Abstract
The α4 integrin, α4β7, plays an important role in recruiting circulating lymphocytes to the gastrointestinal tract, where its ligand mucosal addressin cell adhesion molecule-1 (MAdCAM-1) is preferentially expressed on high endothelial venules (HEVs). Dual antagonists of α4β1 and α4β7,N-(2,6-dichlorobenzoyl)-(l)-4-(2′,6′-bis-methoxyphenyl)phenylalanine (TR14035) andN-{N-[(3,5-dichlorobenzene)sulfonyl]-2-(R)-methylpropyl}-(d)-phenylalanine (compound 1), were tested for their ability to block the binding of α4β7-expressing cells to soluble ligand in suspension and under in vitro and in vivo shear flow. Compound 1 and TR14035 blocked the binding of human α4β7to an 125I-MAdCAM-Ig fusion protein with IC50values of 2.93 and 0.75 nM, respectively. Both compounds inhibited binding of soluble ligands to α4β1 or α4β7 on cells of human or rodent origin with similar potency. Under shear flow in vitro, TR14035 and compound 1 blocked binding of human α4β7-expressing RPMI-8866 cells or murine mesenteric lymph node lymphocytes to MAdCAM-Ig with IC50 values of 0.1 and 1 μM, respectively. Intravital microscopy was used to quantitate α4-dependent adhesion of fluorescent murine lymphocytes in Peyer's patch HEVs. When cells were prestimulated with 2 mM Mn2+ to activate α4β7 binding to ligand, anti-α4 monoclonal antibody (mAb) [10 mg/kg (mpk) i.v.] blocked adhesion by 95%, and anti-β1 mAb did not block adhesion, demonstrating that this interaction was dependent on α4β7. TR14035 blocked adhesion to HEVs [ED50 of 0.01–0.1 mpk i.v.], and compound 1 blocked adhesion by 47% at 10 mpk i.v. Thus, α4β7/α4β1antagonists blocked α4β7-dependent adhesion of lymphocytes to HEVs under both in vitro and in vivo shear flow.
The ability of lymphocytes to arrest under conditions of vascular flow enables their movement into both normal lymphoid tissues and sites of inflammation. Lymphocyte recruitment in the vasculature is regulated by the differential expression and activation of homing receptors (selectins and integrins) on lymphocytes that interact with counter-receptors of the Ig superfamily on high endothelial venules (HEVs) (Bargatze and Butcher, 1993; Bargatze et al., 1995). This interaction mediates a multistep process, involving rolling and tethering of leukocytes to endothelial ligands, rapid activation of integrins by locally released chemokines, stable adhesion of activated integrins to endothelial ligands, and transendothelial migration through the vessel wall (Bargatze et al., 1995; Warnock et al., 2000). Although all integrins expressed on leukocytes can mediate firm adhesion, α4β7 and α4β1 are members of a small subset of integrins that can also mediate rolling (Berlin et al., 1995).
Mucosal addressin cell adhesion molecule-1 (MAdCAM-1), expressed on HEVs of Peyer's patch and other gut-associated lymphoid tissues (GALTs), is the principal ligand for α4β7, an integrin highly expressed on gut-homing memory lymphocytes (Berlin et al., 1993;Shyjan et al., 1996; Briskin et al., 1997). α4β7 binds to MAdCAM-1 with higher affinity than to VCAM-1 or the CS-1 subdomain of human fibronectin. Although both α4β7 and α4β1 can bind VCAM-1 and CS-1, α4β1 does not bind MAdCAM-1 (Berlin et al., 1993). Both l-selectin and α4β7 mediate the initial attachment and rolling of lymphocytes by interacting with MAdCAM-1, whereas α4β7also mediates the firm adhesion of lymphocytes via this ligand (Rott et al., 1996).
Lymphocyte trafficking in the GALT not only enables normal immune responses (Butcher and Picker, 1996) but also contributes to unwanted inflammation (Podolsky and Fiocchi, 2000). Gut inflammation can induce dramatic changes in the extent and selectivity of lymphocyte recruitment to the gut wall (Briskin et al., 1997; Picarella et al., 1997). For example, the expression of MAdCAM-1 can be up-regulated by as much as 5-fold on blood vessels at sites of intestinal inflammation (Briskin et al., 1997), and proinflammatory cytokines facilitate the recruitment of lymphocytes and other leukocytes to sites of active inflammation (Podolsky and Fiocchi, 2000). The tissue-specific distribution of MAdCAM-1 and selective interaction with gut-homing memory lymphocytes expressing α4β7 suggest a contributing role of this ligand-receptor pair to inflammatory bowel diseases such as Crohn's disease and ulcerative colitis.
Blockade of α4β7 and MAdCAM-1 with antibodies defines their role in models of inflammatory bowel disease. Monoclonal antibodies (mAbs) directed against α4 or α4β7 block lymphocyte homing to intestinal sites in naı̈ve mice (Hamann et al., 1994), and mAbs against β7 or MAdCAM-1 reduce inflammation in mouse models of colitis (Picarella et al., 1997; Kato et al., 2000). In the cotton-top tamarin, which spontaneously develops colitis, an anti-α4β7mAb effectively resolved the established colitis (Hesterberg et al., 1996). Perhaps the strongest argument for the importance of α4 integrins in mediating inflammation of the gut is derived from recent clinical trials with Antegren (anti-α4; Elan/Biogen, Cambridge, MA). In a blinded placebo-controlled phase II trial of 248 patients with moderate-to-severe Crohn's disease, Antegren administered at a single dose of 3 mpk i.v. resulted in a 46% remission rate after 6 weeks, versus 27% remission with placebo (Ghosh et al., 2001).
