Specific Phosphorylations Transmit Signals from Leukocyte β2 to β1 Integrins and Regulate Adhesion*

Background: Integrins regulate leukocyte adhesion in a stepwise manner. Results: β2 Integrins can signal to β1 integrin very late antigen 4 and inhibit ligand binding. Conclusion: Phosphorylation on both the α and β chains of β2 integrins are needed, but the signal is transmitted through the β2 chain, leading to dephosphorylation of β1. Significance: Our findings show a mechanism of integrin cross-talk. The regulation of integrins expressed on leukocytes must be controlled precisely, and members of different integrin subfamilies have to act in concert to ensure the proper traffic of immune cells to sites of inflammation. The activation of β2 family integrins through the T cell receptor or by chemokines leads to the inactivation of very late antigen 4. The mechanism(s) of this cross-talk has not been known. We have now elucidated in detail how the signals are transmitted from leukocyte function-associated antigen 1 and show that, after its activation, the signaling involves specific phosphorylations of β2 integrin followed by interactions with cytoplasmic signaling proteins. This results in loss of β1 phosphorylation and a decrease in very late antigen 4 binding to its ligand vascular cell adhesion molecule 1. Our results show how a member of one integrin family regulates the activity of another integrin. This is important for the understanding of integrin-mediated processes.

Leukocyte extravasation plays a key role in inflammation. The recruitment of leukocytes into inflamed tissue requires leukocyte adhesion to the vascular endothelium as a sequence of events, including the capture of circulating leukocytes and subsequent leukocyte rolling, arrest, firm adhesion, and transmigration. These processes are mediated by adhesion receptors, of which selectins and integrins are of pivotal importance (1,2).
Integrins are transmembrane heterodimeric receptors that can communicate in two directions across the plasma membrane. Outside-in signaling occurs by binding of ligands or, e.g., activating antibodies to the integrins, whereas inside-out signaling is due to ligand binding to non-integrin receptors (such as chemokine receptors or the T cell receptor), which, through intracellular signaling, convey the message to integrins, resulting in changes in activity. Integrins can exist in at least three different conformations: inactive, extended, and extended open (3)(4)(5)(6).
The integrin cytoplasmic domains are short and devoid of catalytic activity. However, integrin function is regulated by protein interactions with the cytoplasmic domains. Many adaptor proteins, including filamin, ␣-actinin, and 14-3-3 protein family members, compete for the relatively few binding sites on the integrin tails. Talin and the kindlin family of proteins are considered to be important in the final stages of activation (1,6,12,13).
LFA-1 and VLA-4 mediate distinct steps in the adhesion cascade. VLA-4 is the predominant integrin regulating rolling, and it participates in adhesion strengthening, whereas LFA-1 is needed for firm adhesion and migration (2). Importantly, integ-rins have the ability to modulate their adhesive properties within seconds after chemokine stimulation (14).
Little is known about how integrins act in concert. Some earlier studies have shown that LFA-1 may regulate VLA-4 activity in T cells, but the mechanisms have remained largely unknown (15). We have shown previously that integrin phosphorylations are essential in the regulation of ␤ 2 integrin activity (6, 16 -18). Phosphorylation of Thr-758 in the ␤ 2 chain leads to 14-3-3 binding, recruitment of T cell lymphoma invasion and metastasis 1 (Tiam1) protein, and up-regulation of Ras-related C3 botulinum toxin substrate 1 (Rac1) (19).
We now show that LFA-1 and CR4 ␣ chain phosphorylation is needed for chemokine-induced cross-talk to VLA-4. In contrast, signaling through the T cell receptor and outside-in activation do not need ␣ chain phosphorylation. However, after both inside-out and outside-in activations, signaling is transmitted through ␤ 2 chain phosphorylation and the 14-3-3/ Tiam1/Rac1 pathway. This results in the inhibition of ␤ 1 phosphorylation, VLA-4 binding to filamin, and inactivation of VLA-4.
