Identification, Characterization, and Epitope Mapping of Human Monoclonal Antibody J19 That Specifically Recognizes Activated Integrin α4β7*

Background: The activation-specific antibodies against integrins are powerful tools in integrin studies. Results: An activation-specific antibody J19 against α4β7 was identified and characterized. Conclusion: J19 specifically binds to the activated α4β7 by recognizing the epitope only exposed in extended conformation. Significance: J19 is a potentially powerful tool for studying α4β7 function and treatment of α4β7-related inflammatory diseases. Integrin α4β7 is a lymphocyte homing receptor that mediates both rolling and firm adhesion of lymphocytes on vascular endothelium, two of the critical steps in lymphocyte migration and tissue-specific homing. The rolling and firm adhesions of lymphocytes rely on the dynamic shift between the inactive and active states of integrin α4β7, which is associated with the conformational rearrangement of integrin molecules. Activation-specific antibodies, which specifically recognize the activated integrins, have been used as powerful tools in integrin studies, whereas there is no well characterized activation-specific antibody to integrin α4β7. Here, we report the identification, characterization, and epitope mapping of an activation-specific human mAb J19 against integrin α4β7. J19 was discovered by screening a human single-chain variable fragment phage library using an activated α4β7 mutant as target. J19 IgG specifically bound to the high affinity α4β7 induced by Mn2+, DTT, ADP, or CXCL12, but not to the low affinity integrin. Moreover, J19 IgG did not interfere with α4β7-MAdCAM-1 interaction. The epitope of J19 IgG was mapped to Ser-331, Ala-332, and Ala-333 of β7 I domain and a seven-residue segment from 184 to 190 of α4 β-propeller domain, which are buried in low affinity integrin with bent conformation and only exposed in the high affinity extended conformation. Taken together, J19 is a potentially powerful tool for both studies on α4β7 activation mechanism and development of novel therapeutics targeting the activated lymphocyte expressing high affinity α4β7.

Integrins are a family of ␣/␤ heterodimeric adhesion receptors that mediate cell-cell, cell-extracellular matrix, and cellpathogen interactions and transmit signals bidirectionally across plasma membrane (1). Because of their unique function of integrating extracellular environment with cytoskeleton, integrins play important roles in adhesion-dependent cellular processes including cell migration, proliferation, survival, and differentiation (1)(2)(3)(4). Integrin ␣ 4 ␤ 7 is a lymphocyte homing receptor, which can mediate both rolling and firm adhesion of lymphocytes, two of the critical steps in lymphocytes homing to the intestine and gut-associated lymphoid tissues (5,6). Its ligand, mucosal addressin cell adhesion molecule-1 (MAdCAM-1), 3 is preferentially expressed on high endothelial venules of gut-associated lymphoid organs and on lamina propria venules, helping lymphocyte traffic to mucosal organs (7). The activation of integrin ␣ 4 ␤ 7 is a critical step in the progression of inflammatory bowel disease (8,9). Thus, ␣ 4 ␤ 7 is a promising therapeutic target for the treatment of inflammatory bowel disease.
