A Small Molecule α₄β₁ Antagonist Prevents Development of Murine Lyme Arthritis without Affecting Protective Immunity¹

Lyme disease caused by three genospecies of the spirochete Borrelia burgdorferi sensu lato is the most common tick-borne disease in humans in the Northern Hemisphere. Infection with B. burgdorferi provokes a multiorgan inflammatory ailment, but the precise underlying mechanisms are not well understood (1). The disease is characterized by some or all of the following clinical manifestations: an initial erythematous rash, lymphocytoma cutis, acrodermatitis chronica atrophicans, neurological disorders, carditis, and, especially, arthritis. Intermittent or chronic oligoarticular arthritis primarily affecting large joints such as the knees is a debilitating late complication. Although most patients with acute Lyme arthritis can be treated effectively with antibiotics, ∼10% of patients have persistent knee swelling for months to years after >2 mo of medication. This condition has been termed antibiotic treatment-resistant Lyme arthritis (2).

In a murine model of Lyme borreliosis, used to delineate mechanisms of inflammation and protective immunity, genetically susceptible mouse strains like C3H/He develop severe carditis and arthritis when infected with B. burgdorferi. The disease pattern arising in joints of mice resembles human Lyme arthritis, with edema, hyperproliferation of synovial cells, as well as massive infiltration of neutrophils and monocytes, but markedly few lymphocytes into tissue and joint space (3). The severity of inflammation peaks at 2–3 wk after infection and regresses spontaneously in the presence of B. burgdorferi-specific Abs.

Leukocytes are recruited from circulating blood to the inflamed tissue by sequential adhesive interactions with the endothelium in postcapillary venules. This process involves capture on, rolling along, and firm adhesion to the microvascular endothelium, followed by transmigration through the vessel wall and further migration across basal lamina/extravascular tissue (4). All of these steps, along with a costimulation of cellular activation, are orchestrated by cell adhesion molecules (CAMs)³ expressed on both leukocytes and endothelial cells (EC). CAMs are often divided into four major subsets responsible for the different steps in extravasation, each revealing a partial functional redundancy of its members. The selectins and heavily glycosylated selectin ligands mediate the initial phases (tethering and rolling, i.e., slowing the passage of leukocytes), whereas leukocyte integrins and their endothelial counter-receptors of the Ig superfamily are mainly involved in firm attachment of leukocytes to endothelial surfaces following activation by chemokines or other stimuli (5). Members of the Ig superfamily (PECAM-1, junctional adhesion molecules) are engaged furthermore in diapedesis, when leukocytes responding to a chemoattractant concentration gradient polarize and squeeze through the endothelial junctions (6). Blockade of any of the hierarchically proceeding steps leads to an interruption of leukocyte extravasation.

The expression level and relevance of CAMs during B. burgdorferi infection has been studied in several settings. In patients with antibiotic treatment-resistant Lyme arthritis-enhanced expression of P-selectin, vascular adhesion protein-1, ICAM-1, LFA-1, VCAM-1, and α₄β₁ was detected in affected joints (2). In the murine model of acute Lyme arthritis, B. burgdorferi was shown to up-regulate CAMs such as E-selectin, P-selectin, ICAM-1, and VCAM-1 in synovia (7). However, E- and P-selectin double-deficient mice displayed no significant differences to their wild-type littermates during infection with B. burgdorferi in terms of arthritis development and tissue spirochete levels (8).

In addition, some reports described in vitro the enhanced expression of functional P-selectin, E-selectin, ICAM-1, and VCAM-1 on murine as well as human ECs upon contact with intact B. burgdorferi, spirochetal lysates, and OspA or OspB, respectively (9, 10, 11, 12, 13). Inhibition experiments conducted by Burns and Furie (14) and by Sellati et al. (11) using mAbs to α₄, CD18, and E-selectin demonstrated that monocytes and CD4⁺ lymphocytes mainly engage α₄ and CD11/CD18 integrins for migration across B. burgdorferi-activated endothelium, whereas extravasation of neutrophils involves E-selectin- and CD11/CD18-dependent pathways.

The integrin α₄β₁ (very late Ag-4, VLA-4, CD49d/CD29) is a key cell surface receptor that is expressed constitutively on most leukocytes such as lymphocytes, monocytes/macrophages, eosinophils (15, 16), mast cells (17), basophils (18), and NK/NKT cells (19). On neutrophils, a minor baseline expression is enhanced upon activation/emigration of these cells (20, 21, 22, 23, 24). α₄β₁ is a noncovalent heterodimer composed of an α₄ (155 kDa) and a β₁ (150 kDa) chain. It is known that the subunits, like in most integrins, are conformationally mobile, moving between high-affinity, active and low-affinity, inactive states in response to stimuli (25). α₄β₁ binds to vascular CAM-1 (VCAM-1, CD106), an inducible EC surface protein, to the CS1 region in the alternatively spliced type III connecting segment of the extracellular matrix (ECM) protein fibronectin (26, 27), and to a number of other ligands (28, 29, 30, 31, 32, 33), thereby mediating cellular adhesion and activation through a variety of cell-cell and cell-matrix interactions (16).

