Dissection of α4β7 integrin regulation by Rap1 using novel conformation-specific monoclonal anti-β7 antibodies

Integrin activation is associated with conformational regulation. In this study, we developed a system to evaluate conformational changes in α4β7 integrin. We first inserted the PA tag into the plexin-semaphorin-integrin (PSI) domain of β7 chain, which reacted with an anti-PA tag antibody (NZ-1) in an Mn2+-dependent manner. The small GTPase Rap1 deficiency, as well as chemokine stimulation and the introduction of the active form of Rap1, Rap1V12, enhanced the binding of NZ-1 to the PA-tagged mutant integrin, and increased the binding affinity to mucosal addressing cell adhesion molecule-1 (MAdCAM-1). Furthermore, we generated two kinds of hybridomas producing monoclonal antibodies (mAbs) that recognized Mn2+-dependent epitopes of β7. Both epitopes were exposed to bind to mAbs on the cells by the introduction of Rap1V12. Although one epitope in the PSI domain of β7 was exposed on Rap1-deficienct cells, the other epitope in the hybrid domain of β7 was not. These data indicate that the conversion of Rap1-GDP to GTP exerts two distinct effects stepwise on the conformation of α4β7. The induction of colitis by Rap1-deficient CD4+ effector/memory T cells suggests that the removal of constraining effect by Rap1-GDP on α4β7 is sufficient for homing of these pathogenic T cells into colon lamina propria (LP).


Results
Development of a system to measure the active conformation of α 4 β 7 by inserting a PA tag. α 4 β 7 showed low-affinity state to MAdCAM-1 in Ca 2+ /Mg 2+ , and the addition of Mn 2+ increased the binding affinity of α 4 β 7 to MAdCAM-1 (Fig. 1a). To probe the conformational state of α 4 β 7 using the PA-tag-NZ-1 system (Fig. 1a), a proB cell line (BAF cells), in which the β 7 chain was knocked-out using CRISPR/ Cas9-mediated genome editing, and β 7 chains cDNA which were inserted PA tag into 4 locations (PAins 1: 23/24, PAins 2: 29/30, PAins 3: 427/428, PAins 4: 431/432) were introduced into BAF cells (Fig. 1b,c). These insertion mutants of β 7 -expressing cells were stained with NZ-1 or FIB504 (conventional mAb against mouse/ human β 7 , which recognizes the binding sites of α 4 β 7 with MAdCAM-1) in the presence of 1 mM Ca 2+ /Mg 2+ or 0.5 mM Mn 2+ , and analyzed by flow cytometry. All mutants of β 7 were approximately equally expressed on the cell surface (Fig. 1c). PA expression on the surface of PAins2-expressing cells (PAins2 cells) was reduced in the low-affinity α 4 β 7 with the bent conformation in the presence of Ca 2+ /Mg 2+ and exhibited an eightfold increase in the high-affinity α 4 β 7 with the extended conformation in the presence of Mn 2+ (Fig. 1d). In the cells expressing other insertion mutants of β 7 , PA was exposed to be recognized by NZ-1 in the bent conformation, the same as in the extended conformation of β 7 (Fig. 1d). These data showed that PAins2 in the PSI domain of β 7 was an appropriate PA tag insertion design.
As exogenous addition of antibodies that preferentially bind to the extended conformation can shift the equilibrium toward the high-affinity state of integrins, they often activate integrins from outside the cell. Therefore, we confirmed the conformational change in α 4 β 7 in the presence of Mn 2+ using the Fv-clasp of NZ-1. The Fv-clasp of NZ-1, an artificially designed small antibody fragment of 37 kDa, was used as a reporter of conformational change, as it did not affect the equilibrium between the high-and low-affinity states 19 . Using Fv-clasp of NZ-1, the surface expression of PA tag exhibited a 11-fold increase in the presence of Mn 2+ , compared to that in the presence of Ca 2+ /Mg 2+ (Fig. 2), indicating that the surface expression of PA precisely reflects the active conformation of α 4 β 7 .

Rap1 deficiency induced a conformational change in α 4 β 7 .
