The Unique Disulfide Bond-stabilized W1 β4-β1 Loop in the α4 β-Propeller Domain Regulates Integrin α4β7 Affinity and Signaling*

Background: Integrin α4β7 is unique for mediating rolling and firm adhesion of lymphocytes pre- and post-activation. Results: Disrupting the disulfide bond-stabilized W1 β4-β1 loop in the α4 β-propeller domain impaired α4β7-mediated, low-affinity ligand binding and bidirectional signaling. Conclusion: The W1 β4-β1 loop regulates integrin ligand binding and signaling. Significance: Our findings reveal a particular molecular basis for α4β7-mediated rolling cell adhesion. Integrin α4β7 mediates rolling and firm adhesion of lymphocytes pre- and post-activation, which is distinct from most integrins only mediating firm cell adhesion upon activation. This two-phase cell adhesion suggests a unique molecular basis for the dynamic interaction of α4β7 with its ligand, mucosal addressin cell adhesion molecule 1 (MAdCAM-1). Here we report that a disulfide bond-stabilized W1 β4-β1 loop in α4 β-propeller domain plays critical roles in regulating integrin α4β7 affinity and signaling. Either breaking the disulfide bond or deleting the disulfide bond-occluded segment in the W1 β4-β1 loop inhibited rolling cell adhesion supported by the low-affinity interaction between MAdCAM-1 and inactive α4β7 but negligibly affected firm cell adhesion supported by the high-affinity interaction between MAdCAM-1 and Mn2+-activated α4β7. Additionally, disrupting the disulfide bond or deleting the disulfide bond-occluded segment not only blocked the conformational change and activation of α4β7 triggered by talin or phorbol-12-myristate-13-acetate via inside-out signaling but also disrupted integrin-mediated outside-in signaling and impaired phosphorylation of focal adhesion kinase and paxillin. Thus, these findings reveal a particular molecular basis for α4β7-mediated rolling cell adhesion and a novel regulatory element of integrin affinity and signaling.

Integrin ␣ 4 ␤ 7 mediates rolling and firm adhesion of lymphocytes pre-and post-activation, which is distinct from most integrins only mediating firm cell adhesion upon activation. This two-phase cell adhesion suggests a unique molecular basis for the dynamic interaction of ␣ 4 ␤ 7 with its ligand, mucosal addressin cell adhesion molecule 1 (MAdCAM-1). Here we report that a disulfide bond-stabilized W1 ␤4-␤1 loop in ␣ 4 ␤-propeller domain plays critical roles in regulating integrin ␣ 4 ␤ 7 affinity and signaling. Either breaking the disulfide bond or deleting the disulfide bond-occluded segment in the W1 ␤4-␤1 loop inhibited rolling cell adhesion supported by the lowaffinity interaction between MAdCAM-1 and inactive ␣ 4 ␤ 7 but negligibly affected firm cell adhesion supported by the highaffinity interaction between MAdCAM-1 and Mn 2؉ -activated ␣ 4 ␤ 7 . Additionally, disrupting the disulfide bond or deleting the disulfide bond-occluded segment not only blocked the conformational change and activation of ␣ 4 ␤ 7 triggered by talin or phorbol-12-myristate-13-acetate via inside-out signaling but also disrupted integrin-mediated outside-in signaling and impaired phosphorylation of focal adhesion kinase and paxillin. Thus, these findings reveal a particular molecular basis for ␣ 4 ␤ 7 -mediated rolling cell adhesion and a novel regulatory element of integrin affinity and signaling.
Integrins are a family of ␣/␤ heterodimeric cell adhesion molecules that mediate cell-cell, cell-matrix, and cell-pathogen interactions and signal bidirectionally across the plasma membrane (1). Different from most integrins that mediate only firm cell adhesion upon activation, a small subset of integrins, including ␣ 4 ␤ 7 , ␣ 4 ␤ 1 , ␣ 6 ␤ 4 , and ␣ L ␤ 2 , can mediate rolling and firm cell adhesion pre-and post-activation (2)(3)(4)(5). Integrin ␣ 4 ␤ 7 is expressed exclusively on lymphocytes, and its major ligand, mucosal addressin cell adhesion molecule 1 (MAdCAM-1) 3 , is specifically expressed on high endothelial venules of Peyer's patches and postcapillary venules in intestinal laminae propriae (6). The unique two-phase cell adhesion mediated by integrin ␣ 4 ␤ 7 makes it indispensable in the homing of lymphocytes to the intestine and the associated lymphoid tissues and plays critical roles in gut immune homeostasis and the pathogenesis of intestinal inflammatory disorders (7)(8)(9).