Because the association of α₄β₁/VCAM-1 in numerous animal models of chronic inflammatory, autoimmune, and allergic human diseases (e.g., multiple sclerosis (MS), asthma, inflammatory arthritis, atherosclerosis, and inflammatory bowel disease) is well documented (reviewed in Ref.34), the efficacy of blocking α₄β₁ with specific agents has been evaluated in many studies. The use of Abs directed against the α₄ subunit of α₄β₁ for intervention in several animal models was successful (35, 36, 37, 38, 39) and has led to the development of the humanized mAb Natalizumab (approved in November 2004) that was shown to be effective in both MS and inflammatory bowel disease (40, 41).

Natural ligands of α₄β₁ (fibronectin, VCAM-1) and anti-α₄ mAbs have served as the basis for the design of highly selective, peptide-based small molecule α₄β₁ antagonists with low or even subnanomolar IC₅₀ values (42). To date, targeting of α₄β₁ with synthetic antagonists has been examined in mouse models of lung inflammation/asthma (43, 44) and experimental autoimmune encephalomyelitis/MS (45, 46) with predominantly beneficial results. However, to our knowledge, no small molecule antagonists specific to α₄β₁ have been tested so far for treatment in an animal model of an infectious disease.

Therefore, in this study, we attempted to determine the effects of the newly developed α₄β₁ inhibitor S18407 on the onset and resolution of B. burgdorferi-induced Lyme arthritis, the antibacterial immune response, and the elimination of spirochetes in systemically treated C3H/HeJ mice. S18407 is a small molecule α₄β₁ antagonist, which belongs to the class of hydantoin-based inhibitors. The compound represents a prodrug that is physiologically metabolized at a rate of ∼25% to the active carboxylic acid S16197 inherently exhibiting poorer oral bioavailability. S16204 is the sodium salt of S16197 with an improved solubility in water (Fig. 1).

Female C3H/HeJ mice were obtained from Harlan Winkelmann. All of the mice were housed under specific pathogen-free conditions and were 4–5 wk old at the time of infection. Animals destined for treatment were selected randomly and divided into three groups consisting of 6–12 mice each. All of the experiments were conducted according to the institutional guidelines for animal welfare of the Institute for Clinical Microbiology, Immunology, and Hygiene at the University of Erlangen-Nuremberg and approved by the District Government of Middle Franconia.

The following cell lines were obtained: mlEnd.1, murine mesenteric lymph node-derived EC line, created in the same fashion as the line bEnd.3 (47, 48); U-937, human promonocytic leukemia cell line (49) (American Type Culture Collection (ATCC)); JY, EBV-transformed human B lymphoblastoid cell line (50); KL/4, LFA-1-transfected K562 cell line (highly undifferentiated human chronic myelogenous leukemia cell line) (51); MOLT-4, human thymic T lymphoblastoid leukemia cell line (52) (ATCC); and HT-1080, human fibrosarcoma cell line (53) (ATCC).

Cell attachment assays as well as cell-free binding assays were performed to test the inhibitory efficacy of the drug S16197 to the interaction of selected pairs of adhesion molecules in vitro.

Ninety-six-well tissue-culture-treated microtiter plates (round-bottom; Corning/Costar) were coated with purified fusion proteins consisting of the Fc portion of human IgG and the extracellular domain of mucosal addressin CAM-1 (MAdCAM-1) (2 μg/ml) or ICAM-1 (10 μg/ml), or with human vitronectin (2.5 μg/ml) for 60 min at room temperature (RT) (all dilutions in HEPES-buffered saline (HBS) with 20 mM HEPES and 150 mM NaCl; pH 7.4). Wells were aspirated and blocked with heat-denatured BSA (10 mg/ml) for 30 min at RT. Serial dilutions of the compound S16197 were prepared in HBS containing 0.4 mM Mn²⁺. For the KL/4-ICAM-1 assay, an activating anti-β₂ mAb (KIM185) was added at 50 μg/ml to the HBS/Mn²⁺ buffer. JY and KL/4 cells were cultured in RPMI 1640 and resuspended at 1 × 10⁶ cells/ml in DMEM/25 mM HEPES.

Fifty microliters of cells and 50 μl of the diluted compound were added to the wells for 20 min at 37°C with 5% (v/v) CO₂. To determine the reference value for 100% attachment enabling quantification of the percentage of cells specifically bound, cells were added directly to uncoated/unblocked control wells. The control wells were fixed without washing with 20 μl of 50% (w/v) glutaraldehyde. Unbound and loosely attached cells in the rest of wells were aspirated, and specifically bound cells were fixed with 100 μl of 5% (w/v) glutaraldehyde for 30 min. After aspirating the fixative, the wells were washed gently three times with dH₂O, and attached cells were stained for 60 min with 0.1% (w/v) Crystal Violet in 100 mM MES (pH 6.0). Finally, the wells were washed four times with dH₂O before adding 100 μl of 10% (v/v) acetic acid to solubilize dye. Results were assessed by spectrophotometric measurement of absorbance at 570 nm using a plate reader.

Ninety-six-well microtiter plates with round-bottom (MaxiSorp; Nunc) were coated with 10 μg/ml goat anti-human Ig (MP Biomedicals) in 50 mM Tris buffer (pH 9.5) and incubated for 60 min. Plates were washed with PBS and blocked for 30 min with 1% BSA in PBS. After washing with PBS, the huVCAM-1-IgG fusion protein (VCAM-1 (I–VII)-IgG; Biotech Australia) was added at 0.37 μg/ml in blocking buffer and incubated for 1.5 h. Plates were washed again with PBS, and a Fc receptor blocking buffer (1 mg/ml γ-globulin; Sigma-Aldrich) in binding buffer (100 mM NaCl, 100 μM MgCl₂, 100 μM MnCl₂, 1 mg/ml BSA in 50 mM HEPES buffer; pH 7.5) was added for 30 min.