A conformational equilibrium between bent (low-affinity) and extended (high-affinity) states of integrin was found to be regulated by inside-out signaling such as Rap1. Using PAins2 cells, we examined the effects of Rap1 on the conformational state of α 4 β 7 . To this end, we introduced the GTP-binding form of Rap1, Rap1V12, Rap1 GTPase activating protein (GAP), Spa-1,  www.nature.com/scientificreports/ and knocked down of Rap1a/b in PAins2 cells (Fig. 3a). We confirmed that CXCL12-induced Rap1 activation was inhibited in Spa-1-expressing PAins2 cells (Fig. 3b). Next, we examined the binding activity of α 4 β 7 on each transfectant to soluble MAdCAM-1 in the presence or absence of CXCL12. As shown in Fig. 3c, CXCL12 stimulation elevated the binding of α 4 β 7 on control cells to soluble MAdCAM-1, indicating that chemokine stimulation shifted the equilibrium of α 4 β 7 toward high-affinity state. As expected, overexpression of Rap1V12 increased the binding of α 4 β 7 to soluble MAdCAM-1, without CXCL12 stimulation (Fig. 3c). The inhibition of Rap1 activation by overexpression of Spa-1 suppressed CXCL12dependent increase in the binding of α 4 β 7 to MAdCAM-1 (Fig. 3c). Knockdown of Rap1 also significantly increased the binding activity, but the effect was weak compared to that of Rap1V12-expressing cells (Fig. 3c). These data suggest that Rap1-GDP locks α 4 β 7 in the low-affinity state and that Rap1-GTP further promotes an equilibrium toward high-affinity state of α 4 β 7 .
Subsequently, we examined changes in the surface expression of the PA tag in each transfectant. CXCL12 stimulation exhibited a 1.3-fold increase in PA surface expression, and overexpression of Spa-1 completely inhibited PA surface expression induced by CXCL12 stimulation (Fig. 3d), indicating that the conversion of Rap1-GDP to GTP is necessary for the surface expression of PA. Overexpression of Rap1V12 significantly promoted PA surface expression (Fig. 3d). PA surface expression was higher in Rap1-deficient cells as compared with that in Rap1V12-expressing cells (Fig. 3d), suggesting that the loss of Rap1-GDP induced a conformational change in α 4 β 7 and that this change is different from the Rap1V12-induced conformation of α 4 β 7 . These results indicate that Rap1-GDP suppresses conformational changes to active form of α 4 β 7 .
In addition, talin is reported to bind Rap1-GTP and integrin, and trigger integrin activation 20 . Therefore, we examined the effect of the knockdown (KD) of talin. The abundance of talin protein in talin KD cells was reduced to 5% of control cells (Fig. S1a). As shown in Fig. S1b, the silencing of talin reduced basal surface expression of PA, whereas CXCL12 increased surface expression of PA in talin-KD cells at a same proportion as control cells. This result suggests that talin is a basic cytoskeletal component necessary for active conformation of α 4 β 7 , rather than a downstream effector molecule of chemokine-mediated signaling.
Identification and characterization of rat mAbs to detect Rap1-dependent conformational changes in α 4 β 7 . To establish hybridomas producing mAbs that exclusively reacted with α 4 β 7 in an Mn 2+ -dependent manner, immunogenic N-terminal amino acids (1-458) of β 7 -MBP fusion protein were injected into rats. A hybridoma producing rat mAb G3 (γ2/κ) for Mn 2+ -dependent conformation of α 4 β 7 was established. As shown in Fig. 4a, the G3 epitope was almost not detected in the low-affinity α 4 β 7 with a bent conformation in the presence of Ca 2+ /Mg 2+ but increased 4.8-fold in the high-affinity α 4 β 7 with the extended conformation in the presence of Mn 2+ .
