cells were stained for tac expression. The fluorescence intensity and area of positively transfected cells were determined as described in Materials and Methods. We found that moderate levels of expression of wild-type tac-β1 inhibited cell spreading (Fig. 6) as previously reported (LaFlamme et al., 1994; Berrier et al., 2000). When we compared the ability of various tac-β1 mutants to inhibit cell spreading, we found that the β1-758,760 (^*/A) and the β1-763,765 (^*/A) mutants with mutations in the membrane-proximal motif inhibited cell spreading similar to wild-type tac-β1 (Fig. 6A). The β1-783,795 (Y/F) mutant also inhibited cell spreading, but was not as potent an inhibitor as the wild-type tac-β1 (Fig. 6A). By contrast, tac-β1 chimeras containing either the β1-775,776 (^*/A), β1-775, 777 (^*/A), or β1-783 (Y/A) mutant were each poor inhibitors of cell spreading (Fig. 6A). Thus, alanine substitutions in the WDT motif or the membrane-proximal NPXY motif of the β1 tail abrogated the ability of tac-β1 to regulate cell spreading.

In a second series of experiments, we compared the ability of tac-β3 mutants and additional tac-β1 mutants to inhibit cell spreading on fibronectin. We found that wild-type tac-β1 and tac-β3 inhibited cell spreading as expected (Fig. 6B); however, tac-β1 consistently inhibited cell spreading better than tac-β3. Tac-β1 and tac-β3 chimeras containing either the β1-787 (V/A), β1-787-789 (^*/A), β1-787-789 (^*/A), 793(P/I), β3-723,726 (^*/A), β3-751-753 (^*/A) or the β3-747 (Y/A) mutant failed to inhibit cell spreading on fibronectin. However, it is important to note that the spread cells expressing the β1-787 (V/A) mutant tended to have smaller areas than cells expressing the control receptors. These results suggest that three regions of the β cytoplasmic domain are important for regulating cell spreading: the membrane-proximal region in the β3 tail, the NPXY motif in the β1 and β3 tails and the analogous VTT and TST motifs in the β1 and β3.

The β1-787 (V/A) mutant and the β3-723,726 (^*/A) mutant are of particular interest because they behaved similar to wild-type tac-β tail chimeras in their ability to trigger tyrosine phosphorylation, regulate β1 integrin conformation and inhibit β1 function in cell attachment (Fig. 2), suggesting that these mutations inhibit protein interactions that specifically regulate cell spreading. Because the β1-787 (V/A) mutation was previously shown to inhibit the interaction of β1-tail specific binding protein ICAP-1 (Chang et al., 1997), we were interested in whether the diminished ability of tac-β1 containing this mutation to inhibit cell spreading was specific for spreading mediated by β1 integrins. To test this hypothesis, we used two CHO cell lines that stably express either wild-type integrin αIIbβ3 (A5 cells) or αIIbβ3 containing a β3 cytoplasmic domain truncation, αIIbβ3Δ727 (ETC12 cells). A5 cells adhere and spread on fibrinogen via αIIbβ3 and ETC12 cells spread on fibronectin via their endogenous α5β1 integrin (Ylanne et al., 1993). Therefore, we compared the ability of tac-β1 containing the β1-787 (V/A) mutant to inhibit cell spreading of A5 cells on fibrinogen and ETC12 cells on fibronectin. We found that the β1-787 (V/A) mutant failed to inhibit cell spreading both on fibrinogen and on fibronectin (Fig. 6C). This result suggests that either the binding of ICAP-1 to the β1 tail is required to nucleate protein interactions, which are common to both β1 and β3 and are required for cell spreading, or that the β1-787 (V/A) mutation inhibits the binding of a protein other than ICAP-1 that binds to both β1 and β3 tails to regulate cell spreading.