We have identified a dual α4β7/α4β1antagonist by evaluating the ability of the compound to block the binding of human or murine α4β7-expressing cells to soluble human or murine MAdCAM-Ig. Small molecule antagonists of α4β7 that block the static adhesion of human α4β7-expressing cells to the CS-1 subdomain of human fibronectin, human VCAM-Ig, human MAdCAM-Ig, or murine MAdCAM-Ig have been described previously (Shroff et al., 1996, 1998; Carson et al., 1997; Harriman et al., 1999; Martin et al., 1999). The ability of α4β7 antagonists to block the binding of murine α4β7-expressing cells to soluble murine MAdCAM-Ig under static conditions has also been reported (Martin et al., 1999), but the ability of compounds to block ligand binding under in vitro or in vivo shear flow conditions has not been examined. We used in vitro shear flow assays to quantitate the adhesion of both human and murine α4β7-expressing cells to human and murine MAdCAM-Ig, and we characterized the ability of compounds to inhibit adhesion to HEVs in an in vivo model.
Materials and Methods
Antibodies and Cell Lines
The following purified monoclonal antibodies were obtained from BD PharMingen (San Diego, CA): 4B4 (mouse anti-human β1), FIB27 (rat anti-mouse β7 that cross-reacts with human β7), DATK32 (rat anti-mouse α4β7), Ha2/5 (hamster anti-rat β1 that cross-reacts with murine β1; Mendrick and Kelly, 1993), MEL-14 (rat anti-mouse l-selectin), and isotype controls (hamster IgM, rat IgG2b, and rat IgG2a). HP2/1 (mouse anti-human α4) was obtained from Beckman Coulter, Inc. (Fullerton, CA). PS/2 (rat anti-mouse α4) (Miyake et al., 1991) and a matched isotype control (rat anti-human Ras Ab) were supplied by LigoCyte (Bozeman, MT). The following cell lines were used: RPMI-8866 cells (human B cell line) obtained from J. Wilkins (University of Manitoba, Winnipeg, MB, Canada), TK-1 cells (murine T cell line) obtained from I. Weissman (Stanford University, Stanford, CA) (Holzmann and Weissman, 1989), and Jurkat (human T cell line) and RBL-2H3 cells (rat mucosal-type mast cell line) from American Type Culture Collection (Manassas, VA).
Expression and Purification of Cellular Adhesion Molecule-Immunoglobulin Fusion Proteins
Human MAdCAM-Ig.
Domains 1 and 2 of human MAdCAM-1 (GenBank no. U43628) were amplified by PCR using human small intestinal cDNA (Invitrogen, Carlsbad, CA) as a template and the following primer sequences: 5′-PCR primer, 5′-ATTAGGAATTCGCCACCATGGATTTCGGACTGGCCCTCCTGCTGG-3′; and 3′-PCR primer, 5′-AATTGGGATCCACTTACCTGTGGAGGTCGGGCTGTGCAGGACG- GGGATG-3′. PCR was performed in the presence of 10% DMSO with KlenTaq (CLONTECH, Palo Alto, CA) in a thermocycler (MJ Research, Waltham, MA) by using 40 cycles with the following parameters: 45 s at 94°C, 45 s at 60°C, and 90 s at 72°C. The resulting PCR product of 660 bp was digested with EcoRI and BamHI and ligated into a pIg (R & D Systems, Minneapolis, MN) expression vector. The pIg vector contains the genomic fragment that encodes the hinge region, CH2 and CH3 of human IgG1 (GenBank no. Z17370), and the fragment encoding human MAdCAM-1 (hMAdCAM-1) was ligated proximal to the IgG1 region. The sequence of the resulting hMAdCAM-1 fragment fused to human IgG1 was verified using Sequenase (U.S. Biochemical Corp., Cleveland, OH). The fragment encoding the entire MAdCAM-Ig fusion was subsequently excised from the pIg vector with EcoRI and NotI and ligated to pcDNA3.1/neo (Invitrogen). The resulting vector, pcDNA3.1/neo-MAdCAM-Ig, was transfected into CHOKI cells (CCL61; American Type Culture Collection) by electroporation and grown under selection with 0.7 mg/ml G418 (Invitrogen). Culture supernatants from single cell clones were assayed by Ig enzyme-linked immunosorbent assay, and a high expressing clone (1 μg/ml) was adapted to CHO-SFM II serum-free media (Invitrogen) for large-scale expression. hMAdCAM-Ig was purified from crude culture supernatants by affinity chromatography on protein A/G Sepharose and dialyzed into 50 mM sodium phosphate buffer, pH 7.6.
Murine MAdCAM-Ig.
The entire extracellular domain of murine MAdCAM-1 (domain 1, domain 2, mucin, and domain 3) (GenBank no. L21203) was amplified by PCR using murine small intestinal cDNA (Invitrogen) as a template and the following primer sequences: 5′-PCR primer, 5′-CCGAGATATCGCCACCATGGAATCCATCCTGGCCCTCCTG-3′; and 3′-PCR primer, 5′-CCTTGGATCCACTTACCTGTGGTGGAGGAGGAATTCGGGGTCA-3′. PCR was performed in the presence of 10% DMSO with KlenTaq (CLONTECH) in a thermocycler (MJ Research) by using 30 cycles with the following parameters: 1 min at 94°C, 1 min at 60°C, and 2 min at 72°C. The resulting PCR product of 1095 bp was digested with EcoRV and BamHI and ligated into a pIg (R & D Systems) expression vector. After verification of the sequence, the fragment encoding the entire MAdCAM-Ig fusion was excised from the pIg vector with EcoRI and NotI and ligated into pCMV-I/IRES/GFP puro vector (Cell & Molecular Technologies, Inc., Phillipsburg, NJ). The resulting vector carrying the gene for GFP was used to transfect CHOKI cells that were grown in the presence of 4 μg/ml puromycin. Transfected cells were sorted by flow cytometry on the basis of their high GFP expression, and the GFP expression was followed during subsequent culture, before the isolation of single cell clones. Culture supernatants from single cell clones were assayed by Ig enzyme-linked immunosorbent assay and immunoblot using anti-murine MAdCAM-1 mAb (MECA-367; BD PharMingen). A high expressing clone (1 μg/ml) was adapted to CHO-SFM II serum-free media (Invitrogen) for large-scale expression, and mMAdCAM-Ig was purified as described for hMAdCAM-Ig.
Human VCAM-Ig.