The rolling and firm cell adhesion mediated by ␣ 4 ␤ 7 are dependent on the dynamic regulation of integrin affinity (2,3). The inactive and activated ␣ 4 ␤ 7 support the rolling and firm adhesion of lymphocytes via the low-affinity and high-affinity interaction with its ligand, MAdCAM-1, respectively (6,10,11). Integrin affinity transition is associated with the conformational rearrangement of the integrin molecule. In the resting state, integrin has a low-affinity bent conformation, with the headpiece facing down toward the cell membrane. Upon activation, integrin undergoes a series of conformational rearrangements and extends upward in a switchblade-like opening motion, which leads to the increased integrin affinity (12)(13)(14). Integrin affinity is dynamically regulated by inside-out signals from the cytoplasm. Several intracellular effector molecules, such as talin and kindlins, have been shown to activate integrin through the interaction with integrin cytoplasmic domains (15)(16)(17)(18). In addition to inside-out signaling, extracellular metal ions can also regulate adhesion by integrins (19). Compared with the low-affinity state in Ca 2ϩ ϩ Mg 2ϩ , addition of Mn 2ϩ or removal of Ca 2ϩ strikingly increases the affinity and adhesiveness of almost all integrins (20 -22). Studies have shown that integrin affinity is regulated by divalent cations via a cluster of three divalent cation-binding sites in the integrin ␤ 7 I domain, with the metal ion-dependent adhesion site (MIDAS) at the center and flanked by the synergistic metal ion-binding site and the adjacent to MIDAS (10,11,19,(23)(24)(25). The divalent cation at MIDAS directly coordinates the acidic side chain of Asp-42 in MAdCAM-1 and is essential for both rolling and firm cell adhesion (26). Binding of Ca 2ϩ at the adjacent to MIDAS stabilizes the closed ␤I domain conformation to support rolling adhesion (10). On the contrary, the occupancy of the synergistic metal ion-binding site by divalent cation is required for integrin activation to support firm adhesion (10). In addition, the synergistic metal ion-binding site cation links the specificity-determining loop through a cationinteraction with Phe-185 in the ␤ 7 I domain, which has been reported to be critical for ␣ 4 ␤ 7mediated firm cell adhesion and signaling (27).
Despite the advances in understanding the mechanism by which ␣ 4 ␤ 7 regulates its affinity to support rolling and firm adhesion of lymphocytes, the precise molecular basis for the regulation of low-and high-affinity ␣ 4 ␤ 7 -MAdCAM-1 interactions remains elusive because of lack of the ␣ 4 ␤ 7 ⅐MAdCAM-1 complex structure. As shown from the crystal structure of the ␣ 4 ␤ 7 headpiece, the ␤-propeller domain of ␣ 4 differs from those of the previously characterized ␣ IIb , ␣ V , and ␣ X integrins, especially in the loops on the face of ␤-propeller domain that bind the ␤I domain and contribute to the formation of the ␣ 4 ␤ 7 ligand-binding pocket (28) (Fig. 1A). A particular one is the W1 ␤4-␤1 loop, which is stabilized by a disulfide bond that exists exclusively in ␣ 4 /␣ 9 subfamily (28) (Fig. 1). Considering the structure specificity of this loop, we hypothesize that this disulfide bond-stabilized W1 ␤4-␤1 loop might contribute to unique two-phase cell adhesion mediated by ␣ 4 ␤ 7 . Here we demonstrated that the disulfide bond-stabilized W1 ␤4-␤1 loop is essential for rolling cell adhesion mediated by the low-affinity interaction between inactive ␣ 4 ␤ 7 and MAdCAM-1 but not for firm cell adhesion supported by the high-affinity interaction between Mn 2ϩ -activated ␣ 4 ␤ 7 and MAdCAM-1. Either breaking the disulfide bond or deleting the disulfide bond-occluded segment in the W1 ␤4-␤1 loop not only blocked the global conformational rearrangement and activation of ␣ 4 ␤ 7 triggered by talin or phorbol-12-myristate-13-acetate (PMA) via insideout signaling but also disrupted integrin outside-in signaling. Thus, the disulfide bond-stabilized W1 ␤4-␤1 loop is a novel regulatory element of integrin affinity and bidirectional signaling and plays an essential role in supporting the ␣ 4 ␤ 7 -mediated rolling adhesion.