Different from most integrins supporting only firm adhesion of cells upon activation, integrin ␣ 4 ␤ 7 can mediate both rolling and firm adhesion of lymphocytes (10,11). The resting integrin ␣ 4 ␤ 7 supports rolling adhesion of lymphocytes via its low affinity interaction with MAdCAM-1. Upon activation, ␣ 4 ␤ 7 binds to MAdCAM-1 in high affinity, which results in firm cell adhesion. The transition from rolling adhesion to firm adhesion is regulated by the shift of integrin from low affinity to high affinity state (11,12). Affinity regulation is associated with the conformational rearrangement of the integrin molecule (13,14). Previous studies have shown that integrin extracellular domains exist in at least three distinct global conformational states that differ in affinity for ligand: low affinity bent conformation with a closed headpiece, intermediate affinity extended conformation with a closed headpiece, and high affinity extended conformation with an open headpiece (15)(16)(17)(18)(19). The equilibrium among these different states is regulated by integrin inside-out signaling and extracellular stimuli, such as divalent cations (20,21). Compared with the low affinity state in Ca 2ϩ ϩ Mg 2ϩ , removal of Ca 2ϩ or addition of Mn 2ϩ strikingly increases ligand binding affinity of almost all integrins (11). Electron micrographic studies of integrins ␣ V ␤ 3 and ␣ 5 ␤ 1 demonstrate that integrin activation is coupled with the switchblade-like extension of the extracellular domain and a change in angle between the ␤I and hybrid domains (18,19). Crystal structures of integrin ␣ IIb ␤ 3 headpiece in the high affinity conformation demonstrate that the C-terminal ␣7-helix of the ␤I domain moves axially toward the hybrid domain, causing the ␤ hybrid domain to swing outward by 60°, away from the ␣ subunit (15,22). The conformational rearrangement in the integrin headpiece destabilizes the bent conformation and induces integrin extension in which the headpiece extends and breaks free from an interface with the leg domains that connect it to the plasma membrane. This conversion from the low affinity to the high affinity conformation of integrin can be mimicked by the introduction of glycan wedges into the interface between the hybrid and the I domains of ␤ 7 , ␤ 3 , and ␤ 1 integrins, which activates integrins by stabilizing the outward swing of the hybrid domain and the high affinity headpiece conformation (12,23). This wedge mutant integrin therefore can be used as a target for screening activation-specific antibodies that exclusively recognize the activated integrin. Up to date, there is no well characterized activation-specific antibody against integrin ␣ 4 ␤ 7 . Thus, a well characterized activation-specific antibody to ␣ 4 ␤ 7 will be extremely useful for both studies on ␣ 4 ␤ 7 activation mechanism and development of drug delivery system targeting the activated lymphocyte for the treatment of inflammatory bowel disease.
In this study, we discovered the activation-specific human mAb J19 against integrin ␣ 4 ␤ 7 by screening from the human scFv phage library using the activated wedge mutant ␣ 4 ␤ 7 as a target. J19 IgG specifically bound to the ␣ 4 ␤ 7 activated by Mn 2ϩ , DTT, ADP, or CXCL12 but not to the low affinity integrin. Moreover, J19 IgG did not interfere with ␣ 4 ␤ 7 -MAdCAM-1 interaction, suggesting it was a mAb distinct from "ligand mimetic" group, and we demonstrate that J19 IgG recognizes an activation-dependent epitope on ␣ 4 ␤ 7 consisting of residues from both ␣ 4 and ␤ 7 subunits. This epitope is buried in the low affinity integrin with bent conformation and only exposed in the extended conformation induced by integrin activation. These data also provide strong supporting evidence for the conformational rearrangement during integrin ␣ 4 ␤ 7 activation.
Plasmid Construction and Transient Transfection into 293T Cells-J19 IgG expression constructs were built on the backbone of pIRES2-EGFP (Invitrogen). cDNAs encoding human IgG1 light chain and heavy chain constant regions were amplified by RT-PCR from human endothelial cell total mRNA. cDNAs of human ␣ 4 , ␣ E , ␤ 7 , and ␤ 1 subunits were inserted into vector pcDNA3.1/Hygro (Ϫ) (Invitrogen), respectively. Chimeric ␤ 1 /␤ 7 subunits were generated by overlap extension PCR (26,27). Chimeric ␣ 4 /␣ E subunits were generated by an improved PCR mutagenesis strategy for sequence swapping (28). Chimeras were named according to the origin of their segments. For example, ␤ 7 542␤ 1 indicates that residues 1-542 are from ␤ 7 subunit and residues 543 to the C terminus are from the ␤ 1 subunit. Amino acid sequence numbering was according to the mature ␣ 4 or ␤ 7 sequence. The ␤ 7 site-directed mutations were generated by using QuikChange (Stratagene). All of the constructs were confirmed by DNA sequencing. Transient transfection of 293T cells was performed as described (11).
Protein Purification and Analytical Gel Filtration-The J19 IgG was purified by protein A (Pierce) affinity chromatography. 293T cells were transiently transfected with J19 IgG expressing construct and cultured in DMEM supplemented with 10% "ultralow IgG" fetal bovine serum (Invitrogen).
Analytical gel filtrations were performed using precalibrated Superdex 200 on an ÄKTA purifier system running Unicorn 5.11 software at a flow rate of 0.5 ml/min at room temperature (29,30). The elution profiles were monitored in-line by UV adsorption at 280 nm. Hepes-buffered saline (HBS) buffer (150 mM NaCl, 20 mM Hepes, pH 7.4) containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ or 0.5 mM Mn 2ϩ was used throughout. 5.2 g of purified WT or wedge mutant integrin ␣ 4 ␤ 7 was loaded.