Next, we tested the cross-reactivity of the G3 mAb with human α 4 β 7 using Jurkat cells transfected with human β 7 . The surface expression level of β 7 was determined in the Jurkat cells using FIB504 (Fig. 4a). G3 epitope expression in the Jurkat cells was extremely low in the presence of Ca 2+ /Mg 2+ and increased fivefold in the presence of Mn 2+ (Fig. 4a). To identify the epitope of G3, we constructed murine β 1 /β 7 chimeras and the deletion mutant (∆1-19) of β 7 . These mutants were co-expressed with endogenous α 4 in β 7 -knockout BAF cells, and the surface expression level was confirmed by the immunostaining with FIB504 (Fig. 4b). As shown in Fig. 4b, the deletion of N-terminal 1-19 a.a. of β 7 , which does not exist in β 1 , let G3 mAb lose the reactivity to murine β 7 . Thus, flow cytometric analysis of these transfectants showed that the β 7 segment 1-19 a.a. located in the PSI domain was required for binding of G3 mAb to β 7 (Fig. 4b), which was close to the PA grafting site (Fig. 1c). As expected, G3  www.nature.com/scientificreports/ epitope expression increased 1.2-fold with CXCL12 stimulation (Fig. 4c). Overexpression of Rap1V12 enhanced the expression of G3 epitope to 1.9-fold ( Fig. 4c). Rap1-deficiency also increased the expression of G3 epitope to 2.8-fold (Fig. 4c). Consistent with the results using the PA-tag-NZ-1 system, these data indicate that Rap1-GDP locks the conformation of α 4 β 7 in inactive state. We also established a hybridoma producing rat mAb H3 (γ2/κ) to detect Mn 2+ -dependent conformation of α 4 β 7 . As shown in Fig. 5a, the expression of H3 epitope was almost not detected in the low-affinity of α 4 β 7 with the bent conformation in the presence of Ca 2+ /Mg 2+ and increased 33-fold in the presence of Mn 2+ , suggesting that H3 mAb recognized the high-affinity α 4 β 7 . Subsequently, we explored the cross-reactivity of H3 mAb with human α 4 β 7 using Jurkat cells transfected with human β 7 . H3 epitope on Jurkat cells was not expressed in the presence of Mn 2+ (Fig. 5a), indicating that H3 mAb did not recognize the active conformation of human β 7 . To identify the epitope of H3, we constructed murine/human β 7 chimeras. These chimeras were co-expressed with endogenous α 4 in β 7 -knockout BAF cells, and the surface expression level was confirmed by immunostaining with FIB504 (Fig. 5b). Flow cytometric analysis of these transfectants showed that the β 7 segment 373-393 a.a. located in the hybrid domain was required for binding of H3 mAb to β 7 (Fig. 5b). As expected, the expression of H3 epitope also increased 1.4-fold with CXCL12 stimulation (Fig. 5c). Overexpression of Rap1V12 enhanced the expression of H3 epitope to 3.7-fold (Fig. 5c). However, Rap1 deficiency did not increase the expression of H3 epitope with or without CXCL12 stimulation (Fig. 5c). These data indicate that the expression of H3 epitope requires Rap1-GTP.
In our previous paper 10 , we demonstrated that T cell-specific Rap1-deficient mice developed severe colitis with infiltration of CD4 + T EM cells into colon LP and that adoptive transfer of these cells into normal mice induced colitis. Previous study also demonstrated that α 4 β 7 -MAdCAM-1-dependent rolling was significantly promoted in Rap1-deficient CD4 + T EM cells 10 . In the present study, the injection of a MAdCAM-1 inhibitory mAb into T cell-specific Rap1-deficient mice prevented the development of colitis (Fig. 6a, Fig. S3). These findings confirmed that the α 4 β 7 -MAdCAM-1 interaction was critical for the development of colitis in T cell-specific Rap1a/bknockout mice. Therefore, we explored whether a conformational change in α 4 β 7 was observed in pathogenic T cells. Since CCR9 and its ligand CCL25 are found to play essential roles in gut-homing of T EM cells 21 , we used CCL25 for the stimulation of CD4 + T EM cells. As shown in Fig. 6, G3 epitope significantly increased in pathogenic T cells as compared to that in wild-type T EM cells, although the surface expression of α 4 β 7 was elevated in the pathogenic T cells. CCL25 increased the expression of G3 epitope in control cells but not in pathogenic T cells (Fig. 6b). As expected, the expression of the epitope recognized by H3 mAb was reduced and not induced by CCL25 stimulation in pathogenic T cells, although the addition of Mn 2+ induced H3 epitope in these cells at a similar level to that in wild-type T EM cells (Fig. 6c). These data suggest that active conformation in α 4 β 7 induced by Rap1 deficiency is sufficient for the infiltration of the pathogenic T cells into colon LP through rolling and arrest on MAdCAM-1-expressing endothelial cells.