The plates were washed with PBS and preincubated for 20 min with serial dilutions of compound S16197 in binding buffer before adding U-937 cell suspension (1 × 10⁶ cells/ml, preblocked in Fc receptor blocking buffer). After 20 min liquid was poured off gently, and plates were drained in a tank with stop buffer (100 mM NaCl, 100 μM MgCl₂, 100 μM MnCl₂ in 25 mM Tris; pH 7.5). Exposure to stop buffer was repeated once before staining cells with the DNA-specific fluorochrome Hoechst 33258 (Sigma-Aldrich; 16.7 μg/ml in PBS with 4% (w/v) formaldehyde and 0.5% (v/v) Triton X-100) for 15 min. Liquid was poured off, and plates were drained twice in stop buffer as described above. Number of cells was assessed by measurement of fluorescence in a microtiter plate cytofluorometer (Millipore; single measure, filter A/A (360/40/460/40)).

Integrins (α₄β₁, MOLT-4 cell extract; α₂β₁, HT-1080 cell extract; α₅β₁, placental extract) were diluted in PBS without NaHCO₃ and immobilized on 96-well microtiter plates (Immulon 3; Dynex) overnight at RT. Dilutions applied were 1/100 for α₄β₁, 1/200 for α₂β₁, and 1/400 in the case of α₅β₁. Plates were blocked for 3 h at RT with 5% BSA, 0.05% Na-azide in TBS (150 mM NaCl, 25 mM Tris-HCl; pH 7.4). After washing three times with TBS containing 0.1% BSA and 2 mM Mn²⁺, the appropriate counterligands of the integrins were added with or without compound S16197 and incubated for 3 h under the following conditions: VCAM-1-IgG, 0.5 μg/ml at 37°C; biotinylated fibronectin fragments 50 kDa and H/120, each 0.1 μg/ml at RT; and biotinylated collagen I, 1 μg/ml (rough estimate) at RT. Plates were washed three times as before and incubated for 20 min with peroxidase-conjugated ExtrAvidin (Sigma-Aldrich; diluted 1/500 in TBS) or in the case of VCAM-1-IgG with a peroxidase-conjugated anti-Ig-Fc Ab (diluted 1/1000 in TBS). After washing four times as before, the peroxidase substrate ABTS (11 mg of ABTS dissolved in 0.5 ml of dH₂O + 10 ml of 0.1 M NaOAc/0.05 M NaH₂PO₄ (pH 5.0) + 100 μl of 0.3% H₂O₂) was added. Enzymatic reaction was stopped after 15 min with a solution of 2% SDS in TBS, and absorbance was measured spectrophotometrically at 405 nm in a plate reader.

To assess the inhibitory efficacy of S16204 to the adhesion of specific leukocytes to B. burgdorferi-activated ECs, 3 × 10⁴ mlEnd.1 cells were seeded onto Lab-Tek chamber slides (Nunc) and grown to confluency. Cells were activated by coincubation with 1 × 10⁷ spirochetes in 100 μl of DMEM for 50 h at 37°C. Murine thioglycollate-elicited peritoneal cells containing ∼65% neutrophilic granulocytes, or U-937 cells (each 1 × 10⁶ in 100 μl of medium) were preincubated with the drug at a final concentration of 3 nM for 30 min at RT and added to ECs that were washed three times with medium before. Leukocytes were permitted to adhere for 20 min under shear (85 rpm on a horizontal shaker; Bühler) at RT. Cells were then washed cautiously three times in DMEM for removal of nonadherent cells and fixed with 2.5% glutaraldehyde in DMEM for at least 2 h. Measurements were done in quadruplicate. The assays were evaluated by direct counting of four random visual fields.

The N40 isolate of B. burgdorferi was grown at 32°C in BSK-H medium containing 6% rabbit serum (Sigma-Aldrich) (55) and underwent fewer than five in vitro passages before inoculation. Spirochetes were enumerated with a blood cell counting chamber by dark-field microscopy and diluted with sterile medium. Mice were infected by s.c. injection of 5 × 10⁵ bacteria in 50 μl of BSK-H into the right hind footpad.

Because inflammation predominantly emerges unilaterally dependent on inoculation site, only joints derived from the affected side of the body were used for examination in all respective experiments.

Before application, S18407 was ground up in an agate mortar followed by suspending in a solution of 5 mg/ml hydroxyethylcellulose (HEC) in tap water. Groups of 6–12 mice were treated once daily starting on the day of inoculation for a period of 21 days. At this time the extent of ankle swelling in untreated control mice was, as expected, spontaneously decreasing. Doses of 3 mg or 30 mg S18407 per kilogram body weight or the vehicle HEC alone were administered in a total volume of 200 μl per mouse by the oral route using a bulb-headed cannula (1.0 × 55 mm, straight; Acufirm).