Cell Culture and Transfection-Stable transfectants of the K562 cell line (22) expressing WT CR4 or S1158A-␣ X ␤ 2 (CR4 S/A) or empty plasmid have been described previously (23). HMEC cells were a gift from Antti Vaheri (University of Helsinki, Finland) and have been described previously (24). Cells were grown on plates coated with 0.5% gelatin in RPMI 1640 medium supplemented with 10% FBS, 100 units/ml penicillin/ streptomycin, and 2 mM L-glutamine. The human T cell lymphoma cell line clone J␤2.7, which lacks the CD11 chain (25), was a gift from N. Hogg and was grown in RPMI 1640 medium supplemented with 10% FBS, 2 mM L-glutamine, and 100 units/ml penicillin/streptomycin. The J␤2.7 cell lines expressing LFA-1 WT or the phosphorylation mutant S1140A were produced by virus transductions. HEK293T cells were transiently transfected using the following protocol. 30 g of DNA (12.5 g of retroviral vector, 12.5 g of GAG-Pol, and 5 g of Env) were mixed with 120 l of polyethylenimine (Polysciences, Warrington, PA) and DMEM to a total volume of 1.5 ml and incubated for 10 min at room temperature. 1 ml of DMEM with 10% FBS, 2 mM L-glutamine, and 100 units/ml penicillin/streptomycin and 250 l of the DNA mixture were added to each well. The supernatants were harvested 24 and 48 h after transfection. For transduction, 10 5 J␤2.7 cells were grown for 4 -6 h in the presence of 1 l of Polybrene (Sigma-Aldrich), after which 1 ml of fresh medium and 1 ml of virus supernatant were added to each well. 24 -72 h after transduction, the cells were rinsed and expanded in fresh medium.
Cell Adhesion Assays-For adhesion assays with J␤2.7 cell lines, culture wells were coated with VCAM-1-Fc, and 10,000 cells were added and allowed to adhere for 30 min with or without preincubation with 10 g/ml ␣ 4 -blocking 2B4 or ␤ 1 -blocking mAb 13 antibodies for 30 min or SDF-1␣ (50 ng/ml) for 15 min. Adhered and spread cells were counted from ten screens using an Evos microscope (Invitrogen). For live cell imaging, cells were followed for 10 -120 min, and time lapse movies were taken with the Evos microscope. Adhesion assays of K562 cells to VCAM-1 were performed as described previously (23).
Briefly, K562 cells transfected with empty plasmid, WT CR4, or S1158A-␣ X ␤ 2 (CR4 S/A) were allowed to bind to VCAM-1-Fc coated on Nunc MaxiSorp 96-well plates (Thermo Fisher Scientific). 100,000 cells were incubated in the wells without treatment or in the presence of antibodies against ␣ 4 (2B4) or ␤ 1 (Mab 13), after which the wells were washed and the bound cells were lysed and detected using phosphatase substrate (Sigma-Aldrich).
Flow Adhesion Assays-For flow adhesion assays, HMEC cells were seeded on flow chamber channels (-Slide VI 0.4 , Ibidi GmbH, Martinsried, Germany) coated with 0.5% gelatin and grown until confluency, after which cells were activated with TNF-␣ (Sigma-Aldrich), 50 ng/ml, overnight. Alternatively, chambers were coated with 6 g/ml VCAM-1. J␤2.7 cell lines were stained with carboxyfluorescein diacetate succinimidyl ester (Invitrogen) according to the instructions of the manufacturer and treated with SDF-1␣ for 10 min or left untreated. The cell suspensions were then injected into a flow system using a silicone tubing loop connected to a multiphaser NE-1000 syringe pump (New Era Pump Systems, Inc., Farmingdale, NY), allowing cells to flow over HMEC-1-or VCAM-1-coated Ibidi microslides VI 0.4. A continuous shear flow rate of 0.3 dynes/ cm 2 was used. Adhering cells were monitored by fluorescence microscopy (EVOS fl, Invitrogen), and firmly attached cells were counted at two different time points from eight separate screens.
Migration Assays-For the chemokine-induced migration assays, 5-m pore size Transwell membranes were incubated with soluble VCAM-1, ICAM-1, or BSA (10 g/ml or the indicated concentrations) overnight at 4°C, washed once with PBS, and placed in the wells of a 24-well plate containing 600 l of buffer (RPMI 1640 medium/10% FBS) with 15 ng/ml SDF-1␣. Cells were stained with carboxyfluorescein diacetate succinimidyl ester, calcein blue (eBioscience, San Diego, CA), fluorescein, or eFLuor 670 (eBioscience) according to the instructions of the manufacturer. In some cases, cells were preincubated with 10 g/ml of the ␣ 4 -blocking antibody 2B4 for 15 min at 37°C. 60,000 cells were placed on the membrane and allowed to migrate toward SDF-1␣ at 37°C for 1 h. Migrated cells were collected and counted.