Domains 1 and 2 of human VCAM-1 (GenBank no.M30257) were amplified by PCR using the human VCAM-1 cDNA (R & D Systems) as a template and the following primer sequences: 3′-PCR primer, 5′-AATTATAATTTGATCAACTTACCTGTCAATTCTTTTACAGCCTGCC-3′; and 5′-PCR primer, 5′-ATAGGAATTCCAGCTGCCACCATGCCTGGGAAGATGGTCG-3′. PCR was performed for 30 cycles using the following parameters: 1 min at 94°C, 2 min at 55°C, and 2 min at 72°C using a DNA thermal cycler (model 480; PerkinElmer Instruments, Norwalk, CT). The resulting PCR product of 650 bp was digested with EcoRI andBclI and ligated into a pIg (R & D Systems) expression vector. After verification of the sequence, the fragment encoding the entire VCAM-Ig fusion was excised from the pIg vector withEcoRI and NotI and ligated to pCI-neo (Promega, Madison, WI). The resulting vector, pCI-neo/VCAM-Ig, was transfected into CHOKI cells by calcium phosphate DNA precipitation and grown under selection with 0.4 mg/ml G418. Culture supernatants from single cell clones were assayed for the ability to support Jurkat cell adhesion, and a high expressing clone (1 μg/ml) was adapted to CHO-SFM II serum-free media. VCAM-Ig was purified as described for hMAdCAM-Ig.
Ligand Binding Assays for α4β7 and α4β1
A ligand binding assay for α4β7 was performed by incubating RPMI-8866 cells (7.5 × 105cells/well) or TK-1 cells (1 × 106cells/well) with <200 pM iodinated human or murine MAdCAM-Ig in a 96-well filter binding format. A ligand binding assay for α4β1 has been described previously (Hagmann et al., 2001) and was performed by incubating Jurkat cells (5 × 105 cells/well) or RBL-2H3 cells (4 × 104 cells/well) with <100 pM iodinated VCAM-Ig in a 96-well filter binding format. Purified VCAM-Ig and MAdCAM-Ig were labeled with 125I using Bolton Hunter reagent and purified using high-performance liquid chromatography gel filtration chromatography. Specific radioactivities were in excess of 1100 Ci/mmol. Compounds were evaluated by incubating radioligand compound (prepared in DMSO; <1% DMSO final concentration), cells, and binding buffer (25 mM HEPES, 150 mM NaCl, 3 mM KCl, 2 mM glucose, and 0.1% bovine serum albumin, with 1 mM MnCl2, pH 7.4) at 25°C for 30 min (α4β1 assays) or 45 min (α4β7 assays) in a 96-well multiscreen MHVBN filtration plate (Millipore, Bedford, MA). After filtration and a single wash with binding buffer, the filtration plates were dried and transferred to adaptor plates (Packard BioScience, Meriden, CT). After adding 100 μl of Microscint-20 (Packard Bioscience) to each well, the plates were sealed, placed on a shaker for 1 min, and counted on a Packard BioScience TopCount. Wells containing cells + radioligand + 1 μM compound or DMSO alone served as controls to calculate 100 and 0% inhibition, respectively.
Quantitative FACS Analysis
A total of 106 RPMI-8866 or Jurkat cells were incubated for 30 min on ice in FACS buffer (phosphate-buffered saline with Ca2+/Mg2+, 5% fetal bovine serum, 100 μg/ml goat IgG, and 0.05% sodium azide) containing saturating levels of the following polyethylene-conjugated antibodies: FIB504 rat anti-mouse β7 (2.4 μg/ml; cross-reacts with human β7), MAR4 mouse anti-human β1 (80 μg/ml), 9F10 mouse anti-human α4 (10 μg/ml), and mIgG1 isotype and rIgG2a isotype controls. Similarly, a total of 106 TK-1 or murine mesenteric lymph node (MLN) lymphocytes were incubated for 30 min on ice in FACS buffer with Fc block (10 μg/ml; BD PharMingen) containing DATK32-PE rat anti-mouse α4β7 (10 μg/ml) or rIgG2a-PE isotype control antibodies. All polyethylene-conjugated antibodies were obtained from BD PharMingen. Cells were washed in FACS buffer and resuspended in FACS buffer containing 1 μg/ml propidium iodide. Cells were analyzed by FACScan flow cytometer (BD Biosciences, Franklin Lakes, NJ). Standardized quantum R-PE microbeads (Flow Cytometry Standards Corp., Fishers, IN) were analyzed by flow cytometry and used to create a calibration curve that relates mean fluorescence intensities to molecules of equivalent soluble fluorescence for use in calculating receptor density values.
In Vitro Shear Flow Adhesion Assay
An in vitro shear flow system adapted from published methods (Berlin et al., 1993) was used to evaluate the ability of compounds to block the binding of human RPMI-8866 cells or murine MLN lymphocytes (2 × 106 cells/ml) isolated from BALB/c mice to capillary tubes coated with murine or human MAdCAM-Ig, respectively. Cells at a density of 6 × 106cells/ml were preincubated with test compounds (0.06–6 μM; final DMSO <1%), neutralizing mAb, or isotype control mAb at 0.6 μM in buffer (HBSS without Ca2+/Mg2+, 20 mM HEPES, 2 mM Mn2+, and 2% human serum, pH 7.0) for 10 min at 37°C. A 500-μl cell suspension was then injected into a closed loop flow system that contained 2.5 ml of assay buffer (HBSS with Ca2+/Mg2+, 20 mM HEPES, and 2% human serum, pH 7.0). Murine assays included preincubation of cells with anti-l-selectin at 0.6 μM to blockl-selectin-dependent rolling due to interactions with the mucin domain of mMAdCAM-Ig. A silicone tubing loop and roller pump were used to circulate cells through a MAdCAM-Ig-coated capillary tube mounted on an inverted microscope stage. MAdCAM-Ig was titrated (5, 10, and 50 μg/ml), and 50 μg/ml was selected for use because it supported 50 to 150 interacting cells/field after 15 min of continuous shear flow. The shear rate was 2 dynes/cm2 for the first 5 min and 1.2 dynes/cm2 for the last 10 min, simulating physiological shear flow, where noninteracting lymphocytes have a midline velocity of 4000 μm/s in murine HEVs (Bargatze et al., 1995). Cells were monitored for 15 min by videomicroscopy, and the number of adherent cells was determined at 1-min intervals by analysis of individual frames using either a customized Proteo-Flow computer analysis package (LigoCyte Pharmaceuticals) or manual counting directly from the monitor screen. Control adhesion for individual experiments was based on the isotype or 1% DMSO control treatment. Data are represented as area under the curve (AUC) from time 0 to 15 min. Data were normalized by calculating percentage of control adhesion within each experiment, with two to six tests evaluated per treatment.
