Dihydroorotate dehydrogenase is a high affinity binding protein for A77 1726 and mediator of a range of biological effects of the immunomodulatory compound.

A protein with high affinity (K 12 nM) for the immunomodulatory compound A77 1726 has been isolated from mouse spleen and identified as the mitochondrial enzyme dihydroorotate dehydrogenase (EC 1.3.3.1). The purified protein had a pI 9.6-9.8 and a subunit M of 43,000. Peptides derived from the mouse protein displayed high microsequence similarity to human and rat dihydroorotate dehydrogenase with, respectively, 35 and 39 out of 43 identified amino acids identical. Dihydroorotate dehydrogenase catalyzes the fourth step in de novo pyrimidine biosynthesis. The in vitro antiproliferative effects of A77 1726 are mediated by enzyme inhibition and can be overcome by addition of exogenous uridine. The rank order of potency of A77 1726 and its analogues in binding or enzyme inhibition was similar to that for inhibition of the mouse delayed type hypersensitivity response. It is proposed that inhibition of dihydroorotate dehydrogenase is an in vivo mechanism of action of the A77 1726 class of compounds. This was confirmed using uridine to counteract inhibition of the murine acute graft versus host response.

Leflunomide (N-(4-trifluoromethylphenyl)-5-methylisoxazol-4-carboxamide, HWA 486) is a novel immunomodulatory and anti-inflammatory compound currently under evaluation in phase III clinical trials for the treatment of rheumatoid arthritis. It has shown dose-dependent clinical efficacy in a 6-month, double-blind, placebo-controlled study on patients with longstanding active arthritis, 81% of whom had failed previous disease modifying anti-rheumatic drug therapy (1). In addition it has been shown to be effective in controlling the development of autoimmune disorders and delaying transplant rejection in animals (2)(3)(4)(5). Its primary metabolite A77 1726 (Structure 1), which mediates the immunosuppressive and disease-modifying effects of the parent drug, inhibits proliferation of cell lines and mitogen-or cytokine-stimulated lymphoid cells in vitro (6,7) by inhibiting progression from the G 1 to the S phase of the cell cycle (7). Although the biochemical mechanisms by which A77 1726 exerts its effects are unknown, they have been shown to differ from those of other immunosuppressive agents such as corticosteroids, cyclosporin A, rapamycin, or mycophenolic acid (7).
The results presented here describe the purification of a high affinity binding site for A77 1726 and its identification as the mitochondrial enzyme dihydroorotate dehydrogenase. Evidence is presented that identifies the enzyme as a mediator of the in vitro and in vivo effects of the compound (see Structures 1 and 2).

Materials
[2,6-phenyl- 3 Structure 1) were synthesized in the Chemistry Department, Hoechst Roussel Ltd., Swindon, United Kingdom, except for leflunomide, which was prepared in the Chemistry Department, Hoechst AG, Werk Kalle-Albert, Wiesbaden, Germany. SP-Sepharose HP, PBE118 polybuffer exchange matrix, Pharmalyte pH 8 -10.5 and PD-10 desalting columns were purchased from Pharmacia Biotech Inc. Hydroxylapatite (Biogel HTP) was obtained from Bio-Rad. Trifluoroacetic acid and Biobrene Plus were from Applied Biosystems and endoproteinase Asp-N from Boehringer Mannheim. Coenzyme Q10, dihydroorotic acid, and dichlorophenolindophenol (DCIP), 1 were obtained from Sigma. Nonyl glucoside was obtained from Sigma or Calbiochem depending on availability or was synthesized in the Chemistry Department of Hoechst Roussel Ltd. using published procedures (8,9). All other chemicals were obtained from commercial sources.

Methods
All procedures in which native proteins were handled or processed apart from the protein assays themselves were carried out either at 4°C or on ice.
Preparation of Mouse Spleen Membranes and Soluble Preparations-Crude mitochondrial/microsomal membranes from spleens of CD1 mice (Charles River) were prepared by homogenization and differential centrifugation (10). Homogenization buffer was 25 mM sodium phosphate, 0.25 M sucrose, 10 g/ml soybean trypsin inhibitor, 2 g/ml aprotinin, pepstatin A, and leupeptin, pH 7.4. Slow speed centrifugation was at 470 ϫ g for 10 min, and the membranes were washed by resuspension in homogenization buffer plus 150 mM NaCl, 1 mM EDTA, and 1 mM EGTA before final resuspension in homogenization buffer. For binding studies membranes were prepared in Tris-HCl buffer without the NaCl wash step.