Peptide Transfections-60,000 cells in 100 l of buffer (RPMI 1640 medium/5% FBS) were transfected in 48-well plates. For transfection, 2 g of nonphosphorylated or Thr-758-phosphorylated ␤ 2 peptide was incubated for 20 min at room temperature with 1 ml enhancer and 1 ml Turbofect transfection reagent (Fermentas, Thermo Fischer Scientific) in 50 l of RPMI medium and added to the cells. After incubation, the transfected cells were added to Transwell filters, and migration assays were performed as above. To block the interaction between 14-3-3 proteins and their targets, cells were transfected with the R18 peptide or the control peptide P621 (2 g/transfection) prior to migration. To inhibit Tiam1, cells were preincubated with the NSC23766 inhibitor (100 M, TOCRIS Bioscience) for 2 h at 37°C.
Quantifications-Quantifications were done using ImageJ (v10.2) software. The fluorescence intensity of cells stained with mab24 and KIM127 is an average of 100 cells for each condition and cell line done in triplicate. Cell extensions were quantified as extensions longer than the cell body from 10 frames of adhered cells in triplicate. All statistical analyses were performed in Excel with unpaired Student's t test. In the figures, the mean Ϯ S.D. is given.

Binding of J␤2.7 Cells to VCAM-1 through VLA-4 Can Be
Blocked by LFA-1-For these studies, we mainly used the J␤2.7 Jurkat cell line, which contains a mutation in the ␣ L gene, leading to lack of expression of ␣ L and, thus, no functional LFA-1 on the cell surface. The cells spread and adhere strongly to VCAM-1 (Fig. 1A). The VCAM-1 binding integrin VLA-4 expression on J␤2.7 cells was verified by flow cytometry (data not shown). By blocking ␣ 4 or ␤ 1 , adhesion is inhibited, showing that the cells bind to VCAM-1 through VLA-4 ( Fig. 1A).
This cell line was used to study how the presence of LFA-1 affects VLA-4 functions. A stable J␤2.7 cell line expressing the ␣ L chain of LFA-1 (LFA-1) was produced by virus transduction. The expression of ␣ L and ␤ 2 was detected by flow cytometry, and binding of LFA-1 to its ligand, ICAM-1, was shown by adhesion to ICAM-1 under flow (data not shown). Therefore, LFA-1 cells express functional LFA-1. To study how the expression of LFA-1 affects VLA-4, J␤2.7 cells, or LFA-1 cells were allowed to adhere to VCAM-1-coated wells and adhered cells were counted (Fig. 1B). Cells were either unactivated or activated by the chemokine stromal cell-derived factor 1␣ (SDF-1␣), which has been shown to activate LFA-1 (26). Cells adhere and spread much faster on VCAM-1 in the absence of activated LFA-1. The presence of SDF-1␣ -activated LFA-1 leads to a clear reduction of cell adhesion to VCAM-1, pointing to a trans-dominant inhibition of VLA-4 by LFA-1. Inhibition of VCAM-1 binding is not seen after activation with anti-CD2 or anti-CD28, indicating that this inhibition is not a common effect of leukocyte membrane receptors (data not shown).
Inhibition of VCAM-1 binding could also be seen in an adhesion assay under flow on VCAM-1-coated surfaces (Fig. 1C) or endothelial cells activated with TNF␣ (Fig. 1D). The bright field image of the cells spreading on VCAM-1 show that fewer LFA-1 cells adhere to and spread on VCAM-1 (Fig. 1E). Activation by SDF-1␣ of JE6.1 Jurkat cells, which express both LFA-1 and VLA-4, also leads to a reduction of VLA-4 binding to VCAM-1 (data not shown).