Interaction of Lymphocytes with High Endothelial Venules in Murine Peyer's Patches
Compounds were evaluated for their ability to block the interactions of lymphocytes with high endothelial venules in murine Peyer's patches using an intravital microscopy system adapted from published methods (Bargatze et al., 1995). In brief, murine MLN lymphocytes were isolated from normal donor BALB/c mice and fluorescently labeled with rhodamine (MitoTracker Orange CM-H2-TMRos) or fluorescein (carboxy-4′,5′-dimethylfluorescein diacetate) purchased from Molecular Probes (Eugene, OR). After labeling, cells were incubated in EDTA buffer (HBSS without Ca2+/Mg2+, 10 mM HEPES, and 2 mM EDTA) for 10 min at 25°C, washed in buffer (HBSS without Ca2+/Mg2+ and 10 mM HEPES), resuspended in activating buffer (HBSS without Ca2+/Mg2+, 10 mM HEPES, and 2 mM Ca2+/Mn2+), and treated with anti-l-selectin Ab at 200 μg/ml for 5 min for rhodamine-labeled cells to maintain both a saturating in vitro and in vivo concentration of mAb and at 50 μg/ml for 5 min before i.v. injection for fluorescein-labeled cells to provide a saturating in vitro concentration. Syngeneic recipient BALB/c mice were anesthetized and prepared for intravital microscopy by abdominal incision to exteriorize the small intestines, and a selected Peyer's patch was gel mounted and positioned for epifluorescence videomicroscopy. The protocol used to evaluate the effect of mAb or compounds on the binding of murine lymphocytes to Peyer's patch HEVs by intravital microscopy is described in SchemeFS1.
At time −10 min, rhodamine-labeled cells were injected by i.v. bolus into the tail vein and monitored for 10 min to document control adhesion events in each animal. The same high endothelial field was monitored during the entire time course. Compounds were prepared in HBSS containing ≤1% DMSO and 2% polyethylene glycol-400. Compounds (0.01–10 mpk) or mAbs (10 mpk) were then dosed by i.v. injection 5 min before the i.v. injection of 1.5 × 107fluorescein-labeled MLN lymphocytes. In some experiments, fluorescein-labeled cells were also incubated with compounds (100 μM) for 10 min before injection. After monitoring the adhesion of fluorescein-labeled cells for 10 min, anti-α4(PS/2; 10 mpk) was injected i.v., and cell adhesion was monitored for another 10 min to quantitate α4-independent baseline adhesion as a negative control. A single HEV field containing multiple venules (3–5) was evaluated for each animal. Interacting cells were determined at 1-min intervals by analysis of individual frames using either a customized computer analysis package (LigoCyte Pharmaceuticals) or manual counting directly from the monitor screen to obtain statistically significant values when evaluating four to six mice per treatment group.
The ability of mAbs or compounds to reverse established lymphocyte interactions with murine Peyer's patch HEVs was evaluated as described in Scheme FS2. For experiments involving only fluorescein-labeled cells, these cells were preincubated with anti-l-selectin mAb at 200 μg/ml for 5 min before i.v. injection to maintain both a saturating in vitro and in vivo concentration of mAb. At time −14 to −4 min, 1.5 × 107 fluorescein-labeled MLN lymphocytes were injected by i.v. bolus into the tail vein and monitored for 10 min to document control adhesion events in each animal. Compounds (10 mpk) or mAbs (10 mpk) were then dosed by i.v. injection, and adhesion events were monitored for another 10 min. For compound-treated animals, anti-α4 (PS/2; 10 mpk) was dosed by i.v. injection 10 min after compound treatment, and cell interactions were monitored for another 10 min. Interacting cells were quantitated as described above.
Synthesis of Compounds
Compound 1,N-{N-[(3,5-dichlorobenzene)sulfonyl]-2-(R)-methylpropyl}-(d)-phenylalanine, and compound 2,N-{N-[(3-chlorobenzene)sulfonyl] azetidine-2-(S)-carboxyl}-(l)-4-(2′,6′-bis-methoxyphenyl)phenylalanine, were synthesized as described previously (Durette et al., 1998a,b;Hagmann et al., 2001; Kopka et al., 2002). TR14035,N-(2,6-dichlorobenzoyl)-(l)-4-(2′,6′-bis-methoxyphenyl)phenylalanine, was synthesized as described in Sircar et al. (1999a). The structures of the compounds are shown in Fig. 1.
Pharmacokinetic Analysis of Compounds
Female BALB/c mice (20–23 g; n = 4–6/group) were dosed intravenously with compounds prepared in HBSS containing ≤1% DMSO and 2% polyethylene glycol-400. At the end of each experiment, terminal blood samples were taken for the determination of compound concentrations in plasma. Blood was collected into EDTA after euthanasia at either 15 or 20 min postinjection, and plasma was prepared. Plasma samples were immediately acidified by combining 200 μl of plasma with 50 μl of 0.5 M formate buffer (21.5 ml of 88% formic acid/liter of distilled H2O; pH 3.0 adjusted with NaOH), snap frozen on dry ice, and stored at −20°C. Compounds were purified from plasma samples by solid phase extraction (OASIS cartridge; Waters, Milford, MA), and concentrations of compounds were determined by liquid chromatography-tandem mass spectrometry (API 3000; PerkinElmer Sciex, Foster City, CA). Plasma values for compounds are represented as mean ± S.E.M. in nanomolar concentration.
To determine the disappearance of compound in the blood of rodents, male Sprague-Dawley rats (150–200 g; n = 4–6/group) were dosed intravenously with compounds at 1 mpk as described above, and the plasma concentration of compound was calculated at various time points to determine the time when 50% of the compound had disappeared from the blood (t1/2 in min).