For solubilization, the membranes were diluted to 6 mg/ml protein * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and mixed with an equal volume of homogenization buffer lacking sucrose but containing 1% nonyl glucoside, 2 mM EDTA, and 2 mM EGTA. The mixture was stirred for 1 h and then centrifuged at 120,000 ϫ g for 60 min. The supernatant was stored at Ϫ80°C for use in binding and preliminary purification studies or filtered through Millipore AP25 prefilters immediately prior to use in full purification studies.
The eluate fraction was thawed and dialyzed against 1 mM sodium phosphate, 0.5% nonyl glucoside, pH 7.0, over a 2-h period. The fraction was then applied at 1 ml/min to a 4 ϫ 1.6-cm hydroxylapatite column previously equilibrated with dialysis buffer. The column was washed (60 ml) and eluted at 0.5 ml/min in dialysis buffer with a linear 0 -15 mM MgCl 2 gradient (30 ml) and a 15 mM MgCl 2 plateau (10 ml). Remaining proteins were removed by successive washes with 0 mM MgCl 2 (6 ml), 300 mM sodium phosphate, pH 7.0 (15 ml), and 500 mM sodium phosphate, pH 7.0 (15 ml at room temperature).
Proteolytic Digestion of the Murine A77 1726 Binding Site and Internal Sequence Analysis-Direct N-terminal sequence analysis on the purified protein was not possible, and internal sequencing was thus undertaken. The purified A77 1726 binding site was concentrated by chloroform methanol extraction (12) and resuspended in 10 mM Tris-HCl, 0.05% SDS, pH 7.5. The sample was acidified with trifluoroacetic acid (0.1%) and subjected to reverse phase chromatography on an RP-300 1 mm ϫ 100-mm column. The column was eluted with a 9-ml linear gradient from 0.1% trifluoroacetic acid to 50% acetonitrile/0.085% trifluoroacetic acid at 50 l/min. The peak containing the A77 1726 binding site was identified by SDS-PAGE, dried, and digested with 0.24 g of endoproteinase Asp-N for 2 h at 37°C in 10 mM Tris-HCl, 0.025% SDS, pH 7.5. The mixture was acidified with 1% trifluoroacetic acid and the peptides separated by a repeat of the reverse phase separation procedure.
Of the 23 peptide peaks collected, 5 were dissolved in 40% acetonitrile and applied to Polybrene-treated fiberglass filters. However, no phenylthiohydantoin-derivatives were released during attempted sequencing. The remaining 18 samples were dissolved in 0.1% trifluoroacetic acid and then 10 mM Tris-HCl, 0.05% SDS and applied to polyvinylidene difluoride membranes (Prospin, Applied Biosystems) with sequential 80-l washes of 10 mM Tris-HCl, 0.05% SDS, pH 7.5 (twice), and 20% methanol (three times). The peptide on each membrane was sequenced directly in a model 473A microsequencer cartridge (Applied Biosystems).
Photoaffinity Labeling of Mouse Spleen Preparations-Samples were incubated on ice for 1 h in the dark with 0.2-1 nM [ 125 I]RU35072 (see Structure 2) in 25 mM Tris-HCl, 25 mM MgCl 2 , pH 7.4. High affinity binding was defined by competition with 0.1-10 M A77 1726 or its analogue HR325 (compound 4 in Table I). The samples were then irradiated for 5 min at 24 cm with a Mineralite UVG 254 lamp. Noncovalently bound photolabel was removed by dilution with 3 mM HR325 and PD-10 desalting.
Electrophoretic transfer onto Immobilon-P was carried out on a Multiphor II/Novablot semidry system at a constant current of 2 mA/cm 2 for 60 min in 48 mM Tris-HCl, 39 mM glycine, 0.0375% SDS, 20% methanol or 90 min in a CAPS buffer system (13).