Strong Adhesion between VLA-4 and VCAM-1 Causes
Impaired Detachment, Which Can Be Inhibited by LFA-1-J␤2.7 cells adhere strongly to VCAM-1, and cells bound to VCAM-1 cannot be removed by increasing the flow rate extensively, which is the case with LFA-1 adhesion to ICAM-1 (data not shown). This strong adhesion is also seen in cell migration experiments. Cells were allowed to migrate over Transwell filters coated with VCAM-1 toward SDF-1␣ present below. J␤2.7 cells lacking LFA-1 do not migrate over the VCAM-1-coated filters. To detect whether cells stay bound to the filter, they were fluorescently stained with carboxyfluorescein diacetate succinimidyl ester. In cells lacking LFA-1, most cells stay bound to the filter. When cells express LFA-1 or the VCAM-1 interaction is blocked by a VLA-4 blocking antibody, cells migrate through the filters toward SDF-1␣. This indicates that LFA-1 inhibits VLA-4 activity, therefore reducing adhesion between VLA-4 and VCAM-1 ( Fig. 2A).
We next studied the cell morphology of J␤2.7 cells adhering to VCAM-1 by live cell imaging. J␤2.7 cells spread on VCAM-1 with long extensions, which are not seen in LFA-1-expressing cells. When J␤2.7 cells were allowed to adhere and form extensions and then the ␣4-blocking antibody was added, the extensions disappeared (Fig. 2B). To elucidate whether the long extensions are filopodia growing out of the cell body or caused by impaired retraction of adhesion sites, we performed live cell imaging of cells. This clearly showed that the extensions result from strong adhesions that cannot be released as the cells are moving in an opposite direction, therefore leaving a long extension behind (Fig. 2C). When cells express SDF-1␣-activated LFA-1 or when VLA-4 is blocked by an ␣ 4 -blocking antibody,  NOVEMBER 14, 2014 • VOLUME 289 • NUMBER 46 these extensions were not seen, indicating that the strong adhesions between VLA-4 and VCAM-1 cannot be formed (data not shown). LFA-1 needs to be activated to be able to inhibit VLA-4. When LFA-1-expressing cells were allowed to bind to VCAM-1 before the addition of SDF-1␣, strong adhesions could not be released, regardless of later activation of LFA-1. This indicates that LFA-1 can block the formation of VLA-4/ VCAM-1 interactions, but when the strong adhesions and long extensions have formed, activation of LFA-1 cannot disassociate these.

LFA-1 ␤ Chain Regulates VLA-4 by Phosphorylation
Phosphorylation of ␣ L and ␣ X Is Required for the Inhibition of VLA-4 after Chemokine Activation-To elucidate the mechanisms behind the VLA-4 inhibition, we studied the role of integrin ␣ and ␤ chain signaling and the activation state of LFA-1 needed for cross-talk. We have shown previously that the ␣ chain of LFA-1 is phosphorylated on Ser-1140 and that this modification is needed for conformational changes in the integrin after chemokine activation (17). We therefore made a J␤2.7 cell line stably expressing the phosphorylation mutant ␣ L S1140A (LFA-1 S/A). A similar expression of LFA-1 WT and LFA-1 S/A was verified by flow cytometry, and mutated cells showed reduced binding to ICAM-1 compared with WT LFA-1 (Fig. 1, A and C). We next performed static adhesion assays on VCAM-1 with the three cell lines J␤2.7, LFA-1, and LFA-1 S/A in the presence of SDF-1␣. Bright field images of adhering cells were taken after 30 min. After 2 h, cells were fixed and stained with phalloidin. Less cells adhered to and spread on VCAM-1 in LFA-1 cells, whereas the LFA-1 S/A cells resembled those of J␤2.7 cells (Fig. 3A). Adhered cells and the amount of cells with an extension longer than the cell body were quantified. Although LFA-1 expression reduced VLA-4 adhesion to VCAM-1, the phosphorylation mutant LFA-1S/A was not able to inhibit VLA-4, indicating that a phosphorylated Ser-1140 is essential for inhibition (Fig. 3B). We next performed a migration assay. The three cell lines were first stained with different fluorescent markers and mixed before they were allowed to migrate over VCAM-1-coated filters. We first verified that the stains did not affect migration (not shown). In the experiment shown in Fig. 3C, J␤2.7 cells are red, LFA-1 cells are green, and LFA-1 S/A blue. The LFA-1 S/A phosphorylation mutant was not able to release J␤2.7 cells from the VCAM-1-coated filters as LFA-1 WT did, indicating that phosphorylation of the ␣ chain Ser-1140 is needed for LFA-1-mediated inhibition of VLA-4 in SDF-1␣-activated cells (Fig. 3C).