Results
Antagonists Block Both α4β1 and α4β7 Integrins on Human and Rodent Cells.
To determine whether a potent antagonist of α4β1 (Fig. 1; compound 1) (Hagmann et al., 2001) was also capable of blocking α4β7, an assay was developed to measure the binding of MAdCAM-1 to α4β7. RPMI-8866, a human B cell line, was demonstrated by flow cytometry to express high levels of α4β7 (60,000 α4β7 receptors/cell) but low levels of α4β1(4,000 α4β1receptors/cell), and was therefore chosen to evaluate the binding of α4β7 to soluble125I-hMAdCAM-Ig in the presence of the activating divalent cation Mn2+. The concentration of125I-hMAdCAM-Ig used for the binding assay was maintained at <200 pM, based on an IC50 of 1200 pM for competition by unlabeled MAdCAM-Ig. Anti-α4 and anti-β7mAbs blocked 125I-MAdCAM-Ig binding with IC50 values of 110 and 40 ng/ml, respectively, whereas anti-β1 did not block at concentrations up to 3.3 μg/ml, demonstrating the specificity of the interaction for α4β7 (data not shown).
Three compounds were evaluated for their ability to block125I-hMAdCAM-Ig binding to RPMI-8866 cells (Table1). Compound 1, a recently described α4β1 integrin antagonist (Hagmann et al., 2001), blocked binding with an IC50 of 2.93 nM, demonstrating that the compound is a potent antagonist of both α4β1 and α4β7. A structurally related analog, compound 2 (Kopka et al., 2002; Fig. 1), which has an IC50 of 28.5 μM for α4β1 binding to soluble VCAM-1 (Table 1), did not block binding when tested at concentrations up to 100 μM, demonstrating the importance of specific structural features to the activity of compound 1 (Lin et al., 2002). TR14035 (Fig. 1) has been reported to potently inhibit both α4β1 and α4β7 (Martin et al., 1999; Sircar et al., 1999a,b). Consistent with the activity of TR14035 against α4β7, it blocked 125I-hMAdCAM-Ig binding to RPMI-8866 cells and was more potent (IC50 of 0.75 nM) than compound 1.
To compare their potency for α4β1 under similar conditions, the same three compounds were also evaluated for their ability to block 125I-hVCAM-Ig binding to Jurkat cells, which were demonstrated by flow cytometry to express high levels of α4β1 (90,000 α4β1 receptors/cell) but low levels of α4β7(7,000 α4β7receptors/cell). The concentration of125I-hVCAM-Ig used for the binding assay was maintained at <100 pM, based on an IC50 of 350 pM for competition by unlabeled VCAM-Ig. Anti-α4 and anti-β1mAbs blocked 125I-VCAM-Ig binding to Jurkat cells with IC50 values of 80 and 400 ng/ml, respectively, whereas anti-β7 did not block binding at concentrations up to 3.3 μg/ml, demonstrating the specificity of the interaction for α4β1 (data not shown). The activities of compound 1 and TR14035 for α4β1 were confirmed by IC50 values of 0.08 and 0.11 nM, respectively, whereas compound 2 was inactive with an IC50 of 28.5 μM (Table 1).
In preparation for using the compounds in in vivo experiments, the ability of compound 1 and TR14035 to inhibit ligand binding to rodent α4β7 and α4β1 was also tested (Table 1). TK-1, a murine T cell lymphoma line, was reported previously to express α4β7, but not α4β1, and to have enhanced binding to MAdCAM-1 in the presence of Mn2+ (Holzmann and Weissman, 1989; Berlin et al., 1993). Expression of α4β7, but not α4β1, on TK-1 cells was confirmed by FACS (data not shown), and this cell line was therefore chosen to evaluate the binding of α4β7 to soluble125I-mMAdCAM-Ig. Previously published reports indicated that rat RBL-1 mucosal-type mast cells express high levels of α4 that bind both VCAM-1 and MAdCAM-1 (Palecanda et al., 1997). Expression of α4β1 on rat RBL-2H3 cells was confirmed by FACS (data not shown), and these cells were used to evaluate the cross-species binding of rodent α4β1 to soluble125I-hVCAM-Ig, because murine VCAM-Ig was unavailable. Anti-α4 blocked125I-hVCAM-Ig binding to RBL cells, but it was not possible to determine whether α4β1 or α4β7 mediated binding, because anti-rat β1 and anti-rat β7 mAbs were not available. Compound 1 and TR14035 exhibited potencies for inhibition of rodent α4β7 and α4β1 similar to those demonstrated for inhibition of the human receptors (Table 1).
TR14035 Is More Potent than compound 1 in Its Ability to Block α4β7 Adhesion to MAdCAM-Ig under in Vitro Shear Flow Conditions.
Because α4integrins contribute to lymphocyte attachment and rolling under physiological flow, compounds that blocked binding of soluble ligands were further tested for their ability to block the binding of human α4β7-expressing cells to hMAdCAM-Ig-coated capillary tubes under in vitro shear flow (Fig.2, A and B). The assay was modified from systems described previously for α4β7-dependent adhesion under flow (Berlin et al., 1993) by using RPMI-8866 cells and the MAdCAM (domain 1 and domain 2)-Ig fusion protein. Cells were preincubated with neutralizing mAb or compounds in the presence of 2 mM Mn2+ for 10 min at 37°C before injection into the capillary tube adhesion chamber, and the final concentration of Mn2+ in the flow chamber was 0.33 mM. Anti-β7 mAb completely blocked adhesion at concentrations between 1 nM (data not shown) and 0.1 μM (20 μg/ml, 97% inhibition) (Fig. 2B). Anti-α4 mAb was similarly potent, inhibiting adhesion by 99% at 0.1 μM (Fig. 2, A and B). Anti-β1 mAb did not significantly block adhesion (33% inhibition) at concentrations up to 1 nM (200 ng/ml) (data not shown), but did inhibit by 86% at 0.1 μM (20 μg/ml) (Fig. 2B). Although there is no apparent explanation for the blockade of adhesion at the higher concentration of anti-β1, the much higher potency of anti-β7 indicates that human RPMI-8866 cells bind human MAdCAM-Ig-coated capillary tubes in an α4β7-dominant manner under shear flow. TR14035 at 1 μM blocked adhesion of RPMI-8866 cells to MAdCAM-Ig by 100% (Fig. 2, A and B), with an approximate IC50 of 0.01 μM (Fig. 2B). In contrast, compound 1 at 1 μM did not significantly block adhesion (Fig. 2, A and B), indicating that TR14035 was >100-fold more potent than compound 1 in conditions of shear flow.
