Genetic perturbation of the putative cytoplasmic membrane-proximal salt bridge aberrantly activates α₄ integrins

Integrins, α/β heterodimeric glycoproteins, constitute the largest family of cell-surface adhesion molecules that mediate diverse cell-cell and cell-matrix interactions, supporting a wide range of physiologic processes including development, immune regulation, hemostasis, and wound healing.^1,2  A hallmark of the integrin family of adhesion receptors is their unique ability to dynamically up- and down-regulate their adhesiveness and, thereby, to coordinate the orchestrated cellular movements observed in leukocyte migration.³  This ability is achieved largely through conformational changes that are coupled to ligand binding.³  Intracellular signals, elicited by other receptors that sense external stimuli, trigger conformational changes in the cytoplasmic domain, which are then propagated across the plasma membrane to the extracellular domain.^1,2  Under resting conditions, integrins exist largely in a low-affinity, nonadhesive conformation, and upon stimulation are converted to high-affinity conformations. The integrin cytoplasmic domain plays a pivotal role in modulating the activity of the integrins to bind ligand.^4,5  The cytoplasmic domains of the α- and β-subunits in integrins are thought to associate with each other at the membrane-proximal regions in latent low-affinity state (Figure 1A left). This αβ cytoplasmic association functions as a “clasp” that restrains integrins in a default low-affinity conformation. Upon activation, binding to the integrin cytoplasmic domains of signaling molecules (eg, talin) triggers “unclasping,” which leads to the conversion from low-affinity to high-affinity conformations (Figure 1A right).⁶ 

Structural investigations using NMR revealed critical interactions at the association interface between α/β cytoplasmic domains.⁷  The arginine residue in the conserved GFFKR sequence (termed R^GFFKR in this study, Figure 1B) at the membrane-proximal region of the α-subunit forms a putative salt bridge with the β-subunit. This putative membrane-proximal salt bridge is thought to constitute a critical interaction that contributes to clasping the α/β cytoplasmic domains and thereby restrains integrin activation (Figure 1A left). R/A^GFFKR, the alanine substitution of R^GFFKR that would disrupt the clasping salt bridge selectively, constitutively activated α_Lβ₂⁸  and α_IIbβ₃⁵  in transfectants. In addition, β₃-D723A, the alanine substitution of the conserved membrane-proximal aspartate in the β-subunit that would form a salt bridge with R^GFFKR in the α-subunit, led to a constitutively active α_IIbβ₃ integrin in transfectants.⁵  Furthermore, these in vitro results using transfectants have been supported by findings in knock-in mice for α_Lβ₂ and in patients for α_IIbβ₃. The deletion of the α_L-GFFKR sequence in germ line persistently activated α_Lβ₂ in a strain of knock-in mice (Lfa-1^d/d).⁹  β₃-D723H, the mutation found in a group of patients with inherited thrombocytopenia, increased activity of α_IIbβ₃ integrin in patients' platelets.¹⁰ 

In contrast to α_IIbβ₃ and α_Lβ₂ integrins, the physiological role of the putative membrane-proximal salt bridge remains to be elucidated in α₄ integrins (α₄β₁ and α₄β₇). α₄ integrins play a pivotal role in lymphocyte adhesion to, and transmigration across, the endothelium during physiologic lymphocyte migration, including tissue-specific homing to gut-associated lymphoid tissues (GALTs) as well as pathologic leukocyte accumulation during inflammation.^11,12  The activity of α₄ integrins is dynamically regulated. The ability of α₄ integrins to bind ligands is normally maintained in a low-affinity state, and rapidly up-regulated only upon activation with exogenous stimuli.^13,14  Thus, it is of great importance to investigate the physiological significance of the putative membrane-proximal salt bridge in regulating α₄ integrin activation.