The development of joint swelling, grossly correlating with the severity of arthritis, was monitored by measuring the thickness of the infected and, as reference, of the noninfected contralateral tibiotarsal joint by means of a metric caliper (Kroeplin). Values obtained were used to calculate the fold increase in swelling of the infected joint over the contralateral joint. Measurements were taken in the anterior to posterior orientation with extended ankles through the thickest portion.

Rear limbs were excised above and below the tibiotarsal joint and embedded in Tissue-Tek O.C.T. compound into vinyl cryomolds (Sakura Finetek), snap-frozen on liquid nitrogen, and stored at −70°C until analysis by histology. To preserve morphology, ankle joints were dissected in toto without preceding decalcification at −25°C on a HM 500 OM cryostat (Microm) using a tungsten carbide-tipped microtome knife (16 cm, D-profile; Microm). Sagittal sections (10 μm) were mounted on slides previously coated with 50 μg/ml poly-l-lysine in 10 mM Tris-HCl (pH 8.0), fixed with ice-cold acetone for 20 min, and stained with hemalaun (Dr. K. Hollborn & Söhne) and eosin Y (Sigma-Aldrich) for 4 min each. The sections were dehydrated in an ascending sequence of ethanol and twice in Rotihistol (Roth) before embedding in DePeX (Serva). Pictures were taken with a digital video camera (Spot RT Color; Diagnostic Instruments) adapted to an Axiophot microscope (Zeiss) using the MetaVue 4.6 software (Universal Imaging).

DNA was extracted from rear ankle joints of infected mice using the QIAamp DNA Mini kit (Qiagen) according to the manufacturer’s instructions. Elution volume was 200 μl of AE buffer (10 mM Tris-HCl, 0.5 mM EDTA; pH 9.0).

Simultaneous detection and quantification of B. burgdorferi was conducted with the LightCycler PCR system (Roche) using the following primers: forward, 5′-TCTTTTCTCTGGTGAGGGAGCT-3′ and reverse, 5′-TCCTTCCTGTTGAACACCCTCT-3′, which amplify a 70-bp fragment of the Borrelia flagellin B gene (flaB, chromosomal, single copy); and forward, 5′-CCAGCCACAGAATACCATCC-3′ and reverse, 5′-GGACATACTCTGCTGCCATC-3′, which flank a product 154 bp in length corresponding to the murine nidogen-1 gene that was used as reference for normalization. HPLC-purified primers were provided by Metabion. PCRs for flagellin B and nidogen-1 were performed in separate runs. The 20-μl reaction volume contained 1 U Platinum Taq polymerase, 2 μl of 10× reaction buffer, 4 mM MgCl₂ (all obtained from Invitrogen Life Technologies), 0.4 μM each primer, 0.25 mM deoxynucleosid triphosphate mix (Amersham Biosciences), 0.5 × SYBR Green I dye (Roche), 0.5 mg/ml BSA (New England Biolabs), 5% DMSO (Sigma-Aldrich), and 2 μl of extracted DNA in a dilution of 1/10 in H₂O or 2 μl of external standard template. DNA preparations extracted from the reference strain of B. burgdorferi sensu stricto were used to establish a standard curve for flagellin B PCR. Serial dilutions of a pGEM-T Easy plasmid (Promega) harboring a mouse nidogen-1 fragment were included as standards in each normalization PCR for nidogen-1. Applying the LightCycler 5.32 software (Roche), the number of spirochetes was calculated on the basis of the standards and normalized to 10⁵ copies of nidogen-1.

The amplification program consisted of the initial denaturation step at 95°C for 10 min and 50 cycles of denaturation at 95°C for 15 s, annealing at 61°C (flagellin B)/60°C (nidogen-1) for 10 s and extension at 72°C for 15 s. The temperature transition rate was 20°C/s. Fluorescence was measured at the end of each extension step. After each amplification, melting curves were acquired to determine the specificity of PCR products.

Thirty-five days postinfection (p.i.), spleens of mice were removed, and single-cell suspensions were prepared. Cells (4 × 10⁶/ml) were cultured in vitro for 48 h in the presence of B. burgdorferi Ag (10 bacterial equivalents per one spleen cell) or Con A (7.5 μg/ml final; Sigma-Aldrich) in Click-RPMI (Biochrom), supplemented with 2 mM l-glutamine, 10 mM HEPES buffer, 7.5 mM NaHCO₃, 0.05 mM 2-ME, 165 IU/ml penicillin G, 80 μg/ml streptomycin, and 10% selected FCS with a total LPS content <100 pg/ml.

The concentrations of IL-4 and IFN-γ in supernatants of stimulated spleen cells were measured by specific two-site ELISAs, with reference standard curves obtained from known amounts of the respective murine recombinant cytokine. Matched Ab pairs for the detection of cytokines were purchased from BD Pharmingen (anti-mouse IL-4, clone 11B11; anti-mouse IL-4-biotin, clone BVD6-24G2; anti-mouse IFN-γ, clone R4-6A2; anti-mouse IFN-γ-biotin, clone XMG1.2). They were used according to the supplier’s recommendations with a streptavidin/biotin amplification (StreptABComplex/HRP; DakoCytomation) and tetramethylbenzidine (Sigma-Aldrich) as a substrate for HRP.

Results were assessed by spectrophotometric measurement of absorbance at 405 nm wavelength.