In addition, previous study reports that CCL25 and CXCL10 induces different active conformation of α 4 β 7 22 , but there is no difference between the effects of CCL25 and CXCL10 in the expression of G3 and H3 epitopes on CD4 + T EM cells (Fig. S4).

Discussion
In this study, we developed a sensitive system to probe the conformational states of α 4 β 7 using the insertion of PA tag into the PSI domain of the β 7 chain. Using this system, we found that the conformation of α 4 β 7 was regulated by Rap1. To examine the conformational changes of β 7 in primary lymphocytes, we isolated two rat mAbs (G3 and H3) against activation-dependent α 4 β 7 . G3 mAb recognized the epitope in the PSI domain of β 7 , and H3 mAb recognized the epitope in the hybrid domain of β 7 (Fig. 7a). Both epitopes were hidden in the lowaffinity α 4 β 7 with the bent conformation and exposed in the high-affinity α 4 β 7 with the extended conformation induced by the addition of Mn 2+ . The introduction of Rap1V12 induced the exposure of G3 and H3 epitopes to be recognized by mAbs. However, the expression of G3 epitope was increased by depletion of Rap1-GDP, The IMF of binding to G3 mAb was normalized to the IMF of anti-β 7 (FIB504) and is presented as the fold-increase relative to that of wt BAF cells in the presence of 1 mM Ca 2+ /Mg 2+ values of 1. Data represent the mean ± SE of three independent experiments. *P < 0.001, versus Ca 2+ /Mg 2 . (Lower) (left) Flow cytometry profiles of G3 of Jurkat cells expressing human β 7 in the presence of 1 mM Ca 2+ /Mg 2+ , or 0.5 mM Mn 2+ . (right) The IMF of binding to G3 mAb was normalized to the IMF of anti-β 7 and is presented as the fold-increase relative to that of Jurkat cells in the presence of 1 mM Ca 2+ /Mg 2+ values of 1. Data represent the mean ± SE of three independent experiments. *P < 0.001, versus Ca 2+ /Mg 2+ . (b) (Left) (upper) Mapping of G3 epitope. G3 mAb reactivity was determined by co-expressing chimeric murine β 1 /β 7 subunits or ∆1-19 murine β 7 with endogenous α 4 in β 7 -knockout BAF cells in the presence of 0.5 mM Mn 2+ using flow cytometry. (lower) The amino acid sequence of 1-19 a.a. of murine, human β 7 and murine β 1 . The red characters indicate the candidates recognized by G3 mAb. (right) Flow cytometry profiles of the transfectants with anti-β 7 or G3 in the presence of 0.5 mM Mn 2+ . (c) The binding of G3 mAb to α 4 β 7 on the cells introduced with control, Rap1V12, Spa-1-expressing, or Rap1a/b-knockdown cells in the presence or absence of CXCL12. The IMF of binding to G3 mAb was normalized to the IMF of anti-β 7 and is presented as the fold-increase relative to that of unstimulated control cell values of 1. Data represent the mean ± SE of three independent experiments. * 1 P < 0.001, versus unstimulated cells, * 2 P < 0.03, versus unstimulated control cells, * 3 P < 0.002 versus CXCL12-stimulated control cells, * 4 P < 0.001, versus unstimulated control cells. www.nature.com/scientificreports/ whereas the conversion to Rap1-GTP was indispensable for the exposure of H3 epitope. Thus, Rap1-GDP and GTP independently regulated the conformation of α 4 β 7 (Fig. 7). The binding of NZ-1 or G3 mAb to the PSI domain of β 7 was suppressed by Rap1-GDP, and loss of Rap1-GDP was sufficient for maximal expression of these epitopes (Fig. 7). On the other hand, H3 epitope in the hybrid domain of β 7 was not exposed by only deletion of Rap-GDP. The loss of Rap1-GDP had marginal effect on the binding of α 4 β 7 to soluble MAdCAM-1. The overexpression of Rap1V12 as well as the addition of Mn 2+ , which increased the binding of α 4 β 7 to soluble MAdCAM-1, induced the exposure of H3 epitope (Figs. 3c, 5c). Thus, the surface expression of H3 epitope was strongly correlated with the binding activity of α 4 β 7 to soluble MAdCAM-1. These