The ␤ 2 integrin CR4 (␣ X ␤ 2 ) is also expressed in leukocytes, and we have shown recently that it is phosphorylated on serine 1158 in the ␣ chain (23). To find out whether the cross-talk is unique to LFA-1, we studied whether CR4 can regulate VLA-4. We used K562 cells that were transfected with WT CR4 or with the ␣ chain phosphorylation mutant ␣ X -S1158A (CR4 S/A). These cells express ␣ 4 , as detected by the anti-␣ 4 antibody 2B4. Cells stably transfected with WT CR4, CR4 S/A, or empty plasmid were allowed to adhere to VCAM-1 coated on plastic. The mock-transfected K562 cells bound VCAM-1, whereas WT CR4 transfectants showed almost no binding. Expression of CR4 S/A in K562 cells resulted in equal binding as K562 mock cells. Adhesion of mock cells was inhibited to the background level by treatment with anti-␣ 4 or anti-␤ 1 antibodies (Fig. 3D). In an additional assay, K562 cells stably transfected with WT CR4 or empty plasmid were left to adhere to confluent layers of endothelial cells under flow. More mock-transfected K562 cells bound endothelial cells than WT CR4 transfectants. That is, expression of functional CR4 blocked the adhesion of K562 to endothelial cells (data not shown). We conclude that both ␤ 2 integrins studied, LFA-1 and CR4, can inhibit VLA-4 and that this inhibition requires a phosphorylatable ␣ chain.
Transdominant Inhibition of VLA-4 by LFA-1 Needs an Active LFA-1 and pT758-14-3-3Tiam1 Signaling-It has been shown previously that the LFA-1 S1140A mutant is not fully activated after SDF-1␣ induction, and the mab24 activation epitope that reveals the extended open conformation of the integrin head domain (27) can be detected only in LFA-1 WTexpressing cells (17). Therefore, we tested additional ways to activate LFA-1, including inside-out activation with anti-CD3 or outside-in activation with soluble ICAM-1 and ICAM-2 (28) or CBR LFA-1/2 (5), a LFA-1-activating antibody. We then performed an adhesion assay to VCAM-1 under static conditions (Fig. 4A) or under flow (Fig. 4B). LFA-1 was able to reduce VLA-4 binding after each activation. The mutated ␣ chain was able to inhibit VLA-4 when the ␤ 2 integrin was activated with anti-CD3 or by outside-in activation but not by SDF-1␣. Therefore, the inhibition of VLA-4 is not directly mediated through the ␣ chain Ser-1140 phosphorylation itself, but the phosphorylation is needed when cells are activated with SDF-1␣ to fully activate the integrin. We next studied the activation state of LFA-1. Both WT and mutant LFA-1 can be detected with the KIM127 antibody (which recognizes the extended conformation with either a closed or an open headpiece) after SDF-1␣ activation, indicating that LFA-1 is in an activated extended form (Fig. 4C). Only LFA-1, but not LFA-1 S/A, expressed the mab24-epitope, recognizing the extended open headpiece after SDF-1␣ activation (Fig. 4D). When cells were activated with soluble ICAM-1, however, both LFA-1 WT-and LFA-1 S/Aexpressing cells were mab24-positive (Fig. 4E), indicating that the ␣ L phosphorylation mutant can be fully activated by outside-in activation and that there is a correlation between the presence of mab24-positive LFA-1 and LFA-1 inhibition of VLA-4. This implies that the fully activated, mab24-positive form of LFA-1 is present and needed for VLA-4 inhibition but does not exclude the possibility that molecules with lower affinity conformation(s) take part in the cross-talk.