Selection of Integrin-binding Phages-Human scFv phage library was purchased from Geneservice. Antibody screening was done according to the protocol provided by Geneservice with minor modifications. Soluble integrin ␣ 4 ␤ 7 was biotinylated and immobilized to streptavidin-coated Dynabeads (Invitrogen). Phage displaying scFv binding to ␣ 4 ␤ 7 was captured by integrin-labeled beads (so called "panning"). In each round of panning, the binders to WT ␣ 4 ␤ 7 were depleted first followed by screening the specific scFv fragments against wedge mutant. After three rounds of panning, specific binders were identified by monoclonal phage ELISA.
Flow Cytometry-Immunofluorescence flow cytometry was done as described (31). Before staining with antibody, 2.5 ϫ 10 5 cells were washed with HBS containing 5 mM EDTA and then resuspended in either HBS containing 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ or activating HBS containing 2 mM Mn 2ϩ . For activation of ␣ 4 ␤ 7 by other stimuli than divalent cations, 2.5 ϫ 10 5 cells were resuspended and incubated at 37°C for 15 min in HBS (1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ ) with ADP and DTT at a final concentration of 10 and 500 M, respectively. Stained cells were then measured using FACSCalibur (BD Biosciences) and analyzed using WinMDI 2.9 software.
Primary splenic lymphocytes (SPLs) were isolated as previously described (32) and suspended in HBS and stimulated by 0.2 g/ml CXCL12 (R & D Systems) for 5 min at 37°C. The cells were fixed with an equal volume of 2ϫ formaldehyde (7.4%) before staining with 5 g/ml J19 IgG.
Fluorescence Microscopy-Immunofluorescence staining assay was performed as reported (33). Cover glasses were coated with 10 g/ml human MAdCAM-1 fused to the Fc1 and Fc2 regions of human IgG1 (huMAdCAM-1/Fc) in the presence or absence of 2 g/ml CXCL12. The cells were incubated on the cover glass for 20 min at 37°C, fixed with 3.7% polyformaldehyde, and blocked by 10% FBS. The cells were then stained with 5 g/ml J19 IgG or FIB27, followed by Alexa Fluor 488 goat anti-human IgG (H ϩ L) and Cy3-conjugated goat anti-rat IgG (Invitrogen), respectively.
Flow Chamber Assay-The flow chamber assay was performed as described (11,12). A polystyrene Petri dish to be used as the lower wall of the chamber was coated with a 5-mmdiameter, 20-l spot of 10 g/ml purified huMAdCAM-1/Fc in coating buffer (phosphate-buffered saline, 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. The cells were washed twice with HBS containing 5 mM EDTA and 0.5% BSA, resuspended at 1 ϫ 10 7 /ml in buffer A (HBS, 0.5% BSA), and kept at room temperature. The 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. Then shear stress was increased from 0.3 dyn/cm 2 up to 16 dyn/cm 2 . The number of cells remaining bound at the end of 1 dyn/cm 2 was determined.

Generation of High Affinity Integrin ␣ 4 ␤ 7 with Glycan Wedge
Mutation-Previous studies have shown that the introduction of N-glycan at the integrin ␤I/hybrid domain interface will activate integrins and stabilize the high affinity conformation (12,23). To obtain the high affinity human integrin ␣ 4 ␤ 7 , we introduced an N-glycosylation site at Asn-322 in the ␣4-␤5 loop of ␤ 7 I domain by mutating Gln-324 to Thr as previously described (12). The WT and wedge mutant (Q324T) human ␣ 4 ␤ 7 were transiently expressed in 293T cells, and the adhesive behavior in shear flow of those transfectants was characterized by allowing them to adhere to MAdCAM-1 in a parallel wall flow chamber. WT ␣ 4 ␤ 7 293T transfectants behaved as previously described for lymphoid cells expressing ␣ 4 ␤ 7 (36). In 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ , more than 75% of adherent cells expressing WT ␣ 4 ␤ 7 rolled on MAdCAM-1 substrates at the wall shear stress of 1 dyn/cm 2 (Fig. 1A). In contrast, the cells were firmly adherent in 0.5 mM Mn 2ϩ (Fig. 1A). Rolling and firm adhesions represent the low and high affinity interactions of integrin ␣ 4 ␤ 7 with MAdCAM-1, respectively. By contrast with the rolling adhesion of WT ␣ 4 ␤ 7 293T transfectants, 293T transfectants expressing ␣ 4 ␤ 7 wedge mutant showed significantly increased firmly adherent cells in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ , which is similar to the adhesion behavior of WT ␣ 4 ␤ 7 activated by 0.5 mM Mn 2ϩ (Fig. 1A). ␣ 4 ␤ 7 transfectants treated with the ␣ 4 ␤ 7 blocking antibody Act-1 did not accumulate on MAdCAM-1 substrates. Thus, integrin ␣ 4 ␤ 7 is constitutively activated by the glycan wedge introduced into the I/hybrid domain interface of ␤ 7 subunit.