findings indicate that H3 mAb might detect the swing-out of the hybrid domain in the β 7 chain, which is predicted to stabilize the high-affinity conformation 23 (Fig. 7). These data suggest that Rap1-GDP restrained the bent conformation in α 4 β 7 and maintain the binding of α 4 β 7 to MAdCAM-1 in low-affinity and that the conversion into Rap1-GTP further facilitated the active conformation of α 4 β 7 , resulting in binding of α 4 β 7 to MAdCAM-1 in high-affinity (Fig. 7). As previously reported 24,25 , conformation-specific antibodies are useful for the elucidation of the functions and regulatory mechanisms of integrin conformation. The combination of G3 and H3 mAbs, which might differentiate extended closed from extended open conformation of α 4 β 7 , will contribute to various studies of conformational regulation of α 4 β 7 .
The insertion site of PA-tag was between the 29th and 30th in N-terminal of amino acid sequence and G3 epitope was in N-terminal first 19 amino acids. Previous study 26,27 report that AP5 is a mAb that recognizes β3 integrin only in the extended conformation. These findings suggest that N-terminal epitope could be used in many or all beta integrins to obtain antibodies that recognize an active conformation. Since there are many kinds of antibodies including N-terminal amino acids of β chain 28 , it is necessary to clarify whether these antibodies specifically recognize the active conformation of other integrin.
In a previous study, we demonstrated that Rap1-GDP activated LOK and promoted ERM phosphorylation and that the introduction of the active form of LOK or phospho-mimetic ezrin did not prevent conformational changes in α 4 β 7 . Although RIAM (Rap1-interacting molecule) and talin are known to induce active conformation of integrins, they are associated with Rap1-GTP but not Rap1-GDP 8,9,29 . In this study, we suggest that talin is a basic cytoskeletal component involved in active conformation of α 4 β 7 . Studies are needed to shed light on what are the downstream effector molecules of Rap1-GDP/GTP and the roles of cytoskeletal proteins such as RIAM and talin in regulation of conformation of α 4 β 7 .
Previous paper 22 demonstrates that CCL25 and CXCL10 induce different conformation of α 4 β 7 , which favors a MAdCAM-1-and VCAM-1-binding, respectively. According to that paper, CCL25 and CXCL10 activates the different signaling pathways which lead to different phosphorylation states of β 7 and distinct talin and kindling-3 binding patterns 22 . On the other hand, there was no difference in surface expression of G3 and H3 epitopes between CCL25 and CXCL10-stimulated T cells (Fig. S4). Therefore, G3 and H3 mAbs did not seem to discriminate the MAdCAM-1 or VCAM-1-binding conformation of α 4 β 7 . In this study, the conformation of α 4 β 7 recognized by G3 mAb was suggested to be critical for gut-homing of T EM cells, whereas the physiological significance of Rap1-GTP-dependent active conformation recognized by H3 mAb remains to be solved . It is important to clarify the biological implication of conformational regulation.
Rap1 is indispensable for chemokine-dependent integrin activation and naive lymphocyte recirculation, and its deficiency leads to lymphopenia in secondary lymph nodes 10,30 . On the other hand, Rap1-deficient T EM cells exhibit enhanced α 4 β 7 /MAdCAM-1-dependent rolling and arrest on the endothelium, as well as accelerated homing into colon LP 10 . In this study, we found that Rap1 deficiency in T EM cells led to a conformational change in α 4 β 7 , which promoted α 4 β 7 /MAdCAM-1-dependent endothelial transmigration and homing to colon LP 10 . Furthermore, the Rap1-GTP-dependent high-affinity conformation of α 4 β 7 , which was recognized by H3 mAb, was unnecessary for homing of pathogenic T cells into colon LP.