For the isolation of single-cell suspensions, rear hind limbs were excised just above and below the tibiotarsal joint. After removal of skin and thorough dissection with scissors, the tissue particles were digested for 1 h at 37°C in 0.1% collagenase D (Roche) in HBSS. The resulting cell suspension was passed through a 70-μm cell strainer (BD Falcon) to remove insoluble material and washed with Click-RPMI containing FCS. Total yield was determined by counting of isolated cells in a hemacytometer chamber with trypan blue exclusion and ignoring erythrocytes. Cells were analyzed by flow cytometry in the presence of 1 μg/ml propidium iodide on a FACSCalibur cytometer using CellQuest Pro software (BD Biosciences). Low fluorescence detritus and propidium iodide-positive cells considered as nonviable were gated out before analysis. Abs used were anti-mouse FcγRIII/II (CD16/CD32) (clone 2.4G2), anti-mouse CD45-allophycocyanin (clone 30-F11), and anti-mouse Gr-1(Ly-6G and Ly-6C)-PE (clone RB6-8C5) (all obtained from BD Pharmingen).

On days 12 and 35 p.i., serum was collected by bleeding the retro-orbital plexus of B. burgdorferi-infected mice at sacrifice and analyzed for B. burgdorferi-specific Ig isotypes using Ab capture ELISAs. Microtiter plates were coated with sonicated B. burgdorferi at a concentration of 5 μg/ml. Serum samples diluted 1/100 (IgG1 and IgG2a) or 1/5 (IgE) in PBS/10% FCS were added to plates and incubated overnight at 4°C. Bound murine Ig was detected by addition of alkaline phosphatase-conjugated Abs to murine IgG1 (clone G1-6.5) or IgG2a (clone R19-15) (both 1/1000 diluted; BD Pharmingen) for 1 h. Total IgE was measured by a standard sandwich ELISA applying anti-mouse IgE (clone R35-72) and anti-mouse IgE-biotin (clone R35-118) (both 2 μg/ml) as matched Ab pair and purified mouse IgE as standard, all obtained from BD Pharmingen. Plates were developed by incubation with 1 mg/ml p-nitrophenyl phosphate (Sigma-Aldrich) followed by measuring of absorbance in a spectrophotometer at 405 nm.

At various time points after infection indicated in the figure legend, the rear ankle joints were excised, snap-frozen in liquid nitrogen, and stored at −70°C until use. After mechanical processing with scissors, total RNA was isolated from ankles using the RNAqueous kit (Ambion). DNase treatment was performed with the DNA-free kit (Ambion) applying 4 U of DNase I for 1 h at 37°C. Absence of detectable DNA contamination was verified by a subsequent PCR with primers matching a murine genomic sequence.

For cDNA synthesis 8 μl of total RNA were mixed with 500 ng of oligo(dT) primer, 0.5 mM deoxynucleosid triphosphate mix (both obtained from Amersham Biosciences), 4 μl of 5× first-strand buffer, 10 mM DTT, and 50 U of SuperScript II RT polymerase (all obtained from Invitrogen Life Technologies). Samples were incubated for 50 min at 42°C followed by heat inactivation at 70°C for 15 min.

Quantitation of mRNA expression profiles of Vcam-1, Il-6, Il-1β, and cyclooxygenase (COX)-2 in ankle joints was conducted by RT-PCR with the LightCycler PCR system (Roche). In each quantitative PCR run, an external standard curve was generated using a 4 log spanning serial dilution of the vector plasmid pGEM-T Easy (Promega) containing one copy of the respective target sequence per plasmid. The standard curves were created by the LightCycler 5.32 software (Roche) and applied for calculation of amplification efficiency and mRNA expression levels. The content of the housekeeping gene hypoxanthine phosphoribosyltransferase (Hprt) was determined in a separate run on the basis of an external standard as well and used for normalization of the acquired values. In addition, porphobilinogen deaminase (synonym: hydroxymethylbilane synthase) levels in the samples were assessed to validate the results of Hprt normalization (data not shown).

PCR was performed in a final volume of 20 μl containing 1 U Platinum Taq polymerase, 2 μl of 10× reaction buffer, 4 mM MgCl₂ (all obtained from Invitrogen Life Technologies), 0.4 μM each primer (for primer characteristics, see Table I), 0.25 mM deoxynucleosid triphosphate mix (Amersham Biosciences), 0.5× SYBR Green I dye (Roche), 0.5 mg/ml BSA (New England Biolabs), 5% DMSO (Sigma-Aldrich), and 2 μl of cDNA in a dilution of 1/10 in H₂O or external standard plasmid template. PCR parameters were set as described above, adapting the annealing temperatures according to Table I.

Statistical analysis of data was performed by two-tailed Student’s t tests using GraphPad PRISM (GraphPad). Comparison of values obtained from more than two groups of animals was also made by one-way ANOVA. Differences were considered significant at p < 0.05.

The potency of the newly developed compound was initially analyzed in vitro in cell-free binding assays and cell-based attachment assays. As listed in Table II, this compound inhibited the α₄β₁-VCAM-1 interaction in the low nanomolar concentration range (IC₅₀, 1.23 nM for cell-free α₄β₁ and 0.29 nM for U-937-linked α₄β₁, respectively). An even higher inhibitory efficacy of S16197 was observed toward the interaction of cell-free α₄β₁ with the CS1-containing fibronectin fragment H/120 (IC₅₀, 0.22 nM). The binding of JY cells that also express α₄, but in combination with the β₇ subunit, to the α₄β₇ ligand MAdCAM-1 was only blocked at a much higher concentration (IC₅₀, 6.23 μM). No effect of S16197 was detected on the binding properties of other β₁-containig integrins (α₂β₁, α₅β₁, and α₆β₁) or on integrins that contain subunits other than α₄ and β₁ (IC₅₀ > 50 μM). Thus, the compound S16197 targets selectively the heterodimeric molecule α₄β₁.