LFA-1 inhibited VLA-4 both after inside-out activation by SDF-1␣ and anti-CD3 and outside-in activation, e.g. by the activating antibody CBR LFA-1/2. Interestingly, these activations all lead to the phosphorylation of the LFA-1 ␤ 2 -chain on Thr-758 (Fig. 4F). We have previously characterized the signaling pathway downstream of Thr-758-phosphorylated LFA-1. The Thr-758-phosphorylated ␤ 2 binds to the adaptor protein 14-3-3, which, in turn, binds to the Rac1 guanine nucleotide exchange factor Tiam1, which activates Rac1 (19). We therefore studied the role of ␤ 2 chain signaling in the LFA-1-mediated inhibition of VLA-4. J␤2.7 cells lacking LFA-1 were transfected with the phospho-Thr-758 peptide of ␤ 2 (p␤ 2 ), which has been shown to activate the signaling pathway on its own (21). Cells were then allowed to adhere to VCAM-1 under flow, and adhered cells were counted. The presence of the ␤ 2 peptide inhibited VLA-4 activity, indicating that this pathway can mediate LFA-1 inhibition of VLA-4. Cells transfected with the ␤ 2 peptide lacking phosphate showed a slight decrease of VLA-4 activity, which could be caused by the peptide being phosphorylated in the cells. Transfection with a control peptide did not interfere with VLA-4 activity (Fig. 5A). The Thr-758phosphorylated ␤ 2 chain binds to 14-3-3, so we next transfected J␤2.7 cells lacking LFA-1 with p␤ 2 with or without the 14-3-3blocking peptide R18 or R18 alone. Cells were allowed to adhere to VCAM-1 under flow, and adhered cells were counted. The p␤ 2 peptide inhibited VLA-4, and blocking signaling from p␤ 2 to 14-3-3 reduced the inhibition. R18 did not affect VLA-4 NOVEMBER 14, 2014 • VOLUME 289 • NUMBER 46 directly because no effect on VLA-4 to VCAM-1 binding was seen in J␤2.7 cells lacking LFA-1 (Fig. 5B). In the following experiments, we studied the role of Tiam1, which is the next player in the signaling cascade. J␤2.7 or LFA-1 cells were treated with an inhibitor of Tiam1. Adhesion experiments with VCAM-1 under flow were performed and showed that blocking Tiam1 reduced LFA-1 inhibition of VLA-4 (Fig. 5C). This indicates that LFA-1 can inhibit VLA-4 activity through the pT758␤ 2 /14-3-3/Tiam1-pathway. This pathway leads to activation of Rac1 activity. Rac1 has been implicated in VLA-4 reg- Adhered cells and cells with long extensions/impaired detachment were quantified from six screens in triplicate. C, migration depends on phosphorylation of the ␣ L chain. The three cell lines were stained with fluorescent dyes (J␤2.7, red; LFA-1, green; and LFA-1 S/A, blue). Equal amounts of cells were mixed and allowed to migrate toward SDF-1␣ over a VCAM-1-coated filter with or without the VLA-4-blocking antibody (2B4). Cells attached to the filter and migrated cells were quantified from ten screens in triplicate. An average of experiments are shown in the left panels, and a representative fluorescence pictures of one experiment is shown in the right panels. Scale bars ϭ 100 m. D, CR4 ␣ X chain phosphorylation inhibits VLA-4/VCAM-1 binding. K562 cells were transfected with empty plasmid (K562), WT CR4 (CR4), or the ␣ X chain phosphorylation mutant S1158A (CR4 S/A). Cells were allowed to adhere to VCAM-1 with or without the VLA-4-blocking antibodies 2B4 or Mab 13 as described under "Experimental Procedures," and bound cells were detected enzymatically. Cell binding is reported relative to K562 cell binding. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. ulation, and we therefore blocked Rac1 in the three J␤2.7 cell lines with the Rac1 inhibitor EHT 1864. Lack of Rac1 activity caused a large reduction of cell adhesion and spreading in all three cell lines, and differences between cell lines were not significant (data not shown).

LFA-1 ␤ Chain Regulates VLA-4 by Phosphorylation
We next studied the expression and distribution of ␣ 4 in J␤2.7 and LFA-1 cells. Equal amounts of ␣ 4 was seen in all three cell lines, as seen by flow cytometry (Fig. 6C) and Western blot analysis (Fig. 6A). Immunofluorescence staining of the subcellular localization of ␣ 4 , talin, and phalloidin showed that ␣ 4 and talin are present at the membrane at adhesion sites where they colocalize with phalloidin, whereas less ␣ 4 and talin were seen in these structures in LFA-1 cells (Fig. 6D, arrows). We finally looked at the localization of VLA-4-binding proteins in J␤2.7 cells. The strong adhesions sites in the cells, which cannot be detached during migration, showed positive integrin ␣ 4 , talin, 14-3-3, and phalloidin staining at the membrane. Filamin was also present in the adhesion sites but less at the membrane (Fig.  6C, arrows).