Next, we introduced Q324T mutation into the human integrin ␣ 4 ␤ 7 soluble construct with all ectodomains. To eliminate the heterogeneity resulting from partial cleavage of ␣ 4 subunit, a previously described Arg-558 to Ala mutation was introduced into ␣ 4 subunit to remove the protease cleavage site (37). Both WT and wedge mutant ␣ 4 ␤ 7 -soluble proteins were expressed in 293T cells and purified by nickel-nitrilotriacetic acid, Strep-Tactin affinity chromatography, and gel filtration. To study the conformational change of integrin in solution, the isocratic elution profiles of purified WT and wedge mutant ␣ 4 ␤ 7 were compared using analytical gel filtration. Previous studies have shown that the shape change of integrin from the low affinity bent conformation to high affinity extended conformation could lead to the increase in hydrodynamic radius of integrin and the decrease of retention volume in gel filtration (19,30). As expected, the low affinity WT ␣ 4 ␤ 7 was eluted at 10.83 ml in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ , whereas the elution volume of high affinity WT ␣ 4 ␤ 7 in 0.5 mM Mn 2ϩ decreased to 10.67 ml, suggesting the more extended high affinity conformation of ␣ 4 ␤ 7 activated by Mn 2ϩ (Fig. 1B). Similar to WT ␣ 4 ␤ 7 in Mn 2ϩ , the wedge mutant ␣ 4 ␤ 7 showed decreased retention volume (10.72 ml) in Ca 2ϩ ϩ Mg 2ϩ compared with WT ␣ 4 ␤ 7 , suggesting the extended conformation of wedge mutant ␣ 4 ␤ 7 . Thus, the wedge mutant can mimic the high affinity ␣ 4 ␤ 7 by stabilizing the extended conformation of integrin.
Identification of Human Antibody to Activated Integrin ␣ 4 ␤ 7 -To identify an activation-specific antibody against ␣ 4 ␤ 7 , we performed human single fold scFv phage display library (Tomlinson I ϩ J) selection using the high affinity wedge mutant ␣ 4 ␤ 7 as target. The Tomlinson I ϩ J libraries purchased from Geneservice contain over 100 million different human scFv fragments. The library was first depleted by the low affinity WT ␣ 4 ␤ 7 soluble protein and then selected against the high affinity wedge mutant ␣ 4 ␤ 7 . In this way, binders specific for high affinity ␣ 4 ␤ 7 were enriched after each round of selection. After three rounds of selections, the remaining binders were validated by monoclonal phage ELISA using purified WT and wedge mutant protein as target, respectively. Among a number of isolates that showed higher binding signals to wedge mutant than to WT ␣ 4 ␤ 7 (data not shown), phage clone J19 bound specifically to wedge mutant and high affinity WT ␣ 4 ␤ 7 activated by 2 mM Mn 2ϩ , but not to the low affinity WT ␣ 4 ␤ 7 in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ (Fig. 2A).
To further characterize the J19 scFv phage isolate, the J19 scFv was expressed and purified. The binding of purified J19 scFv to integrin ␣ 4 ␤ 7 was analyzed using flow cytometry with K562 cells stably expressing human ␣ 4 ␤ 7 (Fig. 2B). Different from the specificity of J19 phage for high affinity ␣ 4 ␤ 7 , J19 scFv showed similar binding to the low affinity ␣ 4 ␤ 7 in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ and the high affinity ␣ 4 ␤ 7 in 2 mM Mn 2ϩ . The binding of J19 scFv to ␣ 4 ␤ 7 was comparable with that of ␣ 4 ␤ 7 mAb Act-1 (Fig. 2B). Considering that scFv is of a smaller size than the phage particle, it is tempting to speculate that the binding sites of J19 in low affinity ␣ 4 ␤ 7 can be accessed by smaller scFv but not the larger phage particle expressing the same scFv.
