Drs. Stolen, Jalkanen, and Salmi own stock in Biotie Therapies Corporation. Drs. Jalkanen and Salmi formerly held a patent on vascular adhesion protein 1. Drs. Smith, Pihlavisto, and Alaranta own stock and/or hold stock options in Biotie Therapies Corporation.
Vascular amine oxidases are needed for leukocyte extravasation into inflamed joints in vivo
Article first published online: 31 AUG 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 54, Issue 9, pages 2852–2862, September 2006
How to Cite
Marttila-Ichihara, F., Smith, D. J., Stolen, C., Yegutkin, G. G., Elima, K., Mercier, N., Kiviranta, R., Pihlavisto, M., Alaranta, S., Pentikäinen, U., Pentikäinen, O., Fülöp, F., Jalkanen, S. and Salmi, M. (2006), Vascular amine oxidases are needed for leukocyte extravasation into inflamed joints in vivo. Arthritis & Rheumatism, 54: 2852–2862. doi: 10.1002/art.22061
- Issue published online: 31 AUG 2006
- Article first published online: 31 AUG 2006
- Manuscript Accepted: 31 MAY 2006
- Manuscript Received: 6 OCT 2005
- Finnish Academy and the Sigrid Juselius Foundation
- NIH. Grant Number: F32-HL-010495-03
Leukocyte traffic from the blood to the joints is crucial in the pathogenesis of arthritis. A bifunctional endothelial cell–surface glycoprotein, AOC3 (amine oxidase, copper-containing 3; also known as vascular adhesion protein 1), has both adhesive and enzymatic properties. We undertook this study to determine the contribution of AOC3 and its oxidase activity to leukocyte trafficking into inflamed joints in vivo.
We used gene-modified animals, molecular modeling, an AOC3 enzyme inhibitor, oxidase assays, and arthritis models (adjuvant-induced arthritis [AIA] in rats and anti–type II collagen antibody–induced arthritis in mice) to dissect the importance of AOC3 in vivo.
The AOC3 inhibitor fitted well with a covalent binding mode into the active site of the AOC3 crystal structure. It selectively blocked the oxidase activity of AOC3 in enzyme assays. Intraperitoneal and oral administration of the AOC3 inhibitor significantly ameliorated rat AIA. In anti–type II collagen antibody–induced arthritis in mice, the AOC3 inhibitor also improved the outcome of the joint inflammation. The acute semicarbazide-sensitive amine oxidase blockade by the inhibitor had even more pronounced effects than genetic deletion of AOC3. Enzymatic analyses showed that the inhibitor also blocked 2 other structurally very closely related AOCs, but not any of more than 100 other enzymes tested.
These are the first data to demonstrate that the enzymatic activity of the atypical endothelial adhesion molecule AOC3, and possibly that of other closely related ecto-oxidases, is crucial for leukocyte exit from the vessels in inflamed joints in vivo.
Leukocyte exit from the blood into tissues is orchestrated by multiple adhesive interactions between blood-borne leukocytes and the vascular lining. Many steps of this cascade are governed by adhesion and activation molecules (1). Selectins and their oligosaccharide-based counterreceptors can account for initial tethering and rolling (2, 3). Thereafter, binding of chemokines to their 7-transmembrane receptors on leukocytes can induce signals, which result in the activation of integrins that allows shear-resistant firm adhesion of leukocytes to the vessel wall (4, 5). Finally, integrins, immunoglobulin superfamily members, and other molecules mediate the transmigration of the leukocyte through the vessel wall into the tissue (6–8).
AOC3 (amine oxidase, copper-containing 3; also known as vascular adhesion protein 1 [VAP-1]) (EC 22.214.171.124) is a new type of adhesion molecule with 2 interrelated functions (9, 10). It belongs to the cell surface–expressed, semicarbazide-sensitive amine oxidase (SSAO) family (11). These enzymes catalyze oxidative deamination of primary amines in a reaction resulting in the production of the corresponding aldehyde, H2O2, and ammonium (10, 12, 13). On endothelial cells, AOC3 can serve as a traditional adhesion molecule whose function is blocked by monoclonal antibodies (mAb) (14–19). On the other hand, the enzymatic activity of AOC3, which is not inhibited by the anti-AOC3 mAb, also appears to be critical for its adhesive functions in flow-based assays in vitro (16, 17, 20). Enzyme inhibitors targeting AOC3 may therefore offer new ways of preventing harmful inflammation in vivo.