Previous studies have shown that coincubation of B. burgdorferi with mouse ECs (bEnd.3) resulted in enhanced expression of VCAM-1. As a consequence of this up-regulation, α₄β₁-expressing murine L1.2 B lymphoma cells were found to bind much more efficiently to the bEnd.3 cells (9).

Expanding these results, we examined in vitro the efficacy of S16204 to antagonize binding of primary leukocytes to mlEnd.1 ECs that were specifically activated by B. burgdorferi in a cell adhesion assay. As depicted in Fig. 2, both a mixture of inflammatory cells derived from murine peritoneal cavities, consisting mainly of neutrophils, macrophages, and lymphocytes, and the α₄β₁-expressing monocytic U-937 line bound to a marginal extent to resting endothelium. Prestimulation of mlEnd.1 cells with B. burgdorferi for 50 h, however, led to an ∼6-fold increase in the number of cells adhering to the EC monolayer. This adhesion of peritoneal cells and U-937 cells to B. burgdorferi-activated EC was efficiently inhibited by >75% and by nearly 100%, respectively, following addition of 3 nM S16204.

We next investigated whether α₄β₁-mediated processes might be of relevance for the pathogenicity of B. burgdorferi in the mouse model of Lyme arthritis. For this purpose, C3H/HeJ mice, which are known to develop severe arthritis in the course of Borrelia infection, were systemically treated with S18407 at two different doses starting from the day of inoculation and continuing for a period of 21 days.

Ankle swelling, which grossly correlates with arthritis progression, increased rapidly after inoculation in the control groups of animals (vehicle-treated and untreated, respectively) and peaked 2–3 wk p.i., as expected (Fig. 3). In contrast, the clinical course of disease in mice that received the small molecule compound at either dose was markedly alleviated, although a dose-response effect was not evident. The maximum inhibition was achieved by application of 30 mg S18407/kg body weight daily, resulting in significantly lower levels of swelling by days 15 and 19 of infection compared with controls. Of special interest was the finding that the cessation of the treatment did not obviously lead to an exacerbation of arthritis.

Because we have observed essentially no differences between untreated and HEC-treated mice regarding the parameters analyzed in the in vivo experiments described below, for reasons of clarity only data obtained from the untreated mouse group as representative control group are included in Figs. 4-94–9.

To evaluate the arthritis-inhibiting effects of α₄β₁ blockade more closely, we analyzed the rear ankle joints of B. burgdorferi-infected mice histopathologically at the peak level of arthritic swelling on day 16 of infection. In joints of untreated animals (Fig. 4, A and C) the typical histological changes arising in the course of Borrelia infection were identified. Massive and dense cell accumulations were evident in the joint space as well as in tendon sheaths and other areas of connective tissue. In addition, the synovial membrane was thickened due to hyperproliferation, with fragments breaking off from the synovial surface into the joint. Moreover, numerous fibrin clots were discernable in the joint space. Examination at a higher magnification (Fig. 4 C) disclosed neutrophilic granulocytes as the major infiltrating cell population.

In antagonist-treated animals, the overall lesion was less severe compared with their untreated counterparts (Fig. 4,B), although distinct foci of inflammatory cell infiltrates were present especially in areas close to tendons and tendon sheaths, in agreement with the ankle swelling measurements. Besides a reduced total influx of cells, these mice revealed only moderate proliferation of the synovial lining cells, a marginal exfoliation of cells into the joint cavity and fibrin exudates to a much lesser extent (Fig. 4 D). Thus, blockade of α₄β₁ integrin prevented the key hallmarks of B. burgdorferi-induced murine arthritis.

Quantitation of total cells directly isolated out of rear ankle joints and determination of leukocyte (CD45-expressing) and neutrophil (Gr-1high-expressing) fractions were performed on the day of inoculation as well as 2 days and 1 wk thereafter to gain insight in the process of initial cell migration and the influence of α₄β₁ antagonism in the phase before clinical disease onset. In the presymptomatic stage (day 2), no differences were found between treated mice and controls in terms of total articular cell numbers or cell composition: both remained at physiological levels comparable to that in uninfected C3H/HeJ mice (Fig. 5). At 1 wk p.i., however, when signs of arthritic swelling were becoming visible in control mice, the recruitment of cells was markedly attenuated by S18407. The initial infiltration of neutrophils and other leukocytes into joints of treated mice was effectively inhibited, supporting histological findings at the acute stage of disease.

To determine whether the treatment with S18407 affects the expression of proinflammatory mediators that may contribute to the inhibition of cell infiltration and arthritis development, we analyzed the mRNA levels of the key cytokines IL-1β and IL-6 as well as of COX-2 and VCAM-1 in joint tissue by quantitative RT-PCR over a period of 25 days postinoculation.