Here we attempt to disrupt the putative membrane-proximal salt bridge in the integrin α₄ subunit in mouse germ line. The deletion of the entire GFFKR sequence, which was previously used in Lfa-1^d/d knock-in mice,⁹  would not only disrupt the salt bridge via R^GFFKR, but also substantially perturb the α/β cytoplasmic association interface. Thus, to selectively disrupt the putative membrane-proximal salt bridge, we knocked in a point mutation α₄-R/A^GFFKR. We have shown that compared with wild type (WT), lymphocytes from α₄-R/A^GFFKR mice exhibit constitutively enhanced cell adhesion to VCAM-1 and MAdCAM-1, thus demonstrating the importance of the membrane-proximal salt bridge in restraining integrin activation in α₄β₁ and α₄β₇. Furthermore, α₄-R/A^GFFKR lymphocytes exhibit perturbed lateral migration on, and transmigration across, VCAM-1 and MAdCAM-1 substrates as well as delayed transendothelial migration. In vivo, α₄-R/A^GFFKR lymphocytes display an increased capacity to firmly adhere to Peyer patch (PP) venules but a reduced capacity to home to the GALT. These data support the idea that the putative cytoplasmic membrane-proximal bridge plays a pivotal role in the balanced regulation of α₄ integrin adhesiveness to facilitate cell migration.

Supplemental methods are available in Document S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article).

All experiments using mice were approved by the Institutional Review Board of the Immune Disease Institute.

Flow chamber assays were performed as previously described¹⁵  using mouse VCAM-1-Fc (10 μg/mL), MAdCAM-1-Fc (10 μg/mL), or ICAM-1/Fc (50 μg/mL) immobilized on plastic. Cells were resuspended at 10⁶/mL for interactions with VCAM-1 and ICAM-1, and at 4 × 10⁶/mL for interactions with MAdCAM-1, in Hanks balanced salt solution, 10 mM HEPES, 0.5% BSA containing 1 mM Mg²⁺/Ca²⁺ or 1 mM Mn²⁺. Cells were infused into the chamber and allowed to accumulate on the substrate at a shear stress of 0.3 dyne/cm² for 45 seconds. Wall shear stress was increased incrementally every 10 seconds. Microscopic images of cells under flow were recorded and analyzed off-line as previously described.¹⁶ 

Transmigration was studied as described previously with minor modifications using a modified Boyden chamber assay with a 6.5-mm Transwell tissue culture system (Corning, Corning, NY) containing a permeable support insert with a 5-μm pore size.¹⁷  In brief, Transwell inserts were coated with either 40 μg/mL murine MAdCAM-1-Fc, VCAM-1-Fc, or ICAM-I-Fc. Splenocytes (10⁶) were added to the upper chamber and allowed to transmigrate for 4 hours to the lower chamber that contained 2 μg/mL CXCL12. Transmigrated cells were counted using a FACScan flow cytometer (BD Biosciences, San Jose, CA).

Transendothelial migration (TEM) under shear stress was studied as previously described¹⁸  using bEnd.3 mouse brain endothelioma cells (passages 6 to 10; ATCC, Manassas, VA). bEnd.3 cells were plated at 0.5 × 10⁶ cells per 40 mm on circular coverslips coated with 20 μg/mL fibronectin and cultured for 48 hours. The cells were stimulated for 24 hours with 50 ng/mL TNF-α, and 2 μg/mL CCL21 was overlaid for 15 minutes on the bEnd.3 cell monolayer. Coverslips with endothelium were then assembled into an FCS2 parallel wall laminar shear flow chamber (Bioptechs, Butler, PA). For transendothelial migration analysis, splenocytes were perfused for 2 minutes over the monolayer at 0.25 dyne/cm² to allow accumulation. The flow rate was then increased to 2 dyne/cm² and kept constant for an additional 15-minute period during which time the images were recorded. Cell migration was visualized using a 10× phase-contrast objective and images were recorded using a CCD camera at a rate of 1 frame per 5 seconds. Cell adhesion and transmigration were assessed in the resulting videos by counting the number of lymphocytes adherent and laterally migratory over the apical surface of the endothelium (bright diffraction in phase-contrast imaging), as well as those that underwent transmigration to the subendothelial space (transition to a dark appearance in phase-contrast imaging).

Live-cell imaging of laterally migrating cells on VCAM-1 and MAdCAM-1 substrates was performed as previously described.¹⁹  Briefly, 20 μg/mL VCAM-1-Fc or MAdCAM-1-Fc with 16 μg/mL CXCL12 was coated on Delta T chambers (Bioptechs) at 4°C overnight. IL-15–treated memory/effector T cells were generated as previously described.¹⁹  After washing with L-15 medium (Cambrex Bioscience, East Rutherford, NJ), cells were added to the chambers and differential interference contrast (DIC) images were acquired using an Axiovert S200 epifluorescence microscope (Carl Zeiss, Heidelberg, Germany) equipped with a 63× oil objective coupled to an Orca CCD camera (Hamamatsu, Okayama City, Japan) at a frame rate of 15 seconds per frame. Cell migration was analyzed as previously described.²⁰ 