Because the compounds were to be tested in vivo in mice, their ability to block the binding of murine MLN lymphocytes to mMAdCAM-Ig under in vitro shear flow was also determined. Anti-l-selectin mAb was included to block rolling due to the interaction ofl-selectin with the mucin domain of mMAdCAM-Ig. In the presence of 0.33 mM Mn2+, an average of 90 to 200 lymphocytes bound to mMAdCAM-Ig after 15 min (Fig.3A). Because murine MLN lymphocytes express both α4β1 and α4β7, as demonstrated by flow cytometry (data not shown), antibody blockade was used to test the specificity of the interaction for α4β7. Anti-α4 and anti-α4β7 at 0.1 μM (20 μg/ml) blocked adhesion by 79 and 78%, respectively (Fig. 3, A and B). In contrast, anti-β1 at 0.1 μM did not block adhesion (119% of control) (Fig. 3B), indicating that α4β1 did not contribute to the adhesion and that the interactions were specific for α4β7. TR14035 at 1 μM blocked adhesion by 78% (Fig. 3B), with an approximate IC50 of 0.1 μM. Compound 1 was about 10-fold less potent, inhibiting adhesion with an approximate IC50 of 1 μM (Fig. 3B).
Antagonists of α4β1/α4β7 Block α4β7-Dependent Lymphocyte Interactions with Murine Peyer's Patch High Endothelial Venules.
To quantitate α4β7 adhesion under shear flow conditions in vivo, we used in situ videomicroscopic analysis to measure the adhesion of fluorescently labeled murine MLN lymphocytes to Peyer's patch HEVs expressing murine MAdCAM-1. It has been previously reported that anti-α4, anti-β7, and anti-MAdCAM-1 mAbs block adhesion in this model when administered up to 10 min after the injection of the labeled cells (Bargatze et al., 1995). Because other molecular mechanisms than the α4β7-MAdCAM interaction may contribute to adhesion events at longer time intervals, the evaluation of adhesion was limited to a 10-min interval after injecting rhodamine- or fluorescein-labeled cells. As described underMaterials and Methods, results for each mouse (Table2) were based on a positive control segment (binding of rhodamine-labeled cells that were pretreated with Mn2+ and anti-l-selectin before i.v. injection), an experimental segment (binding of fluorescein-labeled cells that were pretreated with Mn2+ and anti-l-selectin before i.v. injection into mice that had been predosed with mAb or compound 5 min before cell injection), and a negative control segment (binding of fluorescein-labeled cells after the i.v. injection of anti-α4 mAb to quantitate α4-independent baseline adhesion). We confirmed that the interaction between MLN lymphocytes and Peyer's patch HEVs was dependent on α4β7, because anti-α4 mAb at 10 mpk blocked adhesion by 95%, and anti-β1 mAb at 10 mpk failed to block adhesion (Table 2).
Compounds that were able to effectively block murine α4β7-dependent adhesion under shear flow conditions in vitro were tested for their ability to block the adhesion of murine MLN lymphocytes to Peyer's patch HEVs under shear flow in vivo (Table 2). TR14035 was dosed at 10 mpk i.v., and 100 μM TR14035 was either added or not to the MLN lymphocytes before injection. The compound blocked binding to Peyer's patch HEVs by 95% when the cells were preincubated with compound and by 88% in the absence of the preincubation step (Table 2). The approximate IC50 for TR14035 was between 0.01 and 0.1 mpk, with a mean plasma value of <1 nM (Table 2). The higher potency of TR14035 in vivo correlated well with the greater in vitro potency of the compound in the murine α4β7 shear flow assay (approximate IC50 of 0.1 μM; Fig. 3B). Compound 1 delivered at 10-mpk blocked binding to Peyer's patch HEVs by 47%, with a mean plasma value of 433 nM. However, at 1 mpk compound 1 did not significantly inhibit adhesion (Table 2). The lower potency of compound 1 in vivo (Table 2) also correlated with its relative potency in the murine shear flow assay (approximate IC50of 1 μM; Fig. 3B). Compound 2, at 10 mpk i.v., did not significantly block adhesion (Table 2), consistent with the inactivity of compound 2 when tested at concentrations up to 100 μM in in vitro assays (Table1). Thus, TR14035 had a 10–100 fold greater potency than compound 1, when the ability of the compounds to block the binding of murine MLN lymphocytes to murine Peyer's patch HEVs in vivo was evaluated. Furthermore, in vivo efficacy of novel compounds was predicted by their in vitro potency in human and murine α4β7 shear flow adhesion assays.
Anti-α4 mAb and TR14035 Can Reverse Established α4β7-Dependent Lymphocyte Interactions with Murine Peyer's Patch High Endothelial Venules.
To determine whether established adhesion events can be reversed in vivo, we administered anti-α4 mAb or TR14035 between 10 and 14 min after the injection of fluorescein-labeled cells and quantitated adhesion for an additional 10 min (Fig.4, A and B). By 5 to 10 min after the cells were injected, adhesion reached a plateau. New adhesive interactions presumably did not occur after this time, because Mg2+ in plasma gradually exchanged with Mn2+ on α4β7. Cation exchange inactivates α4β7 and leads to a mixed population of activated (high-affinity) and unactivated (low-affinity) forms of the receptor (Bargatze et al., 1995). Within 5 min after anti-α4 mAb was injected, the number of adhesion events declined to 46% of control values, and further declined to 19% of control values by 10 min (Fig.4A; Table 3) Within 5 min after administration of TR14035 (10 mpk i.v.), the number of adhesion events declined to 48%, but did not decline further (Fig. 4B; Table 3). In addition, α4-dependent rolling was completely blocked after administering anti-α4 mAb, but not after administering TR14035 (data not shown).