Although several proteins that bind to one or multiple β tails have been identified, the role of many of these protein interactions in specific aspects of integrin function remains unclear. The small size of integrin β cytoplasmic domains (∼50 amino acids) complicates matters, because proteins that interact with β tails more than likely have overlapping binding sites. For this reason, we have focussed our attention on identifying point mutations within β cytoplasmic domains that specifically affect some aspect(s) of cell adhesion, as an initial approach to identifying the relevant protein interactions. In the present study, we expressed wild-type and mutant β cytoplasmic domains in the context of tac chimeras and compared their ability to activate tyrosine phosphorylation, regulate β1 integrin conformation and inhibit integrin function in cell attachment and cell spreading. Although we identified several mutations that inhibited β cytoplasmic domain function, only a few were specific for individual aspects of cell adhesion. We identified mutations in the membrane-proximal region of the β3 tail and in the C-terminal region of the β1 tail that specifically affected the ability of tac-β1 and tac-β3 to inhibit endogenous integrin function in cell spreading. We also identified a mutation in the WDT motif of β1 that inhibited the ability of tac-β1 to regulate β1 integrin conformation and function in cell spreading. In addition, we extended our previous studies and demonstrated that protein interactions with the wild-type β1 and β3 cytoplasmic domains are sufficient to trigger the tyrosine phosphorylation of paxillin and p130CAS in addition to FAK. Although we identified a number of mutations that inhibited the ability of β cytoplasmic domains to trigger tyrosine phosphorylation, we have yet to identify mutations that specifically inhibit tyrosine phosphorylation and not other β tail-dependent phenotypes. This was also true for mutations in the β1 cytoplasmic domain that inhibited the ability of tac-β1 to regulate cell attachment.

Interestingly, the two mutations that specifically abrogated the ability of tac-β1 and tac-β3 to inhibit cell spreading were previously demonstrated to inhibit known protein interactions with β cytoplasmic domains. The β3-723, 726 (^*/A) mutation had previously been shown to inhibit the in vitro interaction of FAK and paxillin with peptides modeled after the β cytoplasmic domain (Schaller et al., 1995). Thus, the binding of FAK or paxillin to β tails may be important for cell spreading. However, Rack1 and talin have also been shown to bind to the conserved membrane-proximal domain of the β1 and β3 tails (Liliental and Chang, 1998; Patil et al., 1999). These mutations may also inhibit the binding of these proteins with β tails.

The β1-787 (V/A) mutation had previously been shown to inhibit the interaction of ICAP-1 with the β1 cytoplasmic domain (Chang et al., 1997EF7). Because ICAP-1 is ubiquitously expressed (Chang et al., 1997EF7; Zhang and Hemler, 1999EF47) and can easily be detected in lysates from normal human fibroblasts (A.L.B., unpublished), tac-β1 may inhibit cell spreading by titrating ICAP-1 from endogenous β1 integrins. However, chimeric receptors containing the β1-787 (V/A) mutant also had a diminished ability to inhibit cell spreading on fibrinogen mediated by recombinant αIIbβ3. Therefore, although one could argue that the binding of ICAP-1 is required for the β1 tail to bind other proteins that normally also bind the β3 tail to regulate cell spreading, it is also possible that the β1-787 (V/A) mutation is affecting a different protein interaction required for cell spreading. For example, valine 787 is in a region of the β1 tail that is important for ILK association (Dedhar and Hannigan, 1996EF10) and for the binding of filamin to β tails in vitro (Loo et al., 1998EF25). Thus, there are several protein interactions that may be important in mediating β cytoplasmic domain function in cell spreading that could be disrupted by these mutations. More studies are required to identify the relevant ones.

Alanine substitutions of the tryptophan and aspartic acid residues in the WDT motif of the β1 tail inhibited the ability of tac-β1 to regulate integrin conformation and function in cell spreading. Interestingly, the WDT motif and the membrane-proximal NPXY motif overlap a putative binding site for talin (Horwitz et al., 1986EF15; Pfaff et al., 1998EF32; Kaapa et al., 1999EF18). Expression of recombinant forms of the head domain of talin can activate αIIbβ3 (Calderwood et al., 1999EF6), suggesting that the interaction of the head domain of talin with integrin β cytoplasmic domains may function in integrin activation. A similar interaction with the β1 cytoplasmic domain may regulate β1 integrin conformation and 9EG7 expression. Additionally, other studies have implicated the binding of talin to the β1 tail in regulating integrin function in cell spreading (Kaapa et al., 1999EF18). Our data indicating that mutations in the WDT motif inhibit the ability of the β tail to regulate integrin conformation and function in cell spreading is consistent with β tail-talin interactions having a role in regulating these processes. However, it is not known whether mutations in the WDT motif inhibit talin binding. In addition, other studies have suggested that the talin head domain does not bind to this region of the β tail, but interacts with the membrane-proximal region instead (Patil et al., 1999EF31). More studies are needed to understand how talin interacts with the β tail and the functional significance of the interaction.