EXPERIMENTAL PROCEDURES
cDNA Construction and Expression-The ␣ 4 site-directed mutations were generated using QuikChange (Stratagene). WT human ␣ 4 cDNA in vector pcDNA3.1/Hygro(-) (Invitrogen) was used as the template. cDNA of the human talin head domain (talin 1-435) was cloned into vector pmCherry-C1 (modified from vector pEGFP-C1) to generate a construct of the talin head domain with N-terminal fused mCherry. All constructs were confirmed by DNA sequencing.
Flow Cytometry-Immunofluorescence flow cytometry was done as described (31). The expression level of integrin ␣ 4 ␤ 7 on transient 293T transfectants was determined by staining with mAb FIB504 and, subsequently, staining with Alexa Fluor 488conjugated mAb goat anti-rat IgG. The expression level of integrin ␣ 4 ␤ 7 on stable CHO-K1 transfectants was determined by staining with PE-conjugated FIB504. Stained cells were then measured using FACSCalibur (BD Biosciences) and analyzed using WinMDI software.
Flow Chamber Assay-The flow chamber assay was performed as described (10,11). A polystyrene Petri dish was coated with a 5-mm-diameter, 20-l spot of 10 g/ml purified h-MAdCAM-1/Fc in coating buffer (PBS, 10 mM NaHCO 3 (pH 9.0)) for 1 h at 37°C, followed by 2% BSA in coating buffer for 1 h at 37°C to block nonspecific binding sites. Cells were washed twice with HBS (20 mM Hepes (pH 7.4)) containing 5 mM EDTA and 0.5% BSA and resuspended at 1 ϫ 10 7 /ml in buffer A (HBS, 0.5% BSA) and kept at room temperature. Cells were diluted to 1 ϫ 10 6 /ml in buffer A containing different divalent cations immediately before infusion in the flow chamber using a Harvard apparatus programmable syringe pump. Cells were allowed to accumulate for 30 s at 0.3 dyne/cm 2 and for 10 s at 0.4 dyne/cm 2 . Then, shear stress was increased every 10 s from 1 dyne/cm 2 up to 32 dynes/cm 2 in 2-fold increments. The number of cells remaining bound at the end of each 10-s interval was determined. The rolling velocity at each shear stress was calculated from the average distance traveled by rolling cells in 3 s. A velocity of 1 m/s, which corresponds to a movement of 1/2 cell diameter during the 3 s measurement interval, was the minimum velocity required to define a cell as rolling instead of firmly adherent. For the experiment of stimulation, 0.1 M PMA (final concentration) was added and incubated for 10 min at 37°C before cells were infused into the flow chamber.
FRET Assay-FRET was measured as described (27). For detecting the orientation of the integrin ectodomain relative to the cell membrane, 293T transient transfected cells were seeded on a poly-L-lysine-coated (100 g/ml) surface in serumfree DMEM and incubated for 30 min at 37°C. 0.5 mM Mn 2ϩ or 0.1 M PMA was added to activate integrin. Adherent cells were fixed with 3.7% paraformaldehyde for 15 min at room temperature, and nonspecific sites were blocked by incubation with 10% serum-rich medium for 10 min at room temperature. Then, cells were stained with 20 g/ml Alexa Fluor 488-conju-gated Act-1 Fab for 40 min at 37°C. After two washes, cells were labeled with 10 M FM4 -64 FX (Invitrogen) for 4 min on ice, washed once, and mounted immediately with Mowiol 4 -88 (Polysciences, Inc.) mounting solution under a coverslip. The mounted slides were kept in the dark and subjected to photobleach FRET acquisition by a confocal microscope (TCS SP5, Leica). FRET efficiency (E) was calculated as E ϭ 1- Pre and Fdonor(d) Post are the mean donor emission intensity of pre-and post-photobleaching.
Cell Spreading and Microscopy-Cell spreading was done as described (32). Glass coverslips were coated with 100 g/ml poly-L-lysine or 10 g/ml h-MAdCAM-1/Fc overnight at 4°C and blocked by 2% BSA for 1 h at 37°C. CHO-K1 stable cells were plated on coated coverslips for 2 h at 37°C and then fixed by 3.7% paraformaldehyde. 0.5 mM Mn 2ϩ was added to the Ham's F12 medium during the spreading if needed. Differential interference contrast and interference reflection microscopy were conducted on an Olympus IX71 microscope with a ϫ63 oil objective coupled to the Retiga Exi Fast 1394 camera (Q-Imaging). For the quantification of cell spreading, outlines of 50 randomly selected adherent cells from each of three separate experiments were generated, and the number of pixels contained within each of these regions was measured by using Image-Pro plus v. 6.0.