G3 mAb recognized the murine and human active form of α 4 β 7 , and the expression of G3 epitope correlated with the infiltration of pathogenic T EM cells into colon LP. Thus, this mAb might be a useful tool to deliver drugs to pathogenic T EM cells. In addition, as the inhibition of the conformational change in α 4 β 7 seemed to be effective in preventing the infiltration of T EM cells into colon LP, the system we developed can be used to screen for drugs The binding of H3 mAb to α 4 β 7 on the cells introduced with control, Rap1V12, Spa-1-expressing, or Rap1a/b-knockdown cells in the presence or absence of CXCL12. The IMF of binding to H3 mAb was normalized to the IMF of anti-β 7 and is presented as the fold-increase relative to that of unstimulated control cell values of 1. Data represent the mean ± SE of three independent experiments. * 1 P < 0.02, versus unstimulated cells, * 2 P < 0.005, versus unstimulated control cells, * 3 P < 0.05 versus CXCL12-stimulated control cells, * 4 P < 0.05 versus CXCL12-stimulated control cells.

Cell lines. Ba/F3 cells (BAF cells) and
Jurkat cells were cultured as previously reported 31 . BAF cells were cultured with RPMI1640 medium containing 10% fetal calf serum, 50 mM beta-mercaptoethanol, and 1% WEHI-3 conditional medium as a source of interleukin 3. Jurkat cells were maintained with RPMI1640 medium containing 10% fetal calf serum. All cell lines were tested for mycoplasma contamination by 4′,6-diamidino-2-phenylindole (DAPI) staining with negative results.

Antibodies and reagents.
Fluorescence-conjugated anti-mouse CD4, CD44, anti-β 7 (FIB504) (BioLegend), anti-Rap1(BD Biosciences), β-actin (Sigma), T7 (MBL) , Flag (Wako), anti-talin (Abcam), APC-conjugated anti-Rat or human IgG, and peroxidase-conjugated goat anti-Rabbit or -mouse IgG (Cell Signaling) were used for flow cytometry and immunoblotting. Anti-MAdCAM-1(MECA-367) (ATCC), G3 and H3 mAb were purified using HiTrap Protein G HP (GE healthcare). Mouse CXCL12, CCL25 and CXCL10 were purchased from R&D Systems. The single-chain Fv of NZ-1 (Fv-clasp) was created by fusing an anti-parallel coiled-coil structure derived from the SARAH domain of human Mst1 kinase to the fragments of V H and V L of NZ-1. NZ-1V H -SARAH and V L -SARAH were separately expressed in E. coli strain BL21, and cultured in standard LB media 19 . After the cell lysis by sonication, the denatured and solubilized V H -SARAH and V L -SARAH chains were then mixed, and the denaturing reagent was diluted to promote protein folding, and correctly-folded, disulfide-bonded Fv-clasp was purified as previously reported 19 . Fv-clasp was fluorescently labeled with Alexa Fluor 647 Amine-Reactive Dye (Thermo Fisher Scientific).
DNA constructs and transfection. cDNA encoding murine β 7 cDNA was subcloned into a pcDNA3.1 vector. Then, β 7 mutants with a PA tag were generated from a pcDNA3.1-murine β 7 construct using inverse PCR.  Fig. 1c. To generate expression constructs of the PAins mutants, they were subcloned into an EcoRI/XbaI site of a lentivirus vector (CSII-EF-MCS; a gift from H. Miyoshi, RIKEN, Wako, Japan). A Rap1V12 mutant and Spa-1 were generated as previously described 32 . The fidelity of all the constructs was verified by sequencing.  WKY rats (8 week old) were injected intramuscularly at the tail base with an antigen emulsion containing MBP-β 7 and Freund's complete adjuvant (BD Biosciences). Then, 2 weeks later, lymphocytes were collected from iliac lymph nodes and fused with a murine myeloma cell line, SP2/0, using GenomeONE (ISHIHARA SANGYO) 33 . Hybridoma clones producing mAbs against the active form of β 7 were screened by flow cytometry of BAF cells using the hybridoma supernatant in the presence or absence of Mn 2+ .