DISCUSSION
Leukocyte cell adhesion to endothelial cells and transendothelial migration is a dynamic, multistep process in which cells are tethered to and roll on the endothelial cells, after which they adhere to and crawl on the endothelium and finally transmigrate. This complicated sequence of events is mediated by different adhesion molecules on the transmigrating cells, regulated through intracellular signaling, ligand binding, and shear flow and by signals expressed by the endothelium (29,30).
T cells express several integrins, of which LFA-1 and VLA-4 are particularly important for the adhesion to endothelial cells. They play different roles in this cascade because VLA-4 can mediate cell tethering and rolling as well as firm arrest (31), whereas LFA-1 participates in the formation of strong adhesions and regulates migration over the endothelial surface (1). These interactions are mediated by changes in integrin conformation and affinity. We show that the activation of LFA-1 can inhibit VLA-4 binding to VCAM-1, an earlier step in the adhesion cascade. Cells lacking active LFA-1 adhere strongly to VCAM-1, which impairs migration because of the inability to detach adhesions and retract the trailing edge in the polarized cells. The expression of activated LFA-1 can inhibit this strong adhesion to the same extent as VLA-4 blocking antibodies and allow cells to migrate.
Regulation between different integrin classes in a given leukocyte was first described in 1993 (15). After that, the phenomenon has been reported in other cell types and between different integrin subclasses. Although, in many cases, the outside-in activation through ligand binding of one integrin decreases the activity of another integrin, there are reports that show an opposite effect (32,33). It has been shown that the activation of LFA-1 can affect VLA-4 activity in T cells (34), but the mechanism has not been known.
We show that an activated LFA-1 is needed for inhibition of VLA-4 (Fig. 7). The downstream signaling from the activated LFA-1 requires the phosphorylation of the ␤ 2 cytoplasmic chain on Thr-758. Both the activation of the T cell receptor and SDF-1␣ chemokine signaling result in the phosphorylation of Thr-758 in LFA-1 and inhibition of VLA-4. In a similar way, outside-in activation by the natural ligands ICAM-1 and ICAM-2 or the activating antibody CBR LFA-1/2 inhibit In this setting, LFA-1 is bound to filamin, and 14-3-3 is associated with VLA-4, which is phosphorylated on Thr-788/789 on the ␤ 1 chain. When LFA-1 is activated, the ␤ 2 chain is phosphorylated on Thr-758, 14-3-3 and Tiam1 bind, and VLA-4 binding to VCAM-1 is inhibited. Simultaneously, Thr-788/789 phosphorylation on the ␤ 1 chain and 14-3-3 binding are reduced, whereas filamin binding is increased.
VLA-4. Phosphorylation of the ␤ 2 chain is needed for the activation of leukocyte integrins, and the signaling cascade downstream of ␤ 2 integrin phosphorylation, leading to rearrangement of the actin cytoskeleton, has been characterized. T cell receptor-induced activation has been found to result in phosphorylated Thr-758 in ␤ 2 , binding of the adaptor protein 14-3-3 and the Rac1 guanine nucleotide exchange factor Tiam1, and, ultimately, to the activation of Rac1, cytoskeleton rearrangements, and integrin clustering (19,21,23). Transfection of a Thr-758-phosphorylated ␤ 2 cytoplasmic peptide was enough to initiate this signaling pathway (19), and we now show that it is sufficient to induce a decrease in VLA-4 ligand binding capacity when transfected into LFA-1-negative cells. This has also been shown in another integrin cross-talk setting. The ␤ 3 cytoplasmic tail alone is necessary and sufficient to mediate transdominant inhibition of ␣ 5 ␤ 1 . The expression of the isolated ␤ 3 cytoplasmic tail exerts an inhibitory effect upon ␣ 5 ␤ 1mediated migration, as well as phagocytosis, in a similar way as ␣ v ␤ 3 ligation (35). In LFA-1-expressing cells, inhibition of VLA-4 by LFA-1 could be overcome by blocking 14-3-3 binding using the R18 blocking peptide, which inhibits its binding to phospho-␤ 2 and other molecules of the signaling cascade. Incubation of the cells with the Tiam1 inhibitor, the next player in the pathway, increased VLA-4/VCAM-1 binding back to the basal level.