Fluorescent imaging was performed as previously described.¹⁹  Briefly, cells spread on 20 μg/mL VCAM-1 or MAdCAM-1 coimmobilized with 10 μg/mL CXCL12 were fixed and stained with mAb to α₄ integrin (PS/2) followed by Cy3-conjugated goat anti–rat IgG and phalloidin–Alexa 488 for actin. Confocal imaging was conducted on a Radiance 2000 Laser-scanning confocal system (Bio-Rad Laboratories, Hercules, CA) using a microscope (model BX50BWI; Olympus, Melville, NY) with a 100× water-immersion objective. Imaging processing was performed with OpenLab software (Improvision, Waltham, MA). Multiple randomly selected fields were imaged for each condition by collecting serial sections through the entire sample Z-axis. These were then projected as complete stacks. For morphologic analysis, cell polarity was identified as the actin-rich crescent-shaped leading edge. The opposing trailing edge was assessed for the presence of extended (> 1 μm) tethers that were rich in both actin and α₄-integrin staining. The number of such projections was counted for each of at least 20 cells in randomly selected fields and averaged.

Intravital microscopy of PPs was performed as previously described.²¹  Briefly, C57BL/6J mice were anesthetized and a bowel segment of the small intestine was positioned for epifluorescence intravital microscopy, to record an individual PP. Preparations were transferred to an intravital microscope (IV-500; Mikron Instruments, San Diego, CA), equipped with a Rapp OptoElectronic SP-20 xenon flash lamp system (Hamburg, Germany) and QImaging Rolera-MGi EMCCD (Surrey, BC). WT or KI splenocytes were labeled with 1 ng/mL calcein-AM for 20 minutes at 37°C, and small boluses (10-50 μL; concentration of 10 × 10⁶ cells/mL) of cells were injected via a left carotid artery catheter. Cell behaviors in PP venules were recorded in 10-minute recordings through 10× or 20× water-immersion objectives (Achroplan; Carl Zeiss). Rolling and sticking fractions in individual vessel segments were determined off-line by playback of digital video files. Rolling fraction was determined as the percentage of cells interacting with PP venules in the total number of cells passing through a vessel during the observation period. Sticking fraction was defined as the percentage of rolling cells that adhered in PP venules for 30 or more seconds.²¹ 

A competitive homing assay was conducted as previously described²²  using WT C57BL/6J-CD45.1⁺ congeneic recipient mice with or without gut inflammation. To induce gut inflammation, mice were fed for 6 days with 3% (wt/vol) DSS in drinking water.²³  On day 5, the same numbers of donor splenocytes (2 × 10⁷) from WT and KI mice labeled with different dyes (CFSE and CMTMR) were mixed and injected intravenously. Eighteen hours later, mice were killed and all lymphoid and nonlymphoid organs were harvested and analyzed as described.²³  In some experiments, fluorescent dyes were switched.

Data are expressed as the mean plus or minus SEM for each group. The Student t test was used for statistical analyses unless otherwise indicated.

To disrupt the critical salt bridge that stabilizes the α/β cytoplasmic association in α₄ integrins in vivo, we used a replacement-type gene-targeting strategy^24,25  to mutate R^GFFKR to alanine in the integrin α₄ gene (Itga4), thereby generating an Itga4-R/A^GFFKR mutated allele in embryonic stem (ES) cells (Figure 1C). We obtained the correct integration of the Itga4-R/A^GFFKR allele into the mouse germ line (Figure 1D and data not shown) and designated mice homozygous for the mutant α₄ allele as α₄-R/A^GFFKR. These mice were born under normal Mendelian ratios, were fertile, and did not exhibit gross abnormalities.

Lymphocytes from the spleen of α₄-R/A^GFFKR mice showed approximately 50% reduction of α₄ and α₄β₇ expression on the cell surface, compared with WT cells (Figure 2A). Only a slight decrease in β₁-subunit expression was observed. Expression of the α_Lβ₂ integrin, L-selectin, and the activation markers CD69 and CD25 was not affected (Figure 2A and data not shown). Similar results were seen in lymphocytes from peripheral lymph nodes (PLNs) and Peyer patches (PPs) (Figure S1).