In addition to the strong activation by Mn 2ϩ , integrin can also be activated by other stimuli like DTT and ADP (38 -40). DTT has been shown to activate integrin in a number of systems (38,39). ADP was reported to induce integrin activation through inside-out signaling by activating PI3K pathway (41). To further study the binding specificity of J19 IgG to ␣ 4 ␤ 7 activated by different stimuli, K562 cells stably expressing ␣ 4 ␤ 7 were treated with Mn 2ϩ , DTT, or ADP and then followed by staining with 5 g/ml J19 IgG. As shown in Fig. 3D, J19 IgG specifically bound to K562 ␣ 4 ␤ 7 cells treated with these stimuli. By contrast, J19 IgG did not bind to the same cells without any stimulation (Fig. 3D). These results suggested that epitope recognized by J19 IgG in integrin ␣ 4 ␤ 7 was only expressed after integrin activation. Moreover, the binding of J19 IgG to K562 ␣ 4 ␤ 7 cells treated with Mn 2ϩ is higher in comparison with those stimulated by DTT or ADP. The different expression level of J19 epitope induced by the above stimuli could be due to the different activation states and conformations of integrin ␣ 4 ␤ 7 .
We next test the cross-reactivity of J19 IgG with ␣ 4 ␤ 7 from other species. The mouse and rat ␣ 4 ␤ 7 were transiently expressed in 293T cells, respectively. The expression level of ␣ 4 ␤ 7 was determined using mAb FIB27 against human and mouse ␤ 7 and mAb HP2/1 against rat ␣ 4 . J19 IgG showed comparable binding to the activated human, mouse, and rat ␣ 4 ␤ 7 but not to inactive ones (Fig. 4B).
J19 IgG Specifically Binds to Chemokine-activated Mouse SPLs-Having shown that J19 IgG was specific for the activated ␣ 4 ␤ 7 expressed on K562 and 293T cell lines, we next tested whether J19 IgG also bound to the high affinity form of ␣ 4 ␤ 7 expressed on primary lymphocytes. The mouse SPLs that highly express ␣ 4 ␤ 7 were isolated, and the activation of integrin by CXCL12 stimulation was done as previously described (33,42). In flow cytometric assay, binding of J19 IgG to SPLs was undetectable before CXCL12 treatment, whereas it significantly increased 5 min after adding 0.2 g/ml CXCL12 (Fig.  5A). Similar results were obtained by immunostaining with J19 IgG and ␤ 7 mAb FIB27 (Fig. 5B). J19 IgG signal was only observed at the surface of SPLs after ␣ 4 ␤ 7 was activated by CXCL12 or Mn 2ϩ , whereas FIB27 bound to ␣ 4 ␤ 7 in an activation-independent manner. Thus, J19 IgG specifically recognizes the activated ␣ 4 ␤ 7 of primary lymphocytes.
Effect of J19 IgG on ␣ 4 ␤ 7 -MAdCAM-1 Interaction-To further characterize J19 IgG, we studied the effect of this antibody on cell adhesion mediated by ␣ 4 ␤ 7 -MAdCAM-1 interaction in shear flow using K562 cells stably expressing human ␣ 4 ␤ 7 . The cells were preincubated with J19 IgG or ␣ 4 ␤ 7 blocking mAb Act-1 and then infused into a parallel wall flow chamber with MAdCAM-1 immobilized on the lower wall. 20 g/ml Act-1 almost completely abolished the cell adhesion at the wall shear stress of 1 dyn/cm 2 , whereas J19 IgG showed no effect on cell adhesion even at much higher concentration of 100 g/ml (Fig.  6). Thus, J19 IgG does not affect the interaction between integrin ␣ 4 ␤ 7 and MAdCAM-1, suggesting that it is distinct from the ligand-mimetic mAb.