We have recently shown that AOC3-deficient mice display diminished inflammatory responses in vivo due to impaired interactions between leukocytes and vascular endothelial cells (21). However, the contribution of AOC3 in arthritis has not been analyzed. Moreover, the AOC3-deficient animals do not allow us to study the contribution of the SSAO catalytic activity as such to the adhesive functions of AOC3. Here we used a small molecule enzyme inhibitor and AOC3-deficient animals to unravel the importance of AOC3 and its oxidase activity in the development of arthritis in vivo and to test the potential of AOC3 blockade in alleviating inflammation.
MATERIALS AND METHODS
Enzyme inhibitors and modeling.
Synthesis of the SSAO inhibitor BTT-2052 has been described (22). It is a trans indane hydrazino alcohol enantiomer with the chemical formula (1S,2S)-2-(1-methylhydrazino)-1-indanol.
From the Protein Data Bank (PDB; online at www.rcsb.org), we downloaded the crystal structures of human AOC3 (PDB code 1us1) and other copper-containing amine oxidases. The superposition of enzyme structures was made with VERTAA in the BODIL modeling environment (23), and the side-chain orientations at the ligand binding site were studied by using the side chain rotamer library (24) incorporated in BODIL. The templates for the active conformation of topaquinone were obtained from the crystal structures of other copper-containing amine oxidases with bound inhibitor molecules (PDB codes 1sii, 1sih, and 1spu) (25, 26). The ligand structure was energy minimized in SYBYL 7.0 (Tripos, St. Louis, MO) with steepest descent and conjugate gradient methods. BTT-2052 was docked manually into the ligand binding site so that the ligand was covalently bound to the topaquinone in a way similar to the inhibitors in the crystal structures of other copper-containing amine oxidases (25, 26). The side-chain conformations of surrounding amino acids were selected so that both intramolecular and intermolecular interactions were energetically as favorable as possible. Some amino acids have different conformations in the crystal structure of bovine plasma copper-containing amine oxidase (PDB code 1tu5) (27) and human AOC3. Therefore, this structure was also used to guide the selection of side-chain conformations.
Male Lewis rats (Harlan, Horst, The Netherlands) weighing ∼150 gm were used in adjuvant-induced arthritis (AIA) studies. Wild-type and AOC3-deficient animals (21) on a pure 129S6 background were used in the mouse models. Age- and sex-matched animals were used in all experiments. All animals were handled in accordance with institutional animal care policy, and local ethics committees for animal experiments approved the studies.
AIA in rats.
Rat adjuvant arthritis was induced by injection of 0.5 mg of dead Mycobacterium butyricum (Difco, Detroit, MI) in 0.1 ml of liquid paraffin into the right hind footpad. The clinical severity of the arthritis was assessed on days 3, 7, 10, 13, 17, 20, 24, and 28 by grading each paw on a scale of 0–4 for changes in redness and swelling, as follows: 0 = no changes, 1 = paws with swelling of joints of the digits or focal redness, 2 = paws with mild swelling of wrist or ankle joints; 3 = paws with severe swelling of the entire paw, and 4 = paws with deformity or ankylosis. The macroscopic score was expressed as the cumulative scores in all paws (4 paws with scores of 0–4; maximum score 16).
Treatment was initiated on day 1, following adjuvant injection on day 0. The rats (n = 10 per group) were treated either with twice daily intraperitoneal (IP) injections of BTT-2052 in sterile water using a dosing volume of 1 ml/200 gm at 12.5 mg/kg, 25 mg/kg, or 50 mg/kg, or by mouth (also twice daily) at 50 mg/kg. Control rats were treated with 150 mM saline or diclofenac sodium at 2 mg/kg by mouth.