Despite the reduced cell-surface α₄ integrin expression in α₄-R/A^GFFKR lymphocytes, WT and α₄-R/A^GFFKR cells showed comparable levels of total α₄ integrin expression, which included cell-surface and intracellular expression, as determined by immunofluorescent cytometry analyses of permeabilized cells (Figure 2B). Quantitative reverse-transcription–polymerase chain reaction (RT-PCR) showed that mRNA expression levels of the α₄ subunit in WT and α₄-R/A^GFFKR splenocytes were comparable (Figure 2C). Similar results with α_Lβ₂ (ie, reduced cell-surface and comparable total expression) were also obtained in Lfa-1^d/d knock-in mice, in which the α_L GFFKR sequence had been deleted.⁹  As previously shown in the α_L-subunit,⁸  heterodimer formation and subsequent translocation to the cell surface might be partially perturbed in the context of the mutant α₄ subunit, possibly through effects on integrin conformation.

To investigate the possibility that a potential increase in an intracellular pool of mutant α₄ integrin subunit might activate the unfolded protein response (UPR) that could modify cellular metabolism and functions,²⁶  we studied UPR in WT and α₄-R/A^GFFKR splenocytes. Using quantitative RT-PCR, we measured the mRNA levels of GRP78²⁷  and the spliced form of XBP-1,²⁸  both of which are known to be up-regulated in response to UPR. We have shown that WT and α₄-R/A^GFFKR cells expressed comparable levels of GRP78 and the spliced XBP-1 (Figure 2D). Thus, at least in this experimental setting, we did not find evidence that α₄-R/A^GFFKR cells are subjected to an increased UPR, compared with WT cells.

We investigated the adhesive interactions of α₄ integrins with their ligands VCAM-1 and MAdCAM-1 under physiological shear conditions using a parallel wall flow chamber.¹⁶  A fraction of WT splenic lymphocytes firmly adhered to VCAM-1 and MAdCAM-1 substrates in approximately 1 mM Ca²⁺ + Mg²⁺, the cation condition in which integrins are predominantly maintained in a nonadhesive state (Figure 3A,B). Under the same setting, approximately 3 times as many α₄-R/A^GFFKR lymphocytes firmly adhered as did WT cells, demonstrating increased basal adhesiveness by α₄ integrins in α₄-R/A^GFFKR lymphocytes (Figure 3A,B). In the current experimental setting, we observed only a small fraction of WT and α₄-R/A^GFFKR lymphocytes that rolled on VCAM-1 and MAdCAM-1 substrates (Figure S2). Increased adhesiveness was specific to the α₄ integrins, as α_Lβ₂ in α₄-R/A^GFFKR splenocytes remained as latent in binding to ICAM-1 as did WT cells (Figure S3). To study the relative contributions of α₄β₁ and α₄β₇ integrins to the enhanced adhesion by α₄-R/A^GFFKR lymphocytes, we used a panel of function-blocking mAbs. Adhesion of α₄-R/A^GFFKR and WT cells to VCAM-1 was blocked completely with anti-α₄ mAb, and partially with either anti-β₁ or anti-α₄β₇ blocking mAb, confirming that the interaction was mediated by both α₄β₁ and α₄β₇ integrins (Figure 3E,G,I). In the presence of either anti-β₁ or anti-α₄β₇ blocking mAb, adhesion of α₄-R/A^GFFKR cells was greater than WT, demonstrating that the adhesiveness of both α₄β₁ and α₄β₇ integrins to VCAM-1 had increased (Figure 3E,G,I). By contrast, adhesion of α₄-R/A^GFFKR and WT cells to MAdCAM-1 was blocked completely with anti-α₄ mAb, or by anti-α₄β₇ blocking mAb, but was rarely affected with anti-β₁ blocking mAb, indicating that enhanced adhesion of α₄-R/A^GFFKR lymphocytes to MAdCAM-1 was solely mediated by the α₄β₇ integrin (Figure 3F,H,J). Upon stimulation with Mn²⁺ capable of mimicking inside-out signaling to induce high-affinity integrin conformations, WT cells exhibited increased firm adhesion to VCAM-1 and MAdCAM-1 (Figure 3C,D). Mn²⁺ stimulation enhanced firm adhesion of α₄-R/A^GFFKR cells to VCAM-1 and MAdCAM-1 (Figure 3C,D), suggesting that α₄ integrins in α₄-R/A^GFFKR cells are not fully activated and are further activatable with agonists. These results convincingly demonstrated that the disruption of the α₄ cytoplasmic membrane-proximal salt bridge increased the adhesiveness of both α₄β₁ and α₄β₇, despite the reduced cell-surface expression of mutant α₄ integrins. Similar results were obtained with cells from PLNs and PPs (data not shown).