In untreated, infected mice, the mRNA expression of IL-1β, IL-6, and COX-2, all of which are known to be crucially involved in the pathogenesis of Lyme disease (56, 57, 58, 59), was strongly increased concomitant with arthritis development, reaching maximum values on day 8 p.i. (12-fold, 40-fold, and even 45-fold higher levels of IL-1β, COX-2, and IL-6, respectively, compared with treated mice) and remaining elevated through the time of acute clinical disease (Fig. 6).

In striking contrast, the expression level of these genes in joints of antagonist-treated animals was only slightly enhanced over uninfected control (IL-1β, 2-fold and IL-6, 6-fold, respectively, on day 6 p.i.) or persisted approximately at a constitutive level (COX-2) during the medication period. Interestingly, after cessation of treatment, transcript numbers of IL-6 and COX-2 escalated, even exceeding those of controls, although without any clinical impact.

Induction of VCAM-1, expressed exclusively on resident cells, was not affected by the antagonist and occurred roughly to the same extent (3- to 6-fold increases) in all mice independent of treatment, but, as opposed to controls, with a decreasing tendency from day 2 to day 15 in joints of S18407-treated mice.

Given the profound inhibitory effect of S18407 on recruitment of immune cells to the site of infection and on the induction of pivotal inflammatory modulators, we investigated the effect of α₄β₁ blockade on the capability of B. burgdorferi-infected mice to eliminate bacteria over the course of 48 days postinoculation. By the end of this time, Lyme arthritis in control mice was completely resolved, and bacterial load in the host was usually diminished to a persisting low level of intact organisms (60). Using quantitative PCR, we found no differences between treated and untreated animals in spirochetal DNA levels in ankles at any of the time points examined (Fig. 7). A drop in bacterial burden to a number of several thousands per 10⁵ mouse nidogen-1 copies was observed in treated as well as control groups as soon as 12 days p.i. and was not further reduced over the whole period analyzed. The persistent spirochetemia in infected host tissues for extended periods despite appropriate activation of phagocytes and induction of adaptive immunity is well known, yet there is at present no explanation for this phenomenon (61). Nevertheless, global immunological mechanisms controlling the replication of B. burgdorferi were apparently not affected by suppression of α₄β₁ function. Likewise, long-lasting protection against the disease was established despite α₄β₁ inhibition during the initial immune response, because we observed no arthritis development in mice reinfected several weeks after primary infection (data not shown).

The differentiation phenotype of CD4⁺ T cells during Lyme borreliosis is widely considered to affect the outcome of infection. The existence of highly polarized Th1 cell cytokine patterns and Th1-associated Ig subtypes has been previously described in arthritic mice infected with B. burgdorferi (62, 63). Other studies, however, have proven that immune responses yielding Th2 cells or IL-4 are not sufficient to protect from Lyme arthritis (64, 65). Therefore, to evaluate possible effects of S18407 treatment on the generation of a defined Th response phenotype, the production of Borrelia-specific Igs (IgG1 and IgG2a) and of the Th cell-derived cytokines IFN-γ and IL-4 was assessed.

Serum titers of Th1-related IgG2a in untreated mice were 6.2- and 3.5-fold higher on day 12 and day 35, respectively, than those of Th2-related IgG1 (Fig. 8). Similarly, mice that received the antagonist exhibited 5.2- and 2.2-fold higher levels of IgG2a at the same time. Production of total IgE largely remained at a low level, supporting the bias at IgG generation. In treated animals, slightly elevated titers of IgG1 and IgE relative to control mice were evident by day 35 p.i., but these differences were not statistically significant.

Consistent with the Th1-driven IgG isotype switching, the quantification of Borrelia-induced cytokine secretion by splenic T cells revealed abundant amounts of Th1-derived IFN-γ and barely detectable IL-4. Depletion of CD4⁺ cells as well as the quantification of IFN-γ-expressing spleen cells using a recently established IFN-γ reporter mouse (66) revealed that the majority (∼80%) of IFN-γ was produced by B. burgdorferi-specific CD4⁺ T cells (data not shown). This cytokine pattern was essentially not affected by α₄β₁ blockade (Fig. 9).

These findings provide clear evidence that untreated as well as treated mice had developed a predominant Th1 immune response to infection with B. burgdorferi.

In an attempt to clarify α₄β₁-dependent mechanisms of inflammation and bacterial clearance in an experimental mouse model of human Lyme borreliosis, we used the highly selective small molecule α₄β₁ antagonist S18407 for treatment of mice after infection with B. burgdorferi. The compound inhibited production of inflammatory mediators and leukocyte influx into joints, yet without any consequences on the host’s ability to control spirochetal rejection. Thus, by inhibition of α₄β₁ overwhelming inflammatory processes, known to be detrimental and most likely responsible for arthritis development in this model, were successfully suppressed while beneficial parts of the immune response to B. burgdorferi remained functional.