To determine whether earlier administration of TR14035 would be more effective in reversing established adhesion events, we administered anti-α4 mAb or TR14035 after fluorescein-labeled cells were allowed to adhere for 4 min instead of 10 to 14 min (Fig. 4, C and D). Within 10 min after administering anti-α4 mAb or TR14035 at 10 mpk i.v., the number of adhesion events observed declined to 10 and 39% of control values, respectively (Fig. 4, C and D; Table 3). In the absence of anti-α4 mAb or after injecting isotype control mAb, previously established adhesion events were stable and did not decay for at least 20 to 30 min (data not shown). To demonstrate that the residual adhesion in the TR14035 treatment group was α4-dependent, we administered a final injection of anti-α4 mAb at 10 mpk i.v., which blocked adhesion by 88% (12% of control) within 10 min (Fig. 4D; Table 3). Thus, although TR14035 was able to cause the detachment of adherent lymphocytes, on a molar basis anti-α4 mAb (0.74 μM at 10 mpk) was more effective than TR14035 (234 μM at 10 mpk) in detaching adherent murine lymphocytes.
Discussion
Our studies confirm previous reports that TR14035 is a potent antagonist of both α4β1and α4β7(IC50 values ranging from 2 to 46 nM) (Martin et al., 1999; Sircar et al., 1999a,b), and further demonstrate that compound 1 is a potent antagonist of both integrins. A study comparing the ability of TR14035 to block the adhesion of α4β7 or α4β1 to CS-1 immobilized on a plate concluded that the compound was 9 times more potent against α4β7than α4β1(IC50 of 5 versus 46 nM for α4β1) (Martin et al., 1999). In contrast, we found that TR14035 was 7 times more potent in blocking α4β1 binding to soluble VCAM-1 (IC50 of 0.11 nM) than in blocking α4β7 binding to soluble MAdCAM-1 (IC50 of 0.75 nM). Differences in the ligands used to evaluate the potency of TR14035 against α4β1 and α4β7 may contribute to the discrepancies in the IC50 values. To compare the activity of TR14035 against α4β1 and α4β7 binding to the same ligand, we tested the ability of the compound to block the adhesion of α4β7- or α4β1-expressing cells to CS-1 immobilized on a plate and found similar potencies against both α4β7(IC50 of 16 nM) and α4β1(IC50 of 12 nM) (L. A. Egger and U. Kidambi, unpublished observations). Taken together, the results clearly indicate that TR14035 is active against both α4β1 and α4β7, with potencies for each dependent on the ligand used and the format of the assay.
The ability of the compounds to inhibit binding of human and rodent α4β7 and α4β1 to soluble ligands in suspension was measured to test for possible species-dependent shifts in potency. Species differences in ligand specificity have been reported for other integrins. For example, murine lymphocyte function antigen-1 (LFA-1; αLβ2) binds human ICAM-2, but not ICAM-1, whereas human LFA-1 binds not only human ICAM-1 and ICAM-2 but also murine ICAM-1 (Johnston et al., 1990). Species specificity for the binding of LFA-1 to ICAM has been mapped to the αL subunit of LFA-1 (Johnston et al., 1990), suggesting that structural differences exist within the ligand binding sites of the same integrins from different species. Previous studies have shown that α4β1 and α4β7 can bind ligands across species (Hession et al., 1992; Shyjan et al., 1996), consistent with the primary structural homology for human and rodent α4, β1, and β7 (76, 92, and 87%, respectively) (Takada et al., 1989; Jiang et al., 1992; Cervella et al., 1993). A high degree of structural homology also extends to the ligands for α4 integrins: human and rodent VCAM-1 have 85% sequence identity, and the α4 integrin binding motif IDS is completely conserved (Briskin et al., 1996). Although the overall homology between full-length human and murine MAdCAM-1 is only 39%, the first two N-terminal Ig-like domains are 57% identical, and the α4 integrin binding motif LDT in domain 1 of MAdCAM-1 is completely conserved (Shyjan et al., 1996). In this report, compounds blocked the binding of human or rodent α4β7 and α4β1 to soluble ligands in suspension with similar potencies, suggesting that the affinity of the receptors for compounds was similar for human or rodent α4.
Differences in the activities of TR14035 and compound 1 against α4β7 on human and rodent cells emerged in in vitro shear flow assays. TR14035 was approximately 10-fold more potent in the assay with human cells than with mouse cells, whereas the reverse was true for compound 1. Variations in receptor density on human and mouse cells may contribute to the species differences in the potency of the compounds. As measured by quantitative FACS analysis, the receptor density of α4β7 on murine lymphocytes was 1000 receptors/cell, 60-fold lower than the receptor density of α4β7 on RPMI-8866 cells (60,000 receptors/cell). Despite the lower density of α4β7 on murine lymphocytes, a robust interaction (90–200 interacting cells after 15 min) between Mn2+-activated lymphocytes and mMAdCAM-Ig was observed (Fig. 3A), similar to that observed with human cells (Fig. 2A). This may be due in part to the use of mMAdCAM(D1D2MD3)-Ig instead of hMAdCAM(D1D2)-Ig, and the mucin domain in the mMAdCAM-Ig construct may contribute to and enhance the affinity of receptor binding under shear flow. Despite the shifts in absolute potency, TR14035 emerged as the more potent of the two compounds with both human and mouse cells under flow conditions in vitro.
The in vivo activity and effective blood concentrations of antagonists of α4β7 were predicted by their in vitro potencies in adhesion assays under shear flow. TR14035 was >100-fold more potent than compound 1 in the human α4β7 shear flow adhesion assay, 10-fold more potent than compound 1 in the murine α4β7 shear flow adhesion assay, and 10–100-fold more potent than compound 1 in in vivo blockade of murine lymphocyte binding to Peyer's patch HEVs (Table 2). The plasma concentration for TR14035 at the observed IC50 for half-maximal inhibition in vivo was >400-fold lower than that for compound 1, correlating well with the higher potency of TR14035 in the in vitro shear flow assays. The in vivo activity of TR14035 is also supported by previous reports demonstrating that TR14035 is orally active (60% bioavailability) in rodent arthritis and colitis models (Sircar et al., 1999b).