As an appropriate balance between adhesion and deadhesion is thought to be critical for efficiently coordinated cell migration, we studied how the persistently increased adhesion of α₄-R/A^GFFKR cells affected transmigration by using a Transwell assay.¹⁷  WT and α₄-R/A^GFFKR cells transmigrated equally well through an ICAM-1–coated Transwell insert into a chemokine CXCL12–containing lower chamber, demonstrating normal migration behavior in general and normal coordination of α_Lβ₂ integrin adhesiveness, specifically in α₄-R/A^GFFKR (Figure 4A). In contrast, when Transwell inserts were coated with VCAM-1 or MAdCAM-1, α₄-R/A^GFFKR cells showed approximately 50% reduction of transmigrated cells compared with WT (Figure 4B,C).

Transendothelial migration (TEM) is a critical step that regulates lymphocyte entry into lymph nodes as well as leukocyte accumulation into inflamed tissues. During lymphocyte TEM, α₄ integrins cooperate with α_Lβ₂ integrin and other adhesion molecules.²⁹  The presence of shear stress enhances the capacity of lymphocytes to transmigrate through endothelial monolayers.³⁰  To study transendothelial migration by WT and α₄-R/A^GFFKR lymphocytes under physiologic shear stress, we used monolayers of bEnd.3 mouse endothelial cells assembled into a flow chamber. The monolayers of bEnd.3 cells were treated with TNF-α to up-regulate surface expression of integrin ligands such as VCAM-1, MAdCAM-1, and ICAM-1. A CCL21 chemokine was immobilized on the endothelial surface to facilitate adhesion to, and transmigration through, endothelial cells. Identical numbers of WT and α₄-R/A^GFFKR cells were infused into the chamber and allowed to accumulate on endothelial monolayers for 2 minutes under 0.25 dyne/cm² shear stress. The migratory behaviors of adherent lymphocytes were then recorded under 2.0 dyne/cm² shear stress for 15 minutes. Although α₄-R/A^GFFKR cells were more adhesive to VCAM-1 or MAdCAM-1 in the absence of chemokine stimulation (Figure 3A,B), comparable numbers of WT and α₄-R/A^GFFKR cells accumulated on bEnd.3 endothelial monolayers (WT, 89.7 ± 4.0; KI, 94.3 ± 2.5 cells per field). Surface-displayed CCL21 is likely to activate α_Lβ₂ and α₄ integrins on lymphocytes to support firm adhesion on bEnd.3 cells that simultaneously express VCAM-1, MAdCAM-1, and ICAM-1. However, when allowed to transmigrate at 2.0 dyne/cm² shear stress for 15 minutes, approximately 50% as few α₄-R/A^GFFKR cells underwent TEM as did WT cells (Figure 4D).

To investigate the mechanism(s) by which TEM of α₄-R/A^GFFKR lymphocytes was perturbed, we studied lateral migration with live-cell time-lapse video microscopy. Although we tried to use naive lymphocytes in which basal integrin adhesiveness is maintained at low levels, they failed to show an efficient lateral migration on integrin ligands in response to chemokine. Similar observations were made in previous reports.³¹  As an alternative, we used IL-15–treated memory/effector T cells, in which integrins are known to show little basal ligand binding, but readily up-regulate upon chemokine stimulation.³² 

WT T cells migrated smoothly on VCAM-1/CXCL12 substrates (Movie S1). By contrast, α₄-R/A^GFFKR T cells migrated poorly on VCAM-1/CXCL12 substrates (Video S2). Whereas WT cells migrating on VCAM-1 were polarized normally, displaying a typically hand mirror–like shape with a flattened leading edge followed by a short tail, α₄-R/A^GFFKR T cells migrating on VCAM-1 exhibited extremely stretched tails enriched in α₄ integrin (Figure 5A). Close image analysis reveals (Figure 5B) that the most remarkable morphologic distinction between 2 types of cells is the strong presence of α₄ integrin–rich trailing edge tethers (Figure 5B arrows) in α₄-R/A^GFFKR cells but not WT cells. Quantitative image analysis demonstrated that α₄-R/A^GFFKR cells displayed significantly more tethers than WT cells (Figure 5C). Similar results were obtained in T cells migrating on MAdCAM-1 substrates (Figure 5A,C). These results strongly suggest that the perturbed detachment of the tail accounts for decreased TEM of α₄-R/A^GFFKR cells across bEnd.3 endothelial monolayers.