Epitope Mapping of J19 IgG-Because of the lack of crossreactivity with ␣ 4 ␤ 1 by J19 IgG and the high homology between ␤ 1 and ␤ 7 subunits, we constructed a panel of ␤ 1 /␤ 7 chimeras to locate the epitope of J19 IgG in ␤ 7 subunit. A schematic of the constructed chimeras is shown in Fig. 7A. These chimeras were all transiently co-expressed with human ␣ 4 subunit in 293T cells, the expression level of which was confirmed by immunostaining with 9F10 against ␣ 4 . Flow cytometric analysis of these 293T transfectants showed that the ␤ 7 segment 331-348 located in ␤I domain was absolutely required for the binding of . J19 IgG recognizes ␣ 4 ␤ 7 heterodimer and cross-reacted with mouse and rat ␣ 4 ␤ 7 . A, 293T cells were transfected with human ␣ 4 ␤ 1 or ␣ E ␤ 7 , and the integrin expression level was determined by indicated antibodies, respectively. The binding of J19 IgG to ␣ 4 ␤ 1 or ␣ E ␤ 7 was analyzed by flow cytometry in the presence of 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ or 2 mM Mn 2ϩ . B, human, mouse, or rat ␣ 4 ␤ 7 was transiently expressed in 293T cells, and the integrin expression level was determined by 5 g/ml indicated antibodies, respectively. Reactivity of J19 IgG with ␣ 4 ␤ 7 293T transfectants was determined by flow cytometry. A representative experiment (of three) is shown as a histogram. The numbers within the panels show the specific mean fluorescence intensity of indicated mAbs. The results are the means Ϯ S.D. of three independent experiments. J19 IgG to ␣ 4 ␤ 7 . All of the chimeras in which segment 331-348 was of ␤ 1 origin failed to be stained with 5 g/ml J19 IgG, whereas all other chimeras in which this segment was of ␤ 7 origin were stained, as well as WT ␤ 7 (Fig. 7A). Within region 331-348, seven amino acids differ between human ␤ 1 and ␤ 7 . Thereafter, we substituted the seven ␤ 7 residues with the cor-responding ␤ 1 residues. Mutations of S331E, A332E, and A333F decreased ϳ30 -50% of J19 IgG binding in comparison with WT ␣ 4 ␤ 7 . By contrast, mutations of the other four residues, L334Q, Q338K, S341K/K342N, almost had no effect on the recognition of ␣ 4 ␤ 7 by J19 IgG. Moreover, the ␤ 7 triple mutation S331E/A332E/A333F completely abolished the recognition of ␣ 4 ␤ 7 by J19 IgG (Fig. 7B). These results strongly suggest that residues Ser-331, Ala-332, and Ala-333 in ␤ 7 I domain represent a direct binding site for J19 IgG.
The J19 IgG binding site in human ␣ 4 subunit was mapped by using ␣ 4 /␣ E chimeras because ␣ 4 and ␣ E share the same ␤ 7 subunit. Considering ␤-propeller domain in ␣ subunit is close to the above mapped J19 epitope in ␤I domain, we first swapped ␤-propeller domain of ␣ 4 and ␣ E subunits, whereas the swap of ␤-propeller domain of ␣ 4 and ␣ E subunits resulted in no expression of both ␣ 4 ␤ 7 and ␣ E ␤ 7 chimeric integrins. The abnormal expression of ␣ 4 /␣ E chimeras is possibly due to the difference in structure between ␣I domain-less integrin ␣ 4 and ␣I domaincontaining integrin ␣ E . Thus, based on J19 binding sites in ␤ 7 subunit and the crystal structures of integrin ␣ IIb ␤ 3 and ␣ V ␤ 3 , several segments in ␣ 4 ␤-propeller domain close to the epitope in ␤ 7 subunit were substituted with corresponding ␣ E sequences, respectively. These chimeric cDNAs were all cloned into pcDNA 3.1 expression vectors and transiently co-expressed with human ␤ 7 subunit in 293T cells, then followed by immunostaining with 5 g/ml J19 IgG. The expression of these chimeras was confirmed by immunostaining with mAb FIB27 against ␤ 7 . Swapping of ␣ 4 segments 211-216 and 240 -246 with those of ␣ E had no effect on the binding of J19 IgG, whereas  swapping of the ␣ 4 184 -190 segment completely abolished J19 IgG binding to chimera ␣ 4 184␣ E 190␣ 4 (Fig. 7C). These results demonstrate that the epitope of J19 IgG in ␣ 4 subunit locates in a seven-residue segment from 184 to 190 in ␤-propeller domain. However, J19 did not bind to chimeric ␣ E (␣ E 184␣ 4 190␣ E ) when swapping the 184 -190 segment of ␣ 4 into ␣ E subunit (Fig. 7C), which might be due to the interference of J19 binding by the ␣I domain on the top of the ␤-propeller domain in ␣ E subunit.