Innate immune cells, especially neutrophils, and their products are thought to play a pivotal role in the development of Lyme arthritis, and the level of polymorphonuclear leukocyte (PMN) infiltration into joints of infected mice has been associated with the severity of disease (67). Preventing or attenuating the recruitment of these cells should therefore be a prerequisite to ameliorate the course of infection. As a consequence of α₄β₁ blockade, we observed histologically as well as by FACS analysis a reduced, but still notable recruitment of neutrophils to infected ankle joints. This suggests that adhesion molecule pathways not involving α₄β₁ are still operative to govern PMN migration under this condition. Indeed, it has been proven that murine neutrophils generally engage α₄β₁ together with the two main β₂ integrins LFA-1 and Mac-1 to enter inflammatory sites and that there is a substantial functional overlap among these integrins (21). Moreover, an additional α₄β₁- and β₂-independent pathway of PMN migration to arthritic rat joints has been proposed (68). Thus, LFA-1, Mac-1, and possibly other receptors not yet identified are likely to be involved in the residual neutrophil migration found in S18407-treated animals. In addition to an effect on decreased invasion, the accelerated clearance of PMN from inflamed tissue as a further consequence of α₄β₁ antagonism may also contribute to the control of leukocyte burden in joints, according to recent results obtained by β₂ integrin inhibition (69). Quite intriguing, the spreading of infiltrated PMN within various compartments of the articular cavity of treated animals showed a pattern uncharacteristic for murine Lyme arthritis. Whereas cells seemed to accumulate in tendon sheaths and synovial membrane, the joint fluid was almost free of pathologic signs. Fibronectin, in joints derived from plasma and synthesized in situ by fibroblast-like synoviocytes, is a major ECM component in normal and pathological synovium. Following VCAM-1-mediated extravasation, CS1 fibronectin was shown to be the most important ligand for α₄β₁-bearing cells to accomplish (possibly in concert with other components of the ECM) migration along a chemokine gradient to the site of activity (70). Under the experimental conditions of B. burgdorferi infection, this interaction of leukocytes to CS1 fibronectin may be blocked more efficaciously by S18407 than binding to VCAM-1, as indicated in vitro in the cell-free binding assays. This is supported by the finding that α₄β₁ binds with lower affinity to CS1 fibronectin than to VCAM-1 (33, 71). β₁ integrins appear to be the dominant class of integrins that mediate PMN interactions with fibronectin, whereas the contribution of β₂ integrins to adhesion to fibronectin is minimal (72). Therefore, impaired β₁-dependent migration of neutrophils through ECM, unlike extravasation, can hardly be compensated by action of β₂ integrins.

Interestingly, despite the apparent, although reduced, recruitment of leukocytes to joint tissue of treated mice, the local mRNA levels of proinflammatory mediators known to be strongly up-regulated by innate immune cells during B. burgdorferi infection were essentially not elevated over baseline. Consistently, formation of edema in ankles, induced particularly by products of COX-2 (J. Gläsner, M. Röllinghoff, and A. Gessner, manuscript in preparation), was largely abolished. Based on these findings, we propose that α₄β₁ blockade in murine Lyme arthritis predominantly acts by impeding (full) activation of α₄β₁-bearing innate immune cells, whereas the interference with cellular trafficking appears to play a minor role.

Numerous studies have disclosed the substantial contribution of outside-in signaling by integrins to the activation of various neutrophil functions upon adhesion to matrix ligands (e.g., respiratory burst, expression of inflammatory mediators, degranulation, and apoptosis inhibition) (69, 73, 74, 75, 76). As shown in some previous in vivo models, blockade of α₄ integrin by mAbs has led to a partial or complete failure of cell activation, and, as in this study, the disease-ameliorating effects could not fully be attributed to an impaired cell recruitment (77, 78, 79, 80). In studies using infectious agents such as Yersinia enterocolitica or Trichinella spiralis, the defect in cell activation was directly correlated with an increase in the tissue burden of the respective pathogen (77, 78). This finding appears to conflict with our results, although the discrepancy may be explained by the nonspecific targeting of α₄β₇ integrin in addition to α₄β₁ by the anti-α₄ mAbs used. Binding of α₄β₇ to MAdCAM-1 plays a critical role in lymphocyte homing to gut-associated lymphoid organs (81, 82), and clearance of intestinal Y. enterocolitica and T. spiralis, unlike B. burgdorferi, is known to be mediated by T cell-dependent mechanisms (83, 84).

α₄β₁-independent host control of spirochetal growth, despite restricted numbers of neutrophils at sites of infection, demonstrates that these phagocytic cells are not essential for elimination of bacteria. This is in line with previous work suggesting that in the Lyme arthritis model, killing spirochetes is secondary to the role of immune regulation by neutrophils (67).

Numerous publications have pointed to the importance of the humoral response in preventing infection by B. burgdorferi, in controlling spirochete levels in tissues, and in clearing arthritic lesions in the mouse (85, 86, 87). Because Borrelia-specific activation of lymphocytes, i.e., priming of T cells to produce large amounts of IFN-γ and secretion of Abs by plasma cells, are not impaired under treatment with S18407, it is likely that, independent of α₄β₁ inhibition, the acquired host defense plays a dominant role for elimination of bacteria from tissues. It further implies that the contribution of α₄β₁ signaling to T cell activation, highlighted by previous reports (88, 89), is functionally compensated by other costimulatory pathways.

In conclusion, this study has shown that α₄β₁-mediated processes are involved in the PMN-dominated development of murine Lyme arthritis, but dispensable for the induction of protective immunity against B. burgdorferi, and that the novel α₄β₁ antagonist S18407 offers a safe approach to efficiently disturb α₄β₁-dependent cell adhesion associated with this infectious disease.

We thank Mrs. Claudia Giessler for her excellent technical assistance.

The authors have no financial conflict of interest.