We have demonstrated proof of principle that the aberrantly activated α₄-R/A^GFFKR perturbs TEM across bEnd.3 cell line under shear stress; however, TNF stimulation of, and apical addition of CCL21 to, bEnd.3 cells might not fully recapitulate in vivo phenotypes of HEVs in the gut during normal lymphocyte homing. To study the in vivo adhesive interactions of α₄-R/A^GFFKR lymphocytes with HEVs in the GALT, such as PPs, we performed an intravital microscopic investigation of PP venules.²¹ 

We injected calcein-labeled α₄-R/A^GFFKR or WT splenocytes into anesthetized WT syngeneic recipient mice that had been prepared for intravital microscopy of PPs. The injected cells were then observed and video recorded entering and interacting with PP venules. Adhesive interactions of α₄-R/A^GFFKR or WT donor splenocytes with PP venules were analyzed off-line. We found that α₄-R/A^GFFKR and WT cells rolled on PP venules at similar frequencies (Figure 6A) and comparable velocities (Figure 6C). However, α₄-R/A^GFFKR cells exhibited higher capacity to firmly adhere to PP venules than WT cells (Figure 6B). These data demonstrated that despite reduced cell-surface expression of α₄ integrin, α₄-R/A^GFFKR cells showed increased firmed adhesion to PP venules in vivo.

α₄ integrin plays a pivotal role in a physiologic lymphocyte recirculation through the GALT as well as a pathologic lymphocyte accumulation to inflamed gut.^11,12  To study the impact of aberrantly activated α₄ integrins on lymphocyte migration to the gut, we used a competitive homing assay. We began with investigating lymphocyte homing to noninflamed tissues. Splenocytes from WT and α₄-R/A^GFFKR mice were fluorescently labeled with CFSE and CMTMR, respectively. Equal numbers of WT and α₄-R/A^GFFKR cells were mixed and intravenously administered into C57BL/6J-CD45.1⁺ mice. Eighteen hours later, WT and α₄-R/A^GFFKR lymphocytes homed equally well to SP and PLNs (Figure 7A), and comparable fractions of WT and α₄-R/A^GFFKR lymphocytes were present in peripheral blood (WT, 0.59% ± 0.18%; and α₄-R/A^GFFKR, 0.57% ± 0.18%; of injected cells [mean ± SEM, n = 6]). By contrast, homing of α₄-R/A^GFFKR lymphocytes to the gut (including MLNs, PPs, small and large intestines) was reduced by approximately 50% compared with WT (Figure 7A). To study whether a subpopulation of α₄-R/A^GFFKR cells with regard to α₄ surface expression might selectively be blocked to enter tissues, we compared α₄ integrin expression of WT and α₄-R/A^GFFKR cells in donor cells before injection and those homed to tissues. The ratio of α₄ expression in WT and KI cells is comparable in both donor cell populations (Table S1).

We then investigated lymphocyte homing to inflamed gut. We induced mild colitis in C57BL/6J-CD45.1⁺ recipient mice by feeding them for 6 days with 3% DSS. At day 5, mice exhibited approximately 4% body weight loss and intestinal edema formation (not shown). We performed a new set of in vivo competitive homing assays using the recipient mice with DSS-induced colitis. Eighteen hours after injection of equal numbers of fluorescently labeled WT and α₄-R/A^GFFKR cells, cells were harvested from inflamed gut and other organs, and the homing indices were then determined. WT and α₄-R/A^GFFKR lymphocytes homed equally well to SP and PLNs. However, α₄-R/A^GFFKR lymphocytes exhibited approximately 50% reduced homing to the gut, compared with WT (Figure 7B). These results demonstrated that the aberrant activation of α₄ integrins perturbed lymphocyte homing to both uninflamed and inflamed gut.