There is also a slight, but not significant, increase in VCAM-1 binding in cells lacking LFA-1 and treated with the Tiam1 inhibitor. Therefore, we cannot exclude the possibility that the inhibition of Tiam1 may also directly affect VLA-4 binding to VCAM-1 and that the full effect of the increase in VCAM-1 binding is not detectable because of the strong interaction between VLA-4 and VCAM-1 in untreated cells. Blocking Tiam1, however, would lead to less activation of Rac1, which should decrease VCAM-1 binding because Rac1 has been implicated previously in VLA-4 activation (36). It is possible that the signaling pathway downstream of LFA-1, which leads to activation of Rac1, may sequester Rac1 activity away from VLA-4, thereby inhibiting VLA-4. Alternatively, a large proportion of Rac1 is consumed by the LFA-1/Rac1 pathway, resulting in less availability for the VLA-4/Rac1 pathway.
Interestingly, ␣ chain phosphorylation was found to play an essential role in the activation cascade when inside-out activation was triggered with the chemokine SDF-1␣ but not by other forms of activation. This is not unique for LFA-1 because the CR4-integrin also requires ␣ chain phosphorylation for transdominant inhibition of VLA-4. SDF-1␣ has also been reported to activate VLA-4-dependent leukocyte adhesion (36,37). However, according to Manevich et al. (37) and our own observations, Jurkat cells (from which the J␤2.7 cell line is derived) show stronger VLA-4/VCAM-1 adhesion than primary T cells. This indicates that VLA-4 in J␤2.7 is already in the high-activity conformation and, thus, cannot achieve any higher affinity.
We also show that, under conditions where LFA-1 is able to inhibit VLA-4, there is an increased amount of the mab24-positive conformation of LFA-1, which is not seen in non-activated cells or SDF-1␣-activated ␣ chain phosphorylation mutants not mediating inhibition. The mab24-positive integrin corresponds to the high-affinity extended open headpiece conformation (1,2). The amount of KIM127-positive reactivity, which detects the extended conformation of the integrin, was the same in all activated cells.
Like other integrins, VLA-4 activities are also regulated by phosphorylation. The VLA-4 ␣ chain was phosphorylated on Ser-988 in the cell lines studied, regardless of the expression of LFA-1 or its activation. The phosphorylation of ␤ 1 on Thr-788/ 789 corresponds to the phosphorylation triplet in ␤ 2 , including Thr-758. This phosphorylation reduces ␤ 1 association with the actin cytoskeleton (38) and increases pressure-induced cell adhesion in cancer cells (39). Importantly, cells expressing the activated form of LFA-1 showed reduced phosphorylation of ␤ 1 Thr-788/789.
Like ␤ 2 , the ␤ 1 cytoplasmic part also interacts with filamin, 14-3-3, and talin (40), but the regulation of these interactions by phosphorylation has not been studied. For ␤ 2 , it is known that the bent inactive conformation interacts with filamin and that, upon Thr-758 phosphorylation, binding of 14-3-3 outcompetes filamin binding (18). We showed that, when LFA-1 mediates inhibition of VLA-4, LFA-1 was phosphorylated on Thr-758 and bound to 14-3-3 but not to filamin. Instead, the binding between VLA-4 and filamin was increased, and the binding to 14-3-3 was decreased. VLA-4 binds to 14-3-3 both through the ␣ and ␤ chains, whereas LFA-1 binds only through the ␤ 2 chain. It has recently been shown that the Thr-758 phosphorylated binding motif in the ␤ 2 integrin tail has a much higher affinity for 14-3-3 than the corresponding ␣ 4 motif (41). Therefore, Thr-758 phosphorylation of ␤ 2 may reduce binding between VLA-4 and 14-3-3 by binding to 14-3-3 protein, enabling binding of filamin to the ␤ 1 polypeptide. In this way, specific integrin phosphorylations can control cell adhesion and migration by spatial segregation of adaptor-protein binding.
Under several clinical conditions, it would be important to affect leukocyte adhesion and migration into tissues. Modulation of ␤ 2 integrins is currently used in the treatment of autoimmune diseases (32). Our results show that this may affect not only LFA-1 but also VLA-4 and, possibly, other integrins, which must be taken into consideration in the interpretation of the clinical outcome.