At the end of the experiment, mice were killed, and 5 uninjected ankle joints were removed from the 50 mg/kg BTT-2052 IP-injected and saline control groups and fixed in 4% phosphate buffered formalin, decalcified, and stained. Hematoxylin and eosin staining was used for general morphology and Giemsa stain for mast cells and inflammatory cells. The sections were examined for histopathologic signs of inflammatory arthritis, focusing mainly on the ankle joints but also considering all smaller joints between the tarsal and metatarsal bones, if available. Histopathologic changes were scored by 3 independent blinded observers, assessing the degree of inflammatory cell infiltration and neovascularization in the synovial layers (absent = 0, minimal = 1, moderate = 2, and marked = 3). Among all leukocytes, we estimated the percentages of different inflammatory cells, including lymphocytes, macrophages, polymorphonuclear leukocytes (PMNs), and mast cells. These cell types were identified by nuclear and cell morphology and by staining characteristics based on Giemsa stain. The magnitude of diffuse synovial cell layer thickening was also scored (<5 cell layers = 0, 5–10 cell layers = 1, 11–15 cell layers = 2, and >15 cell layers = 3). Cartilage and bone erosion was measured by assessing pannus formation, formation of cavities (e.g., neovascularization within articular cartilage), osteolytic lesions, and osteoclast numbers, with a score of 0–3.
Anti–type II collagen antibody–induced arthritis in mice.
For the anti–type II collagen–induced arthritis model in mice, 4 groups of animals were used: 1) wild-type controls (a negative control group), 2) wild-type animals in which arthritis was induced and which received no treatment, 3) wild-type animals in which arthritis was induced and which received SSAO inhibitor treatment, and 4) AOC3-deficient animals in which arthritis was induced, but which did not receive any treatment. For inducing the disease, on day 0, male 129S6 mice (ages 8–9 weeks) were injected intravenously with 400 μl (4 mg) of an Arthrogen-CIA Monoclonal Antibody Cocktail (Chemicon, Temecula, CA) (28, 29). In the morning of day 2, 50 μg (200 μl) of lipopolysaccharide (LPS) was given IP to each mouse according to the manufacturer's protocol to enhance the development of the disease in this relatively arthritis-resistant strain. To exclude any LPS-mediated effects, LPS was also given to the wild-type control group, which did not receive Arthrogen. BTT-2052 was administered IP (0.625 mg in sterile water per mouse, ∼25 mg/kg body weight) 2 times a day starting in the evening of day 2.
Clinical scoring was determined by the consensus of 2 blinded investigators on day 6. In preliminary experiments, maximal inflammation was seen around day 6 in this spontaneously resolving model of arthritis; therefore, day 6 was chosen as the end point in all other studies. The severity of arthritis was scored based on an examination of all 4 paws (29). For each paw, points were given for swelling and redness of the toes (0 or 5 points), knuckles/foot pad (0 or 5 points), and wrist/ankle (0 or 5 points). Thus, every joint was scored as 0 (not affected) or 5 (affected) points, and the macroscopic score for the animal was the cumulative sum of all joint scores (maximum score 60; i.e., 4 paws × 3 joints per paw × 5 points maximum per paw).
On day 6, mice were killed, and all 4 paws (toes, ankle/wrist, and elbow/knee) were removed. The samples were fixed overnight in 10% formaldehyde, decalcified with EDTA, dehydrated, and embedded in paraffin (29). Whole joints were sectioned at 5 μm thickness, and every twentieth section was stained with hematoxylin and eosin. Joint sections were scored semiquantitatively in a blinded manner for leukocyte infiltration. The toes, wrists/ankles, and elbows/knees were scored separately, and the numbers of inflammatory cells in the joint space, synovium, and periarticular tissue were evaluated. The scoring system was as follows: 0 = no or very occasional inflammatory cells, 1 = moderate numbers of inflammatory cells, and 2 = strong or massive inflammatory infiltration.
Cloning of mouse AOC2.