A77 1726-binding Assay and Data Analysis-Membranes or soluble fraction were incubated with 10 or 25 nM [ 3 H]A77 1726 in 10 or 25 mM Tris-HCl, 25 mM MgCl 2 , pH 7.5. Final assay volumes were 0.5 ml, and nonspecific binding was defined with 3 mM A77 1726. Incubations were allowed to reach equilibrium on ice (2 h for membranes and 16 to 24 h for soluble fractions). Assays were terminated by filtration through Whatman GF/F filters with two 8-ml washes with assay buffer. Filters were dried and counted by liquid scintillation. The [ 3 H]A77 1726 binding characteristics and binding pharmacology of membranes and soluble preparation were determined in competition assays. Competing ligand concentrations increased in 1 ⁄3 log unit intervals, and each concentration was assayed at least in quadruplicate. Binding data were analyzed with the EBDA and LIGAND programs (14). In purification studies high affinity binding was defined as that displaceable by 100 nM A77 1726, although this would underestimate the true level slightly.
Dihydroorotate Dehydrogenase Assay-Inhibition of mouse spleen dihydroorotate dehydrogenase activity by test compounds was assessed using a DCIP-linked assay (15). Membranes (0.05 mg of protein) were incubated with 100 M coenzyme Q 10 in 50 mM Tris-HCl, 0.1% Triton X-100, 1 mM KCN, pH 8.0. The reaction was initiated by addition of 500 M dihydroorotate, and the reduction of DCIP (200 M) was monitored by loss of absorbance at 650 nm using a 96-well plate reader at 37°C. Drug concentrations increased in 1 ⁄2 log unit intervals with each concentration tested at least in triplicate. Fractions from purification experiments were assayed in a similar manner, except that the buffer did not contain KCN, and absorbance changes were monitored in a spectrophotometer at 600 nm.
LPS-stimulated Mouse Spleen Cell Proliferation-LPS (12 g/ml)stimulated spleen cells (3 ϫ 10 5 cells/ml) from C57BL/6 mice were incubated for 3 days with increasing A77 1726 concentrations in supplemented RPMI 1640 medium (4 mM L-glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 50 M 2-mercaptoethanol, antibiotics, and 10% fetal calf serum) in the presence or absence of uridine (30 M). Tritiated thymidine incorporation (1 Ci/well) was measured during the final 6 h of the incubation and used as a measure of cell proliferation.
Delayed Type Hypersensitivity (DTH) Test-The method used was based on published information (16). Briefly, on day 0 male CD-1 mice (25-30 g, Charles River) were sensitized by subcutaneous injection near the tail base with 1 mg of methylated bovine serum albumin in 0.2 ml of saline/Freund's complete adjuvant emulsion. On day 7 mice were challenged by subplantar injection into the hind paw with 0.1 mg of methylated bovine serum albumin in 0.05 ml of saline. The DTH paw edema was expressed as the increase in paw weight after 24 h relative to non-sensitized saline/Freund's complete adjuvant-injected mice. Drugs were administered orally once per day on days 4 -6 and twice on day 7, 1 h before and 6 h after challenge.
Protein Determination-Protein concentrations were generally determined using the bicinchoninic acid method (Pierce). The protein concentration in the chromatofocusing eluate was estimated from area under the curve calculations on the absorbance profile (280 nm) of the step.

Characterization of the Binding Sites in Mouse Spleen Membranes and Soluble Preparations-Binding sites for [ 3 H]A77
1726 in mouse spleen membranes were examined in homologous competition assays. The binding isotherms gave best fits to a three-site model (p ϭ 0 versus two-site model) with high (12 nM), moderate (3 M), and low (240 M) affinity sites detected (Table I, part a). The high affinity site represented ϳ75% of the binding curve, whereas the low affinity site was probably saturable nonspecific binding.
The binding characteristics of a series of A77 1726 analogues were tested in competition binding studies. The rank order of potency of the compounds for affinity at the high affinity site was similar to that for the potency of inhibition in the mouse DTH assay (Table I, part b). Compounds 2 and 6 appeared to have higher affinity than expected from their DTH potency; however the differences were small in view of the in vitro/in vivo nature of the comparison.