The Epitope Exposure of J19 IgG Is Coupled with Extension of Integrin ␣ 4 ␤ 7 Ectodomain-The headpiece of integrin folds over its legs and faces down toward the membrane in the low affinity bend conformation and extends upward in a switchblade-like opening upon activation (2,43). We next studied the relationship between J19 IgG epitope exposure and the conformational rearrangement of integrin ␣ 4 ␤ 7 during its activation.
To assess the orientation of integrin ␣ 4 ␤ 7 ectodomain relative to the cell membrane using a FRET system, ␣ 4 ␤ 7 was labeled with Alexa Fluor 488-Act-1 Fab fragment as donor, which binds to the top of ␣ 4 ␤ 7 ␤I domain (44). The outer face of plasma membrane was labeled with FM4 -64 FX (FM) as acceptor (30,35). Compared with the inactive ␣ 4 ␤ 7 in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ , the FRET efficiency of 293T transfectants bearing activated ␣ 4 ␤ 7 in 2 mM Mn 2ϩ was significantly decreased from 22 to 6%, indicating the extension of the ␣ 4 ␤ 7 ectodomain (Fig. 8A). In parallel, immunostaining results showed that J19 IgG bound to activated ␣ 4 ␤ 7 in 2 mM Mn 2ϩ but not to inactive ␣ 4 ␤ 7 in 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ , suggesting that the epitope of J19 IgG was only exposed in the activated ␣ 4 ␤ 7 with extended conformation in 2 mM Mn 2ϩ (Fig.  8B). These data strongly indicate that the J19 IgG epitope is exposed when the integrin head domain moves away from the cell membrane in the extended conformation. . Epitope mapping of J19 IgG. A, mapping of J19 IgG epitope with ␤ 7 chimeras. mAb reactivity was determined with chimeric human ␤ 1 /␤ 7 subunits co-expressed with human ␣ 4 in 293T cells in the presence of 2 mM Mn 2ϩ by using flow cytometry. J19 IgG recognition was measured as specific mean fluorescence intensity and quantitated as a percentage of total ␣ 4 ␤ 7 expression defined by staining with mAb 9F10 to ␣ 4 . B, fine mapping of J19 IgG epitope in ␤ 7 subunit. Human WT ␤ 7 or mutant ␤ 7 containing multiple or single ␤ 7 to ␤ 1 amino acid substitutions was co-expressed with human ␣ 4 subunit in 293T cells. J19 IgG recognition was quantified as in A. C, mapping of J19 IgG epitope with ␣ 4 chimeras. The human ␣ 4 /␣ E chimeras were co-expressed with human ␤ 7 in 293T cells. The transfectants were stained with J19 IgG or mAb FIB27 to ␤ 7 , followed by flow cytometry in the presence of 2 mM Mn 2ϩ . J19 IgG recognition was quantified as in Fig. 7A. The error bars are Ϯ S.D. (n ϭ 3).