RNA-mediated interference and gene introduction via lentiviral transduction. RNA-mediated
interference was used to suppress mouse expression. As previously reported 34 , a 19-nucleotide -specific sense RNA sequences or a scrambled control RNA sequence of (Rap1a: 5′-GAA TGG CCA AGG GTT TGC A-3′, Rap1b: 5′-AGA CAC TGA TGA TGT TCC A-3′, and talin: 5′-CGG TGA AGA CTA TCA TGG T-3′) were introduced into BAF cells using a lentivirus vector with GFP (a gift from Dr. Miyoshi H., RIKEN, Wako, Japan) containing the RNAi construct under control of the H1 promoter cassette, respectively. The production and concentration of lentivirus particles were assessed as previously described 35 . The transduction efficiencies were greater than 90%. A GFP high population was collected by cell sorting and subjected to adhesion assays and immunoblot analysis. Pull-down assays. Rap1-GTP was pulled down with a glutathione S-transferase (GST)-RBD of RalGDS fusion protein, respectively 36 . Briefly, 10 7 cells were lysed in ice-cold lysis buffer (1% Triton X-100, 50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 10 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 1 mM leupeptin, and 0.5 mM aprotinin) and incubated for 1 h at 4 °C with GST-fusion proteins coupled to glutathione agarose beads. The beads were washed three times with lysis buffer and subjected to immunoblot analysis using an anti-Rap1 antibody. Immunoblotting of total cell lysates (5 × 10 4 cells) was also performed.

Assessment of activation epitopes by mAbs staining.
Immunofluorescence flow cytometry was performed as described previously 31 . For NZ-1, G3 or H3 mAbs staining, cells were washed with binding buffer (0.1% BSA, 1 mM CaCl 2 , 1 mM MgCl 2 or 0.1% BSA, 0.5 mM MnCl 2 in HBSS), resuspended in 50 μl of the same buffer, and incubated for 30 min at 37 °C with 10 μg/ml of each mAb in the presence or absence of 0.5 μM CXCL12, CCL25 or CXCL10. Mean fluorescence intensities were measured using a Gallios flow cytometry or CytoFLEX (Beckman Coulter).
Generation of β 7 -deficient BAF cells by the CRISPR/Cas9 system. The guide sequence targeting exon of the mouse β 7 was cloned into pX330 (Addgene #42230) 37 . pX330-U6-Chimeric_BB-CBh-hSpCas9 was a gift from Feng Zhang (Addgene plasmid #42230, https ://n2t.net/addge ne:42230 ; RRID: Addgene_42230). pCAG-EGxxFP was used to examine efficiency of the target DNA cleavage by the guide sequence and Cas9 activity. The resultant guide sequence was cloned into GFP expressing plasmid DNA pX458 (Addgene #48138) 38 . pSpCas9(BB)-2A-GFP (PX458) was a gift from Feng Zhang (Addgene plasmid #48138, https ://n2t.net/addge ne:48138 ; RRID: Addgene_48138). The pX458 plasmid was transfected into BAF cells. 24 h after transfection, cells were sorted GFP-high population, followed by limiting dilution. Expression of full length β 7 protein in each isolated clone was tested by flowcytometry. The sequence of the primer used were as follows: β 7 Exon, For- Figure 7. Model for the regulation of the conformation of α 4 β 7 by Rap1. (a) The structure of the α 4 β 7 headpiece. Crystal structure of α 4 β 7 headpiece was derived from PDB database with PDB code of 3V4P 4 , and created using the PyMOL molecular visualization system. The PSI domain of the β 1 headpiece structure is used instead of the PSI domain of β 7 , and the N-terminal 19 a.a. structure is unknown (surrounded by a dot-line). G3 mAb recognized the epitope in the PSI domain of β 7 (red), and H3 mAb recognized the epitope in the hybrid domain of β 7 (blue). (b) The model for conformational changes of α 4 β 7 . Both G3 (red) and the PA-tag located in PSI domain (green), and H3 (blue) epitopes were all hidden in the low affinity α 4 β 7 with the bent conformation. The surface expression of G3 epitope and PA increased by Rap1 deficiency, which might have induced an "extended-closed form" of α 4 β 7 . Rap1-GDP restrained