The full-length complementary DNA (cDNA) for mouse AOC2 was cloned from mouse lung RNA using reverse transcriptase–polymerase chain reaction (RT-PCR). Briefly, total RNA from mouse lung was extracted using the RNeasy Mini kit (Qiagen, Hilden, Germany). The RNA was reverse transcribed with Expand Reverse Transcriptase (Roche Molecular Biochemicals, Mannheim, Germany) as suggested by the manufacturer, and RT-PCR was performed using the Expand Long Template PCR System (Roche Molecular Biochemicals) according to the instructions of the supplier. The primers used were based on GenBank accession no. AF350446 and were as follows: 5′-CAGTGCCAGCCATGAATCT-3′ (forward) and 5′-CCTCAGGCCTATAAGCCTTC-3′ (reverse). The resulting PCR product of 2,293 bp was cloned into pGEM-T Easy (Promega, Madison, WI) for sequencing and subsequently into expression vector pcDNA3.1 (Invitrogen, Carlsbad, CA) for further studies.
The activities of SSAOs (AOC1 [diamine oxidase] , AOC2 [retina-specific amine oxidase] , AOC3, and lysyl oxidase [LOX] [32,33]) and flavin adenine dinucleotide (FAD)–containing monoamine oxidases (MAOs) A and B (34) were tested using their preferred substrates and either fluorimetric or photometric assays, as appropriate. Tissue lysates, purified enzymes, and transfectants were used as the enzyme source. Recombinant human MAO-A was from the microsomal fraction of baculovirus-infected cells (Sigma, St. Louis, MO), purified AOC1 was from porcine kidney (Sigma), and purified recombinant human AOC3 was from Chinese hamster ovary (CHO) cells. For AOC1, AOC3, and MAO assays, tissue lysates from AOC3-deficient animals and their wild-type littermates were prepared as described elsewhere (21). LOX measurements were made for the entire aorta and lungs as described previously (35). The expression plasmids encoding human and mouse AOC3 have been previously described (11, 36). Human AOC2 was cloned from a lung cDNA library (Elima K, Salmi M, Jalkanen S: unpublished observations).
Fluorimetric SSAO analyses for AOC3, MAO-A, and LOX were performed using the lysates prepared from transfectants or tissues in the presence of preferred substrates and specific inhibitors. A catalytic reaction was initiated by addition of 1 mM benzylamine (AOC3 substrate), 1 mM tyramine (MAO-A substrate), or 10 mM cadaverine (LOX substrate), and the Amplex red H2O2-detecting mixture (Invitrogen, Karlsruhe, Germany). The assays were also done in the presence of 0.5 mM clorgyline (MAO inhibitor), 200 μM semicarbazide (inhibitor of SSAO), or 1 mM β-aminopropionitrile (LOX inhibitor), as appropriate. To evaluate the amount of H2O2 formed via the given enzyme reaction, the values obtained in the presence of specific enzyme inhibitors were subtracted from the total amount of H2O2 formed. The substrate specificity assays of mouse AOC2 were performed by incubating lysates of vector and mouse AOC2 transfectants with different substrates (benzylamine, methylamine, β-phenylethylamine, tyramine, and tryptamine; 1 mM each) or without any substrate in the fluorimetric assay.
Spectrophotometric assays were performed as described elsewhere (37), by mixing the enzyme, substrates, appropriate inhibitors, and the chromogenic solution and then analyzing the formation of H2O2 at an absorbance of 492 nm. In these experiments, different amine oxidases (purified enzymes, transfectants, or tissue lysates) were preincubated with increasing concentrations of BTT-2052 (for 15 minutes at 37°C) before the addition of the appropriate substrate and chromogenic solution.
Ki values for inhibition of the SSAO activity of VAP-1 with the inhibitor were determined using recombinant VAP-1 expressed in CHO cells and 1 mM benzylamine as substrate, as described previously (38). Ki determinations for total MAO inhibition were performed using rat liver homogenates and 0.5 mM tyramine as substrate. Rat liver MAO and human liver MAO activities were previously shown to be comparable (results not shown). The mean ± SD Km value used for VAP-1 was 90 ± 5 μM and that for MAO was 62 ± 4 μM.
RNA isolation and quantitative PCR.