The capacity of lymphocytes to migrate/home represents an important factor in regulating the size and cellularity of lymphoid tissues,³³  which could be affected by the perturbed migration of lymphocytes in α₄-R/A^GFFKR mice. Therefore, we studied the distribution of lymphocytes in lymphoid organs. Macroscopic examination showed that the PPs from α₄-R/A^GFFKR mice were smaller than those of WT (WT, 1.4 ± 0.1 mm; α₄-R/A^GFFKR, 0.9 ± 0.1 mm; diameter, P < .01), whereas SP, PLNs, and MLNs were indistinguishable from one mouse strain to another (data not shown). Consistent with the results of our in vivo homing assay, α₄-R/A^GFFKR mice contained significantly fewer lymphocytes in the gut (ie, PP and IEL compartments) compared with WT (Table 1).

We have demonstrated that disruption of the putative cytoplasmic membrane-proximal salt bridge leads to constitutively activated α₄β₁ and α₄β₇ integrins. Our results with α₄-R/A^GFFKR knock-in mice have convincingly demonstrated that the membrane-proximal salt bridge plays a critical role in restraining default low-affinity state of α₄ integrins in physiologically relevant settings.

The importance of the putative membrane-proximal salt bridge in β₁ integrins including α₄β₁ remains to be questioned. Czuchra et al previously knocked in an alanine substitution, in the β₁ subunit, of a membrane-proximal aspartate D759 that would form a salt bridge with R^GFFKR in the α-subunit.³⁴  Despite the hypothesis that the mutation would activate β₁ integrins, β₁-D759A keratinocytes failed to show an increase in cell adhesion to β₁ integrin ligands; and β₁-D759A bone marrow cells did not show activation-dependent conformational changes in the β₁-subunit. These results led Czuchra et al to argue that the membrane-proximal salt bridge might not be important for physiologic regulation of β₁ integrins. It has yet to be directly demonstrated whether α₄β₁ adhesiveness in β₁-D759A mice is enhanced; however, the effectiveness of perturbing the membrane-proximal salt bridge in α₄β₁ might substantially differ in β₁-D759A and α₄-R/A^GFFKR. The NMR solution structure for the α_IIbβ₃ cytoplasmic domain showed that R^GFFKR (ie, α_IIb-R995) interacts not only with the membrane-proximal aspartate (β₃-D723), but also with 2 other membrane-proximal conserved residues, histidine (β₃-H722) and glutamate (β₃-E726).⁷  Therefore, in β₁-D759A mice, perturbation by the D759A mutation might be at least partially compensated by intact β₁-H758 and/or β₁-E762, thereby creating a much milder effect on the adhesiveness of β₁ integrins. By contrast, the R/A^GFFKR mutation disrupts all 3 electrostatic interactions, thereby inducing more robust activating phenotypes. Alternatively, R/A^GFFKR mutation might destabilize the cytoplasmic α/β association via perturbing the local structural integrity of the α₄ cytoplasmic domain. Future studies are necessary to determine the structure of α₄β₁ and α₄β₇ cytoplasmic domains.

Persistently increased adhesion of α₄-R/A^GFFKR cells perturbed transmigration through VCAM-1 and MAdCAM-1. An appropriate balance between adhesion at the leading edge and deadhesion at the trailing edge is important for efficient cell migration.³  As shown by elongated tails of R/A^GFFKR cells laterally migrating on VCAM-1 and MAdCAM-1, perturbed detachment at the trailing edge is likely to suppress directional forward movements of α₄-R/A^GFFKR cells. Our results are consistent with previous studies involving either other strains of Lfa-1^d/d knock-in mice expressing constitutively active α_Lβ₂⁹  or β₇ (D146A) knock-in mice expressing constitutively active α₄β₇ integrin via a mutation in the negative regulatory metal-binding site ADMIDAS (adjacent to metal ion-dependent adhesion site).¹⁹  Lymphocytes from both knock-in strains showed disturbed lateral migration on, and transmigration across, substrates bearing the corresponding integrin ligands. In addition, aberrantly activated α_Lβ₂ or α₄β₇ integrin delayed TEM in Transwell assays performed in the absence of shear stress. Since the presence of shear stress, a more physiologically relevant condition, promotes TEM, it is of great interest to study the impact of aberrantly activated integrins on shear stress–enhanced TEM. We have shown the proof of principle that aberrantly activated α₄ integrins interfered with shear stress–enhanced TEM. Although the ability of α_Lβ₂ to support TEM appears to be intact in α₄-R/A^GFFKR, impaired detachment of α₄ integrins could interfere with α_Lβ₂-driven diapedesis and migration of α₄-R/A^GFFKR cells.