RESULTS
The Disulfide Bond in the W1 ␤4-␤1 Loop Is Required for Rolling Cell Adhesion Mediated by Low-affinity ␣ 4 ␤ 7 -MAdCAM-1 Interaction-The crystal structure of the ␣ 4 ␤ 7 headpiece has shown that the W1 ␤4-␤1 loop in the ␤-propeller domain of the␣ 4 subunit contains a unique disulfide bond between Cys-81 and Cys-85 ( Fig. 1, A and B). To investigate the role of this disulfide bond-stabilized loop in rolling and firm cell adhesion mediated by human ␣ 4 ␤ 7 , we first substituted Cys-81 and Cys-85 with Ser individually (C81S and C85S) or together (C2S) to break the disulfide bond. The WT and mutant ␣ 4 ␤ 7 were transiently expressed in 293T cells at comparable levels (data not shown), and the adhesive behaviors of these transfectants in shear flow were characterized in a parallel wall flow chamber with human MAdCAM-1 (h-MAdCAM-1/Fc) absorbed to its lower wall. The shear stress was increased incrementally, and the velocity of the cells remaining bound at each increment was determined. The WT ␣ 4 ␤ 7 transfectants behaved as described previously for lymphoid cells expressing ␣ 4 ␤ 7 (22). In 1 mM Ca 2ϩ /Mg 2ϩ , about 88% of the bound cells rolled at a wall shear stress of 2 dynes/cm 2 ( Fig. 2A). In contrast, cells were firmly adherent in 0.5 mM Mn 2ϩ (Fig. 2B). As controls, WT ␣ 4 ␤ 7 transfectants treated with ␣ 4 ␤ 7 blocking antibody Act-1 or with EDTA did not accumulate on MAdCAM-1 substrates (Fig. 2, A and B). All three disulfide bond mutations (C81S, C85S, and C2S) significantly impaired rolling adhesion in 1 mM Ca 2ϩ /Mg 2ϩ . Compared with WT ␣ 4 ␤ 7 transfectants, the number of bound cells decreased by 84% for C81S, 90% for C85S, and 90% for C2S, respectively ( Fig. 2A). In contrast, these mutations showed much less effect on firm cell adhesion in 0.5 mM Mn 2ϩ (Fig. 2B). Thus, the disulfide bond in the W1 ␤4-␤1 loop of the ␣ 4 subunit is essential for the inactive ␣ 4 ␤ 7 to support rolling cell adhesion but not indispensable for the activated ␣ 4 ␤ 7 to mediate firm cell adhesion, suggesting a critical role of the W1 ␤4-␤1 loop in the low-affinity ␣ 4 ␤ 7 -MAdCAM-1 interaction.
Next we investigated the role of the disulfide bond-occluded segment by deleting the three residues simultaneously (Del). As shown in Fig. 2, C and D, the Del mutation abolished the rolling cell adhesion on MAdCAM-1 at 2 dynes/cm 2 in 1 mM Ca 2ϩ / Mg 2ϩ (Fig. 2C). In contrast, it could still mediate decent firm cell adhesion in 0.5 mM Mn 2ϩ (Fig. 2D). Thus, the disulfide bond-occluded segment in the W1 ␤4-␤1 loop is critical for the low-affinity ␣ 4 ␤ 7 -MAdCAM-1 interaction but not for the highaffinity ␣ 4 ␤ 7 -MAdCAM-1 interaction.