Total RNA was isolated from cremaster muscle and Peyer's patches using the Ultraspec RNA isolation kit (Biotecx, Friendswood, TX) and reverse transcribed using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) and DNase I digestion according to the manufacturer's instructions. Probes and primers for mouse AOC2, AOC1, and β-actin were designed using either the Assays-by-Design (AOC1) or the Assays-on-Demand (AOC2 and β-actin) system from Applied Biosystems (Foster City, CA). The sequence for mouse AOC1 was derived by searching the mouse genome with BLAST at the University of California, Santa Cruz web site (http://genome.ucsc.edu/) using the human AOC1 sequence (X78212) as input. PCR reactions were carried out as suggested by Bio-Rad using iQSupermix and were run on an iCycler thermal cycler (Bio-Rad). The quantitative PCR runs were performed at least twice with 2 separate cDNA samples prepared from each RNA preparation. The efficiency of the target amplification versus that of the reference amplification was determined to be approximately equal by performing a validation experiment, as suggested in the Applied Biosystems User Bulletin no. 2. Changes in cycle threshold levels (ΔCt) were calculated by subtracting the average of β-actin Ct values from the average of target gene Ct values, and the relative expression levels of AOC2 and AOC1 in tissues of knockout mice were shown in comparison with those in tissues of wild-type mice.
Except where indicated otherwise, all data are presented as the mean ± SEM. Student's t-tests (unpaired, 2-tailed) were used to compare the effects in inhibitor-treated groups and in the corresponding genotypes without the enzyme inhibitor. In the rat AIA model, the clinical scores were statistically evaluated by analysis of variance followed by the Newman-Keuls multiple comparison test using GraphPad Prism software, version 2.01 (GraphPad Software, San Diego, CA). In competitive enzyme assays, the Ki, Km, and 50% inhibition concentration (IC50) values were computed with curve-fitting programs (GraphPad Prism, version 3.0).
BTT-2052 binds covalently and selectively to AOC3 and inhibits SSAO activity.
We used the small molecule SSAO inhibitor BTT-2052 (22) to study the role of the enzymatic activity of AOC3 in arthritis in vivo. The inhibitor is a trans indane hydrazine alcohol enantiomer (Figure 1A). The availability of the crystal structure of human AOC3 (39) allowed us to model the interaction between the inhibitor and the enzyme. AOC3 is a heart-shaped dimer having a conserved aspartic acid catalytic base at position 386 and a unique topaquinone cofactor (a posttranslational modification of an intrinsic tyrosine) that is necessary for the catalytic reaction at the active site buried deeply within the protein and only accessible via a narrow substrate channel. Since the topaquinone is the inactive conformation in the crystal structure, it was manually rotated into the active conformation using the crystal structures of other copper-containing amine oxidases with bound inhibitor molecules as templates (25, 26). The active conformation of topaquinone is stabilized by the hydrogen bonds to Tyr284 and a water molecule that is, in turn, coordinated to the copper ion (Figure 1B). BTT-2052 was then docked manually into the ligand binding site so that the hydrazine moiety of the ligand was covalently bound to the topaquinone in a way similar to the inhibitors in crystal structures of other copper-containing amine oxidases. As shown in Figure 1B, BTT-2052 forms hydrogen bonds with Asp386 and exhibits good hydrophobic packing with residues Met211, Tyr384, Phe389, and Leu469.
In enzyme assays, the inhibitor BTT-2052 efficiently blocked the enzymatic activity of AOC3, but not that of an unrelated MAO (Figure 2). Moreover, BTT-2052 was tested in a large panel (130 targets) of in vitro receptor binding assays and showed no significant inhibition or stimulation. The structural and functional data thus suggest that the inhibitor BTT-2052 binds covalently to AOC3 and blocks its catalytic activity in a selective manner.
Clinical and histologic improvement of rat AIA by SSAO inhibition.
The in vivo efficacy of SSAO inhibition in treating inflammation was studied using different arthritis models. In rat AIA, the severity of the disease increased in the saline-treated control group for ∼3 weeks after disease induction and remained constant thereafter. In rats treated twice daily with IP injections of BTT-2052 at 25 and 50 mg/kg, the clinical signs of arthritis were alleviated in a dose-dependent manner (P < 0.01) (Figure 3A). The clinical score in the 50 mg/kg group was reduced to that of the group treated with diclofenac (a nonsteroidal antiinflammatory drug used as a positive control). Importantly, when dosed twice daily by mouth at 50 mg/kg, BTT-2052 was also able to significantly reduce the clinical inflammation score compared with that in saline-treated animals, although not to the level in the diclofenac-treated control group (Figure 3B). These results show that SSAO inhibition is highly effective at ameliorating the development of joint inflammation in vivo in a dose-dependent manner.