DISCUSSION
In this study, we screened and characterized an activationspecific human monoclonal antibody to integrin ␣ 4 ␤ 7 by panning a scFv-displaying phagemid library using an active form of ␣ 4 ␤ 7 (wedge mutant) as target. The subtractive selection was performed by depleting the library on the isolated, inactive WT ␣ 4 ␤ 7 protein first and then panning against the constitutively activated ␣ 4 ␤ 7 wedge mutant. The specific scFv clone J19 was isolated from the library and reformatted to full-length human IgG1. The J19 IgG specifically binds to integrin ␣ 4 ␤ 7 activated by different stimuli, other than the inactive ␣ 4 ␤ 7 . The binding epitope of J19 IgG was mapped to two small segments located in ␣ 4 ␤-propeller domain and ␤ 7 I domain, respectively (supplemental Fig. S1). The seven-residue segment from 184 to 190 in ␣ 4 subunit locates between ␤-strands 2 and 3 in ␤-sheet 3 of the ␤-propeller domain. The other segment consists of residues Ser-331, Ala-332, and Ala-333 located in a turn between ␤-strand 5 and ␣-helix 5 at the top of ␤I domain. In the low affinity bend conformation, the integrin ␤-propeller domain and the ␤I domain sit on their leg domains and face the cell membrane, leading to the epitope buried in the inactive integrin. Upon activation, integrin converted from the bent to extended conformation with either closed headpiece (intermediate affinity state) or open headpiece (high affinity state), which led to the exposure of J19 epitope (supplemental Fig. S1). Thus, J19 IgG is an activation-dependent mAb, which recognizes an epitope expressed only in integrin ␣ 4 ␤ 7 with extended conformation.
The high affinity wedge mutant ␣ 4 ␤ 7 used for J19 selection contains a mutation-introduced N-glycan at the integrin ␤ 7 I/hybrid domain interface and mimics the swing-out of hybrid domain in ␤ 7 subunit, which is predicted to activate integrins and stabilize the high affinity conformation. In our study, we showed that J19 IgG bound both wedge mutant and WT integrin ␣ 4 ␤ 7 activated by physiological agonist, such as CXCL12.
Thus, the epitope recognized by J19 IgG is expressed in both high affinity wedge mutant and the physiologically activated WT integrin ␣ 4 ␤ 7 . These results strongly suggest that the hybrid domain swing-out is the key conformational rearrangement during integrin activation and can induce a high affinity conformation mimicking the physiological activation of integrin ␣ 4 ␤ 7 .
Different from the J19 phage and J19 IgG, which specifically recognize the activated integrin ␣ 4 ␤ 7 , J19 scFv showed similar binding to both inactive and activated ␣ 4 ␤ 7 . The loss of specificity to activated ␣ 4 ␤ 7 could be due to the smaller size of scFv compared with phage particle and full-length IgG1. The epitope of J19 in the inactive ␣ 4 ␤ 7 with bend conformation is not accessible to large size J19 phage or IgG, whereas the space is large enough for smaller scFv to access the epitope. This result also provides supporting evidence for the induction of J19 epitope expression by the conformational rearrangements of integrin.
In ␣I domain-containing integrins, the ␣I domain is inserted into a loop between ␤-sheets 2 and 3 at the top of the ␤-propeller domain (45,46). As part of the J19 epitope, the seven-residue segment from 184 to 190 locates in ␤-sheet 3 of the ␤-propeller domain, which will be masked by the ␣I domain in the ␣ E subunit (supplemental Fig. S2). Thus, J19 IgG cannot bind to integrin ␣ E ␤ 7 even after replacement of the seven-residue segment in ␣ E subunit with that from ␣ 4 (Fig. 7C).
Interestingly, the binding of J19 IgG to ␣ 4 ␤ 7 could be enhanced by a number of stimuli, including Mn 2ϩ , DTT, and ADP, but to different levels. The binding of J19 IgG showed stronger binding to ␣ 4 ␤ 7 expressing cells treated with Mn 2ϩ than the same cells stimulated with DTT or ADP, suggesting that integrin activated by different stimuli could have different conformations and expression of J19 epitope. The different expression levels of J19 epitope were consistent with the detection of increased affinity for ligand after those stimulations, as  10). B, the expression of J19 IgG epitope was measured as specific mean fluorescence intensity and quantitated as a percentage of total ␣ 4 ␤ 7 expression defined by staining with mAb Act-1. 293T cells stably expressing ␣ 4 ␤ 7 were stained with 5 g/ml J19 IgG and Act-1 in the presence of 1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ or 2 mM Mn 2ϩ as indicated and followed by flow cytometry. The error bars are Ϯ S.D. (n ϭ 3). measured by a sensitive flow chamber assay (34,35,47). Thus, J19 IgG could serve as a reporter for the activation extent of ␣ 4 ␤ 7 .
In addition to serving as a research tool for in vitro and in vivo studies of integrin activation, the J19 antibody also represents a therapeutic candidate for treatment of inflammatory bowel disease. J19 IgG selectively targets the activated lymphocytes and leaves the inactive ones intact, which may result in fewer potential side effects.