In this report, we demonstrated that the binding of murine MLN lymphocytes to Peyer's patch HEVs was dependent on α4β7. When MLN lymphocytes were prestimulated with Mn2+ to activate α4β7 binding to ligand, anti-α4 mAb (10 mpk i.v.) blocked adhesion by 95%, and anti-β1 mAb (10 mpk i.v.) did not block adhesion. The anti-β1 mAb used in this study has been shown to block the adhesion of β1 integrins to collagen (Mendrick and Kelly, 1993) and the binding of α4β1 to125I-VCAM-Ig in suspension (L. A. Egger, J. Cao, and V. Kidambi, unpublished data). These results suggest that an α4 integrin other than α4β1 is responsible for adhesion of lymphocytes in Peyer's patch HEVs under physiological flow in vivo. Consistent with this is the observation that two compounds demonstrated to antagonize α4β7 were effective at blocking this interaction. It is therefore likely that α4β7 is the primary α4 integrin involved.
The role of α4β7 in regulating homing of lymphocytes to GALT has been clearly defined in β7-deficient mice. Mice deficient for β7 integrins have a severe impairment in the formation of GALT and a reduction in the number of intraepithelial lymphocytes, which express αEβ7, but lymphocyte development is normal. β7−/−lymphocytes derived from these mice fail to arrest and adhere to HEVs as assessed by epifluorescence videomicroscopy in Peyer's patch (Wagner et al., 1996). The lack of apparent pathology in the β7-deficient mice suggests that agents targeting β7 will specifically alter GALT function.
The ability of anti-α4 mAb and TR14035 to reverse established α4β7-dependent interactions between murine MLN lymphocytes and Peyer's patch HEVs supports the use of these agents as therapeutic treatments for blocking established adhesion events in an inflammatory setting. Anti-α4 mAb and TR14035, administered at 10 mpk i.v., reversed established adhesion events by 90% (10% of control) and 61% (39% of control), respectively (Fig. 4; Table 3). The lower potency of TR14035 compared with anti-α4 mAb may be explained by the rapid off-rate of the compound, because the reported t1/2 for TR14035 is 14 min for dissociation from α4β7-expressing cells in the presence of Mn2+ (Martin et al., 1999). When tested at 1 mpk i.v. in the rat, the disappearance of TR14035 and compound 1 from the blood was shown to have at1/2 of 36 and 54 min, respectively (S. Tong, Z. Wang, and J. Wang unpublished data). It is also possible that differences observed with TR14035 and anti-α4 mAb may be due to the enhanced affinity of using a divalent reagent, such as a whole antibody. In addition, α4β7 can exist in multiple affinity states that bind to MAdCAM-1 to mediate rolling and adhesion (de Chateau et al., 2001), and anti-α4mAb is known to block ligand binding to both the unactivated (low-affinity) and activated (high-affinity) states of α4β7. In contrast, TR14035 has been reported to block ligand binding to the activated state of α4β7 (Table 1;Martin et al., 1999), but information on its ability to block the binding of unactivated α4β7 to MAdCAM-1 has not been reported. Because Mn2+ injected with the MLN lymphocytes will exchange with endogenous Mg2+ in 5 to 7 min (Bargatze et al., 1995), it is likely that both activated and unactivated forms of α4β7 are present in the in vivo assay and may contribute to adhesion. Thus, TR14035 was less effective than anti-α4 mAb in reversing established adhesion events involving both activated and unactivated forms of α4β7 in vivo.
In summary, the data presented herein provide the first evidence that small molecule antagonists of α4β7 can block α4β7-dependent lymphocyte interactions under shear flow in vitro and under shear flow in vivo when interacting with murine Peyer's patch HEVs. Because this interaction can be evaluated within a short time interval, this in vivo model can be used to support the development of preclinical candidates. Finally, the human and murine shear flow adhesion assays described in this report have broad utility and can be used to predict the activities of compounds before in vivo testing.
Acknowledgments
We are grateful to Kristy Scheiring, Nancy Quan, and Ji-Yan Xue who contributed to the development of MAdCAM-Ig binding assays, to Zhen Wang and Junying Wang for formulation and mass spectral analysis, and to Marcie Donnelly and Judy Fenyk-Melody for dosing of animals for pharmacokinetic evaluation. We thank John Wilkins (University of Manitoba, Winnipeg, MB, Canada) for providing the RPMI-8866 cell line, Irving Weissman (Stanford University, Stanford, CA) for providing the TK-1 cell line, and Aiyappa Palecanda (LigoCyte Pharmaceuticals, Bozeman, MT) for scientific discussions.
Footnotes
- Abbreviations:
- HEV
- high endothelial venule
- MAdCAM-1
- mucosal addressin cell adhesion molecule-1
- GALT
- gut-associated lymphoid tissue
- mAb
- monoclonal antibody
- mpk
- milligrams per kilogram
- VCAM-1
- vascular cell adhesion molecule-1
- PCR
- polymerase chain reaction
- DMSO
- dimethyl sulfoxide
- bp
- base pair(s)
- GFP
- green fluorescent protein
- FACS
- fluorescence-activated cell sorting
- MLN
- murine mesenteric lymph node
- HBSS
- Hanks' balanced salt solution
- Ab
- antibody
- AUC
- area under the curve
- LFA-1
- lymphocyte function antigen-1
- ICAM
- intercellular adhesion molecule
- compound 1
- N-{N-[(3,5-dichlorobenzene)sulfonyl]-2-(R)-methylpropyl}-(d)-phenylalanine
- compound 2
- N-{N-[(3-chlorobenzene)sulfonyl] azetidine-2-(S)-carboxyl}-(l)-4-(2′,6′-bis-methoxyphenyl)phenylalanine
- TR14035
- N-(2,6-dichlorobenzoyl)-(l)-4-(2′,6′-bis-methoxyphenyl)phenylalanine
- Received November 28, 2001.
- Accepted March 11, 2002.
- The American Society for Pharmacology and Experimental Therapeutics