As reported previously and in the current study, mutation of the membrane-proximal NPXY motif to NPXA appears to completely inhibit the function of β1 and β3 cytoplasmic domains (Reszka et al., 1992EF33; O'Toole et al., 1995EF30; Ylanne et al., 1995EF46; Tahiliani et al., 1997EF41; Schaffner-Reckinger et al., 1998EF36). Because integrin β cytoplasmic domains containing NPXA lose their abilities to bind to talin and filamin in vitro (Pfaff et al., 1998EF32), it is possible that one or both of these protein interactions with β cytoplasmic domains is required for all aspects integrin function requiring integrin β cytoplasmic domains. However, it is also possible that this mutation changes the overall conformation of integrin β cytoplasmic domains, thus nonspecifically inhibiting multiple protein interactions and integrin function in multiple assays. It seems likely that structural information on β cytoplasmic domains will be necessary to understand how mutations in this NPXY motif affect specific protein interactions.

In addition, we have found that alanine substitutions in the VTT motif of the β1, the β1-787-789 (^*/A), or the TST motif of β3, the β3-751-753 (^*/A) mutant, inhibited the ability of chimeric receptors containing these β tails to regulate endogenous β1 integrin conformation and function in cell attachment and cell spreading (also see Mastrangelo et al., 1999b). Additionally, the β1-787-789 (^*/A) mutant also abrogated the ability of tac-β1 to activate tyrosine phosphorylation; however, tac-β3 with the analogous mutation was able to signal, albeit to lower levels than tac-β3 containing a wild-type tail. Consistent with our findings, alanine substitutions at the threonine residues in the VTT motif of the β1 tail in the context of heterodimeric integrins inhibited the expression of the 9EG7 epitope on these integrins, as well as their function in cell attachment (Wennerberg et al., 1998). Interestingly, these mutant β1 integrins were still able to trigger FAK phosphorylation. This difference in phenotype could be due to the additional alanine substitution for valine in the VTT motif of our mutant.

Interestingly, the substitution of phenylalanine for tyrosines in the β1 and β3 tails did not have profound effects in our studies, although the ability of tac-β1 containing the β1-783,795 (Y/F) mutant to inhibit cell spreading was somewhat diminished. However, Wennerberg and colleagues demonstrated that changing tyrosine residues to phenylalanine in the β1 cytoplasmic domain dramatically inhibited the ability of β1 integrins to trigger FAK phosphorylation and paxillin phosphorylation and to mediate cell spreading (Wennerberg et al., 2000EF43). These differences could reflect differences in cell type: the use of GD25 derivatives (Wennerberg et al., 2000EF43) and normal human fibroblasts (present study). We have previously found differences in the behavior of these two cell types in our assays. For example, treating GD25 cells re-expressing the mouse β1 subunit with Mn+2 does not increase 9EG7 expression as in human fibroblasts and treating these same GD25 cells with sodium vanadate does not inhibit the expression of 9EG7 or inhibit integrin function in cell attachment as in human fibroblasts (A.M.M., unpublished). However, these differences may also reflect differences in chimeric receptors and heterodimeric integrins.

Other laboratories have also identified mutations that result in the inhibition of some integrin-mediated processes and not others. A substitution mutant in the putative α-actinin binding region of recombinant αIIbβ3 has been identified that was able to trigger FAK phosphorylation, but was unable to retract a fibrin clot (Lyman et al., 1997EF27). A different substitution mutant in the same region of the β3 tail was identified that inhibited the ability of recombinant αIIbβ3 to be recruited to pre-established focal adhesions and to be internalized, but that still functioned in cell spreading and focal adhesion formation (Ylanne et al., 1995EF46). In addition, a deletion of the membrane-proximal HDRRE motif inhibited the ability of a tac-β1 chimera to be recruited to focal adhesions and to activate tyrosine phosphorylation, but not to trigger cell detachment in response to LPA (David et al., 1999EF9). In the present study, we identified two mutations that specifically affect the ability of tac-β1 and tac-β3 to function as dominant negative inhibitors of cell spreading. Clearly, evidence is accumulating to support the notion that distinct protein interactions with integrin β cytoplasmic domains are required to regulate different aspects of integrin function. The task for the future is to identify the proteins involved and to determine how these interactions with the β cytoplasmic domain are regulated.

We thank Christopher Turner for the REF52 cells, Mark Ginsberg for CHO cells stably expressing αIIbβ3 (A5 cells) and αIIbβ3Δ727 (ETC12 cells), and Jane Sottile for human plasma fibronectin. We also thank Jane Sottile and Michael Dipersio for helpful discussions and critical reading of this manuscript. This work was funded by grants from the National Institutes of Health (GM51540, T32HL07529 and T32GM07033) and the American Heart Association-New York affiliate (0020180T).