In histologic analyses, the saline-treated group showed that an inflammatory response dominated the arthritis, with marked edema and widening of the intraarticular space, including the formation of synovial sacs, which contained neutrophil granulocytes in fibrin cloths. Pannus formation, diffuse synovial thickening, and cartilage erosion along with myelosclerosis were also typical. In the BTT-2052–treated group (50 mg/kg IP), the degree of inflammatory cell infiltration and neovascularization in the synovial layers was reduced (Table 1). Cartilage and bone erosion was also reduced, reflecting the disease-modifying effect of SSAO inhibition (Table 1).
|Treatment group, animal||Inflammation score (range 0–3)†||Inflammatory cells, %‡||Diffuse synovial thickeninge (range 0–3)§||Cartilage/bone erosion (range 0–3)¶|
On the basis of the in vitro results, BTT-2052 (Figure 2) is not a fully selective AOC3 inhibitor. To confirm that MAO inhibition has no effect on arthritis, a classic MAO inhibitor, pargyline, was administered IP to the rats at a dose of 25 mg/kg once a day, which is known to induce a complete MAO inhibition. In this animal group, the signs of arthritis were similar in the saline- and pargyline-treated animals (data not shown), supporting the conclusion that the antiarthritic effect of the AOC3 inhibitor is not due to MAO inhibition.
Alleviation of arthritis in mice by SSAO inhibition and genetic deletion of AOC3.
To compare the effects of SSAO inhibition and genetic deletion of AOC3 in arthritis, the disease was induced in mice using a cocktail of anti–type II collagen antibodies. Four groups of animals were used: 1) wild-type animals with no induction of joint inflammation (negative controls), 2) wild-type animals in which arthritis was induced and which received no treatment (positive controls), 3) wild-type animals in which arthritis was induced and which were treated with the SSAO enzyme inhibitor BTT-2052, and 4) AOC3-deficient animals in which arthritis was induced, but which did not receive any treatment. The clinical severity of the disease was scored blindly and semiquantitatively on day 6 for each paw, based on redness and swelling (Figures 4A and B). When wild-type animals were treated with BTT-2052 starting from day 2 of disease development, there was a significant 85% reduction in the severity of arthritis on day 6 (Figure 4C). Sixty percent of the animals were scored 0, and only 1 animal received a score of 10. AOC3-deficient animals also showed milder arthritis than their wild-type littermates (Figure 4C). While 73% of the wild-type animals had scores ≥10, only 31% of the AOC3-deficient mice had scores over this limit (P < 0.05). Moreover, when every joint was scored separately, there was a statistically significant reduction of clinical inflammation both in the AOC3-deficient mice and in the BTT-2052–treated wild-type mice (Figure 4D).
In this mouse model of acute arthritis, there was marked infiltration of PMNs in most joints of the untreated wild-type mice (Figures 5A–D). The number of infiltrating PMNs was significantly reduced in animals treated with the SSAO inhibitor BTT-2052, and the reduction was most striking in the small joints (Figure 5E). AOC3-deficient animals in which arthritis was induced also had significantly fewer infiltrating leukocytes than their wild-type arthritic controls (Figure 5E). The 2 different arthritis models thus show that genetic deletion of AOC3 or blockade of the AOC3 enzyme activity by a small molecule SSAO inhibitor improves the clinical and histologic outcome of arthritis in vivo.
SSAO inhibitor BTT-2052 blocks several amine oxidases.
Direct comparison of the arthritis-suppressing potential of AOC3 targeting surprisingly revealed that acute SSAO blockade with the enzyme inhibitor caused more robust changes than deletion of the AOC3 gene (Figures 4 and 5). Since there are multiple copper-containing amine oxidases in mice and humans (30–34), these data suggest that the SSAO inhibitor may block other oxidases in addition to AOC3.