The Disulfide Bond-stabilized W1 ␤4-␤1 Loop Is Required for Stable Interaction between Low-affinity ␣ 4 ␤ 7 and MAdCAM-1-To study the influence of the W1 ␤4-␤1 loop on the strength of ␣ 4 ␤ 7 -mediated cell adhesion to MAdCAM-1, we examined the resistance to detachment by increasing wall shear stress (Fig. 3, A-D). In 1 mM Ca 2ϩ /Mg 2ϩ , C81S, C85S, and C2S mutant transfectants detached much more rapidly from MAdCAM-1 than WT ␣ 4 ␤ 7 transfectants (Fig. 3A), suggesting a less stable interaction between low-affinity ␣ 4 ␤ 7 and MAdCAM-1 because of the loss of the disulfide bond in the W1 ␤4-␤1 loop. Consistent with the effects of the W1 ␤4-␤1 loop single-residue mutations on cell adhesion (Fig. 2C), cell resistance to detachment in 1 mM Ca 2ϩ /Mg 2ϩ was mostly decreased by G82A and least affected by K83A, with T84A in between (Fig. 3C). In contrast to the results in 1 mM Ca 2ϩ /Mg 2ϩ , all of the above mutant-expressing cells showed a similar shear resistance as WT ␣ 4 ␤ 7 transfectants in 0.5 mM Mn 2ϩ (Fig. 3, B and D). Thus, these data indicate that the disulfide bond-stabilized W1 ␤4-␤1 loop is required for stable interaction between low-affinity ␣ 4 ␤ 7 and MAdCAM-1 to support efficient rolling adhesion but not indispensible to maintain the stable high-affinity ␣ 4 ␤ 7 -MAdCAM-1 interaction.
To further address the role of the W1 ␤4-␤1 loop in ␣ 4 ␤ 7mediated rolling adhesion, we examined the rolling velocity of ␣ 4 ␤ 7 293T transfectants on MAdCAM-1 at different wall shear stresses (Fig. 3, E and F). In 1 mM Ca 2ϩ /Mg 2ϩ , WT ␣ 4 ␤ 7 transfectants rolled with increasing velocity from 4 to 8 m/s as wall shear stress was increased from 1 to 4 dynes/cm 2 (Fig. 3, E and  F). Compared with WT ␣ 4 ␤ 7 transfectants, the C81S, C85S, C2S, and G82A mutant transfectants showed an obviously increased rolling velocity at each wall shear stress (Fig. 3, E and   F). In addition, the T84A transfectants exhibited a milder increase in rolling velocity, whereas the K83A transfectants showed a slightly increased rolling velocity (Fig. 3F). Taken together, the above data demonstrate that the disulfide bondstabilized W1 ␤4-␤1 loop is required to stabilize the low-affinity ␣ 4 ␤ 7 -MAdCAM-1 interaction for efficient rolling cell adhesion.
The Disulfide Bond-stabilized W1 ␤4-␤1 Loop Is Required for the Activation of ␣ 4 ␤ 7 by Inside-Out Signaling-In addition to the activation by extracellular Mn 2ϩ , integrin can also be activated by intracellular effector proteins, such as talin, via insideout signaling (17,18,33). To investigate the role of the W1 ␤4-␤1 loop in the activation of ␣ 4 ␤ 7 by inside-out signaling, the talin head domain with N-terminal fused mCherry (mCherrytalin) was overexpressed at a comparable level in WT and mutant ␣ 4 ␤ 7 293T transfectants (Fig. 4A), and the cell adhesion at 2 dynes/cm 2 was examined (Fig. 4B). In 1 mM Ca 2ϩ /Mg 2ϩ , the total and firmly adherent cell numbers of WT ␣ 4 ␤ 7 transfectants were both increased significantly after overexpression of mCherry-talin, indicating the activation of ␣ 4 ␤ 7 by talin (Fig.  4B). In contrast, the number of bound cells expressing either the C2S or Del mutant was not increased by mCherry-talin overexpression (Fig. 4B), suggesting impaired integrin activation via inside-out signaling.
To further confirm the impaired activation of the C2S and Del mutants by inside-out signaling, we tested ␣ 4 ␤ 7 activation by PMA in WT, C2S, or Del mutant ␣ 4 ␤ 7 293T transfectants. PMA is reported to activate integrins through inside-out signaling by activating the protein kinase C kinase pathway (34). Consistently, WT ␣ 4 ␤ 7 , but not the C2S and Del mutants, could be activated by PMA (Fig. 4C). Thus, these results suggest that the disulfide bond-stabilized W1 ␤4-␤1 loop is crucial for the activation of integrin ␣ 4 ␤ 7 by inside-out signaling.