In cooperation with α_Lβ₂, α₄β₇ plays a dominant role over α₄β₁ in supporting lymphocyte homing to the GALT.^35,36  Thus, although α₄-R/A^GFFKR mice exhibited constitutive activation of both α₄β₁ and α₄β₇ integrins, aberrantly activated adhesion of α₄β₇ to MAdCAM-1 appears to play a major role in the reduced homing of α₄-R/A^GFFKR cells to the GALT in uninflamed gut. By contrast, adhesive interactions of aberrantly activated α₄ integrins with not only MAdCAM-1 but also VCAM-1 might be involved in reduced lymphocytes homing to DSS-induced inflamed gut.^37,38  How did aberrantly activated α₄ integrin reduce lymphocyte homing to the gut? The most plausible explanation is that TEM through high endothelial venules (HEVs) in the GALT is perturbed by aberrantly activated α₄β₇ integrins that suppress detachment at the trailing edge. This is supported by in vitro TEM experiments under shear stress, which showed delayed TEM by α₄-R/A^GFFKR cells. In homing assays, α₄-R/A^GFFKR cells that had become firmly adhered to HEVs, but had failed to enter GALT interstitial spaces, might eventually become detached. Although the reduced expression of α₄β₇ in α₄-R/A^GFFKR might reduce lymphocyte adhesion to gut HEVs and, thereby, homing to the gut, our intravital microscopic observation that α₄-R/A^GFFKR cells exhibited in vivo an increased cell adhesion to PP venules favors the alternative mechanism (ie, aberrant activation). We believe that the aberrant activation of α₄β₇ represents the primary mechanism interfering with gut homing. As α₄ integrins in α₄-R/A^GFFKR cells are persistently adhesive, independent of activation, it is possible that α₄-R/A^GFFKR lymphocytes are entrapped in α₄ integrin ligand-expressing organs outside the GALT, thereby reducing cell homing to the GALT. However, similar numbers of α₄-R/A^GFFKR and WT donor cells remained in the peripheral blood 18 hours after injection. This result suggests that both WT and α₄-R/A^GFFKR cells in circulation were equally available to the GALT and other lymphoid tissue. Thus, entrapment is unlikely to be the major cause of reduced GALT homing by α₄-R/A^GFFKR cells. Future investigations would be required to better understand how the aberrantly activated α₄ integrin perturbs lymphocyte migration to the gut.

In summary, using knock-in mice we have demonstrated that the putative membrane-proximal cytoplasmic salt bridge is critical for maintaining default nonadhesive states of α₄β₁ and α₄β₇ integrins and consequently in supporting properly regulated integrin adhesive dynamics. The membrane-proximal salt bridge constitutes an important regulatory component within the integrin, one that facilitates the appropriate adhesion and deadhesion balance required for efficient and coordinated migration of lymphocytes to the gut.

We gratefully acknowledge Dr Timothy A. Springer for his valuable advice, Dr Mario R. Capecchi for the self-excision cassette used in pACN-TV, and Dr Klaus Rajewsky for Bruce-4 ES cells. We thank Dr Xiaohui Zhang, Dr Koichi Yuki, Ms Mary Mohrin, Ms Ronnie Yoo, and Ms Whitney Silkworth for their technical assistances.

This work was supported by grants from the Arthritis Foundation (Atlanta, GA; C.V.C.) and the National Institutes of Health (Bethesda, MD), AI063421 and HL048675 (M.S.). Y.I. was partially supported by fellowships from the Nakajima Foundation (Tokyo, Japan), the Japan Intractable Diseases Research Foundation (Tokyo, Japan), the Yamada Science Foundation (Osaka, Japan), and the Uehara Memorial Foundation (Tokyo, Japan).

National Institutes of Health

Contribution: Y.I., E.J.P., and C.V.C. designed and performed research, analyzed data, and wrote the paper; D.P., A.P., and G.C. designed and performed research and analyzed data; U.H.A. designed research and analyzed data; and M.S. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Motomu Shimaoka, 200 Longwood Avenue, Boston, MA 02115; e-mail: shimaoka@idi.harvard.edu.