conformational changes in α 4 β 7 and maintained the binding of α 4 β 7 to MAdCAM-1 under conditions of low-affinity binding (the bent form of α 4 β 7 ). The exposure of H3 epitope was dependent on Rap1-GTP, which was strongly correlated with the binding activity of α 4 β 7 to soluble MAdCAM-1. Rap1-GTP may have induced swing-out of the hybrid domain, which corresponded to the "extended open form" of α 4 β 7 , resulting in the binding of α 4 β 7 to MAdCAM-1 with high affinity. (c) Table summarizing the Rap1-binding guanine nucleotide (GTP or GDP) and each antibody (NZ-1, G3, H3)-binding activity (+ or −) in control, Rap1V12 or Spa-1-expressing cells, and Rap1KD cells in the presence or absence (No Stim) of CXCL12. The binding of the antibody to unstimulated control cells was displayed with (−), and the case that significantly increased, was displayed with (+). Epitope mapping of G3 and H3. Human/murine β 7 , murine β 1 /β 7 chimeras or ∆1-19 murine β 7 were constructed using an In-Fusion HD cloning kit (TaKaRa). The In-Fusion HD enzyme premix fuses multiple PCR-generated sequences and linearized vectors efficiently and precisely, utilizing a 20-bp overlap at their ends. This 20-bp-overlap allows complementary base pairs between two pieces of DNA to anneal, leading to fragment joining. Therefore, when individual DNA fragments derived from human β 7 , murine β 1 or β 7 were amplified by PCR, a 20-bp overlap was engineered by designing custom primers (Table S1). The objective DNA fragments with a 20-bp overlap were joined into a linearized CSII-EF-MCS-IRES2-venus vector. The constructs were transduced to β 7 -knockout BAF cells by lentivirus. The binding of G3 or H3 to the BAF cells expressing the chimera β 7 was measured in the presence of 0.5 mM Mn 2+ by flow cytometry.
Anti-MAdCAM-1 antibody treatment of colitis. T cell-specific Rap1a/b knockout mice aged 4 or 5 weeks were injected intraperitoneally with PBS containing 1 mg of rat IgG or anti-MAdCAM-1 antibody 39 on days 0, 7, and 21. Their body weights were measured every 2 days. Pathological or frozen sections were prepared on day 28.
Histological examination. Colon sections were fixed in 10% buffered formalin and embedded in paraffin. Paraffin-embedded colon sections were cut (3 μm), stained with haematoxylin and eosin and examined on an Olympus IX51 light microscope equipped with a CCD (charge-coupled device) camera. Tissues were graded semiquantitatively as described before 10,40 . Histological grades were assigned in a blinded manner on a scale of 0-5, as follows: grade 0, no changes observed,grade 1, minimal scattered mucosal inflammatory cell infiltrates, with or without minimal epithelial hyperplasia,grade 2, mild scattered to diffuse inflammatory cell infiltrates, sometimes extending into the submucosa and associated with erosions, with mild to moderate epithelial hyperplasia and mild to moderate mucin depletion from goblet cells; grade 3, moderate inflammatory cell infiltrates that were sometimes transmural, with moderate to severe epithelial hyperplasia and mucin depletion; grade 4, marked inflammatory cell infiltrates that were often transmural and associated with crypt abscesses or occasional ulceration, with marked epithelial hyperplasia, mucin depletion; and grade 5, marked transmural inflammation with severe ulceration or loss of intestinal glands.
Immunostaining. Preparation of frozen sections of the colon from colitis mice were performed as described previously. Sections were blocked for 1 h at 20 °C with PBS containing 10% goat serum and 0.1% Triton X-100 and incubated overnight at 4 °C with APC conjugated anti-CD4 antibody. Nuclei were stained with SlowFade Gold antifade reagent with DAPI (invitrogen). Sections were examined on TCS SP8 (Leica).
Statistical analysis. Statistical analysis was performed using two-tailed Student's t-test. P values less than 0.05 were considered significant.