The Disulfide Bond-stabilized W1 ␤4-␤1 Loop Is Required for the Conformational Rearrangement in ␣ 4 ␤ 7 during Integrin Activation-Integrin activation is accompanied by global conformational rearrangements, including the switchblade-like extension of the integrin ectodomain and headpiece opening (12,35). We next used FRET to study the contribution of the W1 ␤4-␤1 loop to integrin conformation. To assess the orientation of the ␣ 4 ␤ 7 ectodomain relative to the plasma membrane, ␣ 4 ␤ 7 was labeled with Alexa Fluor 488-conjugated Act-1 Fab fragment as a donor, which binds to the top of the ␤ 7 I domain (36). The plasma membrane was labeled with FM4 -64 FX as an acceptor (24). In 1 mM Ca 2ϩ /Mg 2ϩ , all WT and mutant (C2S and Del) ␣ 4 ␤ 7 transfectants showed a similar and relatively high FRET efficiency, suggesting the bent conformation of inactive integrins (Fig. 5). Activation of WT ␣ 4 ␤ 7 with 0.5 mM Mn 2ϩ significantly decreased the FRET efficiency, suggesting the extension of the ␣ 4 ␤ 7 ectodomain (Fig. 5A). Interestingly, the FRET efficiency of the C2S and Del mutant ␣ 4 ␤ 7 transfectants was much higher than that of the WT ␣ 4 ␤ 7 transfectants in Mn 2ϩ , suggesting that Mn 2ϩ induces less of an extension of the C2S and Del mutants than WT ␣ 4 ␤ 7 (Fig. 5A). Thus, removal of either the disulfide bond or the disulfide bond-occluded segment impairs the global conformational changes of integrin ␣ 4 ␤ 7 induced by Mn 2ϩ .
Additionally, we examined the conformational change of ␣ 4 ␤ 7 activated by talin or PMA via inside-out signaling (Fig. 5, B and C). Both overexpression of mCherry-talin and stimulation with PMA significantly decreased the FRET efficiency of WT ␣ 4 ␤ 7 transfectants in 1 mM Ca 2ϩ /Mg 2ϩ (Fig. 5, B and C). In contrast, the FRET efficiency of the C2S and Del mutant ␣ 4 ␤ 7 transfectants did not show a significant decrease after mCherry-talin overexpression or PMA treatment (Fig. 5, B and C). These results clearly suggest that the disulfide bond-stabilized W1 ␤4-␤1 loop is essential for the global conformational change during integrin activation via inside-out signaling. Collectively, these data demonstrate that the disulfide bond-stabilized W1 ␤4-␤1 loop in the ␣ 4 ␤-propeller domain is crucial for the global conformational rearrangement coupled with ␣ 4 ␤ 7 activation.
The Disulfide Bond-stabilized W1 ␤4-␤1 Loop Is Essential for ␣ 4 ␤ 7 -mediated Outside-In Signaling-Integrin ligand binding can trigger outside-in signaling to activate multiple intracellular signal proteins, which leads to cytoskeleton rearrangements to support cell spreading (1,33,(37)(38)(39). To assess the requirement of the W1 ␤4-␤1 loop for integrin-mediated outside-in signaling, CHO-K1 cells stably expressing a similar level of WT or mutant (C2S and Del) ␣ 4 ␤ 7 were established that exhibited similar adhesive behaviors on immobilized MAdCAM-1 at a wall shear stress of 2 dynes/cm 2 as 293T transient transfectants, suggesting that the effects of W1 ␤4-␤1 loop mutations on ␣ 4 ␤ 7 -mediated cell adhesion were not cell type-specific (Fig. 6, A and B). Then, ␣ 4 ␤ 7 -mediated cell spreading was studied (Fig. 6, C-E). WT ␣ 4 ␤ 7 -expressing cells spread substantially on MAdCAM-1 in 1 mM Ca 2ϩ /Mg 2ϩ , with an extensive area of close cell substrate contact (Fig. 6C). In contrast, both C2S and Del mutant ␣ 4 ␤ 7 -expressing cells did not spread on MAdCAM-1 and showed the same area of projection as cells in suspension under this condition (Fig. 6, C and E). Addition of 0.5 mM Mn 2ϩ further enhanced the spreading of WT ␣ 4 ␤ 7expressing cells and enabled the mutant ␣ 4 ␤ 7 transfectants to spread on MAdCAM-1 but with a much smaller cell contact area on the substrates (Fig. 6D). In addition, mutant ␣ 4 ␤ 7 -expressing cells exhibited a significantly decreased area of projection compared with that of the WT ␣ 4 ␤ 7 -expressing cells (Fig.  6E). Taken together, these results show that either breaking the disulfide bond or deleting the disulfide bond-occluded segment in the W1 ␤4-␤1 loop leads to deficient cell spreading, suggesting disrupted ␣ 4 ␤ 7 outside-in signaling.