Comparison of wild-type and AOC3-deficient animals showed that the SSAO activity was almost completely abolished in normally AOC3-rich tissues in the AOC3-knockout mice (Figure 6A). There were no significant changes in the enzymatic activity and/or messenger RNA (mRNA) synthesis of copper-containing AOC1 (diamine oxidase) or FAD-containing MAOs A and B between the wild-type and AOC3-deficient animals (Figures 6A and B). Interestingly, we found a compensatory increase in the synthesis of AOC2 mRNA (also known as retina-specific amine oxidase , although it is expressed in multiple tissues [Elima K, Salmi M, Jalkanen S: unpublished observations]) in AOC3-deficient animals (Figure 6B).
In competitive enzyme assays, BTT-2052 was shown to efficiently and dose-dependently inhibit mouse and human AOC3, with IC50 values in the nanomolar range (Figures 6C and D). It did not affect the activity of mouse, human, or rat MAO (Figures 2 and 6C and D). In contrast, it dose-dependently inhibited the activity of purified AOC1 (from pig kidney). Using transient transfections and multiple potential AOC substrates, we found that β-phenylethylamine, tyramine, and tryptamine were potential substrates for mouse AOC2 (Figure 6E). Since the enzymatic activity of mouse AOC2 was low in our expression system, we were not able to directly assess the inhibitory potential of BTT-2052 on it. Nevertheless, since the preferential order of amine substrates is exactly the same with human and mouse AOC2, we were able to test the effect of BTT-2052 on the human enzyme, which gave higher enzymatic activity. These data clearly show that BTT-2052 also inhibits AOC2, although less potently than AOC1. Therefore, either AOC1 or AOC2 or both may also be involved in regulating leukocyte–endothelial contacts in vivo.
Here we have shown for the first time that AOC3 is important for the pathogenesis of arthritis in vivo. When the SSAO activity of AOC3 was blocked by a small molecule inhibitor, the clinical and histologic signs of arthritis were reduced in 2 different models. Since the SSAO inhibitor also blocked arthritis when given by mouth, it could be useful for clinical administration.
The inhibitory effect of anti-AOC3 antibodies on inflammation has been documented in many animal studies and was therefore not analyzed in this study. In contrast, very little is known about the importance of the oxidase activity of AOC3 in supporting leukocyte trafficking in vivo. In in vitro assays, the adhesive function of AOC3 can be blocked either by mAb that do not inhibit the enzymatic activity or by SSAO inhibitors (16, 17, 20). These data have led to the current working model that leukocytes first interact with AOC3 via adhesive epitopes that are detected by the anti-AOC3 mAb, and then via a leukocyte surface–expressed substrate (such as galactosamine or a lysine-containing peptide) that can penetrate through a narrow substrate channel into the catalytic site, buried deep inside the enzyme, to initiate the SSAO reaction. If either of these steps is missing, AOC3 is functionally inactive. AOC3-deficient animals cannot be used to verify this hypothesis in vivo, since both the antibody epitopes and SSAO activity are missing from those animals. Our current data show that SSAO enzyme activity is indeed needed for inflammation in vivo. The enzyme reaction results in the formation of an aldehyde, ammonium, and H2O2. All these bioactive products could be involved in the modulation of leukocyte extravasation. In particular, H2O2 is known to alter the expression of other adhesion molecules (P-selectin, vascular cell adhesion molecule 1), chemokine receptors, and matrix metalloproteinases (40–43) and may thus modulate leukocyte extravasation by signaling effects.
AOC3 blockade caused a partial alleviation of joint inflammation. The partial, rather than complete, effect is typical of blocking of most other adhesion molecules as well. This is inherently related to the nature of the vital extravasation cascade. Thus, during elicitation of a vigorous inflammation, such as that seen in the large joints of arthritic mice, the inflammatory stimuli may be so overwhelming that all, or at least multiple, parallel extravasation pathways are maximally activated. In this case, the redundancy of the extravasation system will allow compensatory, non–SSAO-mediated, pathways to be used to a larger extent to ensure proper defense reactions. In any case, the utility of SSAO inhibitors in alleviating inflammation when administered by mouth supports the feasibility of this approach in controlling arthritis in vivo.