DISCUSSION
The ability of integrin ␣ 4 ␤ 7 to mediate rolling and firm cell adhesion through low-and high-affinity interaction with MAdCAM-1 makes it unique among most integrins that only mediate firm cell adhesion after activation. In this study, we demonstrate that a unique disulfide bond-stabilized W1 ␤4-␤1 loop in the ␤-propeller domain of the ␣ 4 subunit plays an essential role in low-affinity ␣ 4 ␤ 7 -MAdCAM-1 interaction that supports rolling cell adhesion but is not indispensable for firm cell adhesion supported by high-affinity interaction between Mn 2ϩ -activated ␣ 4 ␤ 7 and MAdCAM-1. Moreover, removal of either the disulfide bond or the disulfide bond-occluded short segment not only blocked the global conformational rearrangement and activation of ␣ 4 ␤ 7 triggered by talin or PMA via inside-out signaling but also disrupted integrin outside-in signaling and led to deficient ␣ 4 ␤ 7 -mediated cell spreading. Thus, the disulfide bond-stabilized W1 ␤4-␤1 loop in the ␣ 4 ␤-propeller domain plays an essential role in supporting the rolling cell adhesion mediated by ␣ 4 ␤ 7 and functions as a novel regulatory element of integrin affinity and bidirectional signaling.
According to the crystal structures of integrins, the ligandbinding site of ␣ 4 ␤ 7 is in an overall distinct shape from those of ␣ V ␤ 3 , ␣ IIb ␤ 3 , and ␣ 5 ␤ 1 . The crevice running along the ␣-␤ subunit in ␣ 4 ␤ 7 ligand-binding site is longer, wider, and deeper than those in ␣ V ␤ 3 , ␣ IIb ␤ 3 , and ␣ 5 ␤ 1 , which is contributed largely by the loops on the face of the ␤-propeller domain that bind the ␤I domain and form the ligand-binding site (28). Among the loops, the W1 ␤4-␤1 loop is unique, as it contains a disulfide bond existing only in the integrin ␣ 4 /␣ 9 subfamily. Here we demonstrated that breaking the disulfide bond or deleting the disulfide bond-occluded segment of the W1 ␤4-␤1 loop significantly impaired rolling adhesion mediated by lowaffinity ␣ 4 ␤ 7 on MAdCAM-1 (Fig. 2). Considering the importance of disulfide bonds in stabilizing the three-dimensional structures of proteins (40), our data suggest that the disulfide bond contained in the W1 ␤4-␤1 loop may serve to maintain the optimal conformation critical for the interaction between low-affinity ␣ 4 ␤ 7 and MAdCAM-1. Moreover, as revealed by our results, the three amino acid residues occluded in the disulfide bond exerted different effects on integrin low-affinity ligand binding, with Gly-82 Ͼ Thr-84 Ͼ Lys-83 (Fig. 2). Given the fact that glycine is the amino acid frequently found in ␤-turns, it is tempting to speculate that Gly-82 contributes to keeping the right orientation of the W1 ␤4-␤1 loop to support the optimal interaction between low-affinity ␣ 4 ␤ 7 and MAdCAM-1. Furthermore, because the hydrophilic side chain of threonine with its hydroxyl group helps form hydrophilic interactions, such as hydrogen bonding, it supports the role for threonine in stabilizing the low-affinity ␣ 4 ␤ 7 -MAdCAM-1 binding.
In addition to MAdCAM-1, integrin ␣ 4 ␤ 7 can also bind another ligand, vascular cell adhesion molecule 1 (VCAM-1), and mediate rolling adhesion on this substrate before activation (2,41). To address whether the W1 ␤4-␤1 loop is also involved in ␣ 4 ␤ 7 -VCAM-1 binding, we tested the influence of the W1 ␤4-␤1 loop mutations on ␣ 4 ␤ 7 -mediated cell adhesion on immobilized VCAM-1 in shear flow. Consistent with the results on MAdCAM-1, either breaking the disulfide bond or deleting the disulfide bond-occluded segment inhibited the rolling cell adhesion on VCAM-1 in 1 mM Ca 2ϩ /Mg 2ϩ but barely affected the firm cell adhesion in 0.5 mM Mn 2ϩ (Fig. 7A). These results indicate that the disulfide bond-stabilized W1 ␤4-␤1 loop is also essential for the binding of low-affinity ␣ 4 ␤ 7 to VCAM-1.
One interesting finding of our study is that the disulfide bond-stabilized W1 ␤4-␤1 loop is required for talin-or PMAmediated ␣ 4 ␤ 7 activation but not indispensible for Mn 2ϩ -induced ␣ 4 ␤ 7 activation and firm cell adhesion. This difference could be attributed to the fact that the mechanisms of integrin activation induced by Mn 2ϩ and talin/PMA are different. Mn 2ϩ activates integrins by direct binding to the metal ion binding sites in the ␤I domain, which triggered integrin activation independently of cytoplasmic signaling (19,42,43), whereas talin or PMA activate integrin via inside-out signaling by regulating the binding of intracellular effector molecules to integrin cytoplasmic domains, which triggers the global conformational rearrangement and activation of integrin (15,17,18,33,34). In addition, FRET analysis of the distance between the integrin head domain and the cell membrane in this study also demonstrated that integrin ␣ 4 ␤ 7 stimulated by PMA or overexpression of talin were less extended than Mn 2ϩ -activated ␣ 4 ␤ 7 (Fig.  5), suggesting the different conformations of these integrins. Thus, it implies that integrin ␣ 4 ␤ 7 activated by Mn 2ϩ or talin/ PMA might have distinct requirements for the W1 ␤4-␤1 loop. Another notable finding of our study is that the disulfide bond-stabilized W1 ␤4-␤1 loop in the ␣ 4 ␤-propeller domain is crucial for the global conformational rearrangement of ␣ 4 ␤ 7 triggered by inside-out signaling. Studies have shown that the local conformational changes in the integrin head domain are closely associated with its global conformational rearrangements (27, 31, 44 -46). Mutations around the ligand binding site of integrin could affect the global conformational changes of integrin molecules (27, 31, 44 -46). In this study, as the W1 ␤4-␤1 loop is located on the face of the ␤-propeller domain, which binds the ␤I domain and forms the ligand-binding site of integrin, it is tempting to speculate that deletion or mutations of this loop could exert some effects on the ligand-binding interface formed between the ␣ subunit ␤-propeller domain and the ␤I domain, which might affect the local conformational changes in the head domain of integrin that are required by the global conformational rearrangements triggered by inside-out signaling.
This study also finds that the disulfide bond-stabilized W1 ␤4-␤1 loop is required for ␣ 4 ␤ 7 -mediated cell spreading and outside-in cell signaling (Fig. 6). Integrin-mediated cell spreading is a complex process that involves diverse signaling networks (47). One of the early steps in transducing extracellular cues through integrins to the cytoskeleton is the activation of FAK and paxillin (47). Our study showed that the W1 ␤4-␤1 loop mutations decreased the activation of both FAK and pax-  . p values were calculated by one-way ANOVA with Dunnett post-tests. ***, p Ͻ 0.001. C and D, differential interference contrast and interference reflection microscopy images of CHO-K1 stable transfectants that adhered to immobilized MAdCAM-1 in 1 mM Ca 2ϩ / Mg 2ϩ (C) or 0.5 mM Mn 2ϩ (D). The images are representatives from one of three independent experiments. Scale bars ϭ 50 m. E, quantification of cell spreading area (projection on substrates) of ␣ 4 ␤ 7 CHO-K1 transfectants on the basis of differential interference contrast images. Data are mean Ϯ S.E. (n ϭ 50). p values were calculated by two-way ANOVA with Bonferroni post-tests. ***, p Ͻ 0.001. F, CHO-K1 cells stably expressing WT or mutant ␣ 4 ␤ 7 were plated on poly-L-lysine in serum-free Ham's F12 medium or on MAdCAM-1 in Ham's F12 medium containing 1 mM Ca 2ϩ /Mg 2ϩ or 0.5 mM Mn 2ϩ for 2 h, lysed, and blotted for indicated molecules. Phosphorylated Y397-FAK and Y118-paxillin were blotted as indications of FAK and paxillin activation, respectively. A representative result of three independent experiments is shown.
illin, indicating that ␣ 4 ␤ 7 -mediated outside-in signaling also requires the normal function of the W1 ␤4-␤1 loop. Thus, the disulfide bond-stabilized W1 ␤4-␤1 loop might contribute to a critical interaction between ␣ 4 ␤ 7 and the ligands, which is important for integrin outside-in signaling. In conclusion, our findings reveal that the unique disulfide bond-stabilized W1 ␤4-␤1 loop in ␣ 4 ␤-propeller domain functions as a novel regulatory element of integrin affinity and signaling and uncover a particular molecular basis for ␣ 4 ␤ 7 -mediated rolling cell adhesion.