The SSAO inhibition surprisingly caused larger effects on leukocyte–endothelial interactions and inflammatory disease than did deletion of the AOC3 gene. BTT-2052 was derived from a chemical screen designed to find new SSAO inhibitors (22). It is a water-soluble carbocyclic hydrazine compound that has suitable physicochemical properties to be used in in vivo experiments, and hence, it is superior to the prototype SSAO inhibitors semicarbazide and hydroxylamine. The inhibitor did not affect the hematologic parameters (data not shown) and did not cause any obvious side effects in the current arthritis models or in preclinical tests at concentrations used in vivo. The compound is a potent inhibitor of AOC3 and did not show significant activity against MAO or multiple unrelated molecules including many key receptors, kinases, and other enzymes. However, our data showed that BTT-2052 also inhibits AOC1 and AOC2. These 2 enzymes are expressed in endothelial cells in many tissues (ref.44, and Elima K, Salmi M, Jalkanen S: unpublished observations) and could mediate SSAO-catalyzed reactions involved in leukocyte adhesion in a manner similar to that described for AOC3.
It should be emphasized that currently, no inhibitors specific to a given SSAO species are available, making it impossible to directly demonstrate the involvement of the other SSAO molecules in leukocyte extravasation. The stress-induced glucocorticoid effects may have potentiated the antiinflammatory effects in the SSAO inhibitor–treated group as compared with the control and AOC3-deficient groups in our experimental setting, since we did not include a separate control group that received daily vehicle injections (to reduce the numbers of animals needed). However, we have seen in another model that the SSAO inhibitor indeed has greater effects on the inflammation than does ablation of the AOC3 gene. Using real-time imaging of leukocyte–endothelial cell interactions in inflamed cremaster vasculature, we observed that the inhibition of leukocyte extravasation was less profound in AOC3-null animals than in wild-type animals treated with the SSAO inhibitor (Marttila-Ichihara F, Salmi M: unpublished observations). Moreover, leukocyte extravasation was diminished more in the AOC3-deficient mice treated with the SSAO inhibitor than in AOC3-deficient animals treated with the vehicle. Finally, we also cannot exclude the possibility that the SSAO inhibitor used in the current study, or in the cremaster model, might also inhibit some other molecule involved in inflammation. Nevertheless, our current data clearly show that blockade of the SSAO activity can profoundly alleviate inflammatory reactions in vivo. Findings of the very recent studies using different SSAO inhibitors and different models of inflammation fully support our conclusions (45, 46).
AOC3 has several benefits as a potential target for antiadhesive therapy. It is normally virtually absent from the luminal surface of uninflamed endothelium. Upon inflammation, AOC3 is rapidly translocated from intracellular storage vesicles to the luminal surface, both in mice and in humans (47, 48). Since only surface-expressed enzyme is available for binding neutralizing anti-AOC3 antibodies, and since it is enzymatically active only at this location (37), blocking of its function should have minimal effects on the function of the normal immune surveillance systems. This is, in fact, consistent with the observed phenotype of AOC3-deficient animals (21). In the absence of AOC3, mice are microscopically and macroscopically healthy and have only minor defects in leukocyte trafficking to the gut under normal conditions. Moreover, most adhesion molecules are currently being targeted by humanized antibodies, which have their inherent limitations in clinical use. This is primarily due to the fact that many other adhesion molecules are difficult to target with small molecular compounds. In contrast, the enzymatic nature and available crystal structure of AOC3 greatly facilitate these efforts.
When used in in vitro adhesion assays, anti-AOC3 mAb inhibit the binding of leukocytes to human vessels in several inflammatory disorders, including arthritis (49). The good in vivo efficacy of acute administration of an SSAO inhibitor in alleviating arthritis in rodents therefore suggests that such inhibitors could be useful in the clinical setting. The possibility of targeting adhesive cell-surface enzymes with a small molecule compound offers new venues for antiadhesive therapy that can circumvent many of the difficulties encountered when using neutralizing mAb.
We thank Prof. Mark Johnson for providing the facilities for the modeling studies and Anne Sovikoski-Georgieva for secretarial help.