The high affinity binding site was solubilized with 0.5% nonyl glucoside (45% yield: Table I, part a). The site showed maintained affinity for A77 1726. The moderate affinity site was not solubilized in a detectable form.
Purification of the High Affinity Binding Site-Sequential chromatography on SP-Sepharose HP, hydroxylapatite, and chromatofocusing columns afforded high affinity binding site yields of 65%, 41%, and 72%/step, respectively. Additional losses were observed on the dialysis and desalting steps used to prepare eluates for further chromatography, the overall yield of the procedure was thus 14% with 1988-fold purification (Table  II). The high affinity site started to elute from the chromatofocusing column at pH 9.8, consistent with a preliminary experiment where the site eluted between pH 9.79 and 9.58.
Characterization of the Purified High Affinity Binding Site-The chromatofocusing eluate contained a single band of M r 43,000 on SDS-PAGE following electrotransfer to an Immobilon-P membrane and Coomassie staining (Fig. 1, lane 1). Silver staining was not attempted as an earlier experiment indicated that the M r 43,000 band stained very poorly. The M r 43,000 protein was identified as the high affinity site in photoaffinity labeling studies using the iodoazido compound (Structure 2) on soluble preparation and eluates from small scale purification experiments (Fig. 1, lanes 2-5). Additional studies on a single-step chromatofocusing eluate and an SP-Sepharose-hydroxylapatite eluate, prepared with phosphate rather than magnesium gradient elution, showed complete and almost complete protection, respectively, of the M r 43,000 band by 100 nM HR325 (data not shown).
An attempt at obtaining N-terminal sequence from the M r 43,000 band was unsuccessful. In order to obtain internal sequence, the purification procedure was scaled up 2-fold, yielding 210 pmol of high affinity site, 1% of which produced a faint single band of M r 43,000 on SDS-PAGE and Coomassie staining. The eluate was subjected to reverse phase high performance liquid chromatography to remove detergent-derived nonprotein contaminants, which would prevent peptide detection, together with any minor contaminating proteins not detected on the Coomassie-stained gel. The M r 43,000 protein was subjected to endopeptidase digestion (Asp-N), with the resulting peptides separated by an additional round of reverse phase chromatography. Sequence data was obtained for 7 of the 18 peptides applied to polyvinylidene difluoride membranes (Fig.  2). A search of the Swissprot data bank revealed that only human dihydroorotate dehydrogenase (DHO-DH) had amino acid sequence similarity with all the sequenced peptides. For the five peptides that could be unambiguously aligned with the human sequence, 35 out of 43 assigned amino acids were iden-tical (Fig. 2). A greater degree of similarity was observed when the peptides were compared with rat DHO-DH sequence (EMBL data base, accession no. X80778) with 39 out of 43 residues identical. The remaining two pentapeptides could be in DTH and DHO-DH assays a, A77 1726 binding characteristics of three membrane preparations (mean Ϯ S.D.) and a single soluble preparation each assayed once (see "Experimental Procedures" for detail). The capacity of the moderate and low affinity sites would be underestimated by the assay due to rapid dissociation of bound radioligand during filtration (18), data on the capacity of these sites are thus included for completeness only. The soluble preparation displayed two [ 3 H]A77 1726 binding sites; the lower affinity site was aligned with site 3 of the membrane preparation on the basis of comparable capacity and affinity. b, binding affinities of test compounds were measured in competition binding assays (single determinations); K d estimates at the lowest affinity site are not presented. In each case, this site had lowest affinity for the displacing ligand. For DTH assays compound data refer to inhibition at a drug dose of 100 mg/kg unless otherwise specified by the value in parentheses (mg/kg). Each percentage inhibition value represents a result from a single experiment. IC 50 values in the DHO-DH assay are either from single determinations or are mean Ϯ range (n ϭ 2). All methods were as described under "Experimental Procedures." Compound 3 ϭ A77 1726 (leflunomide was tested in vivo rather than its active metabolite A77 1726, which was used in all in vitro assays), compound 4 ϭ HR325. *, p Ͻ 0.05; **, p Ͻ 0.01 (Student's t test, n ϭ 7-8).
aligned with several possible sequences within the human and rat proteins, with 3 of the 5 residues being identical in each case.
Further Evidence Supporting the Identity of the High Affinity Site as DHO-DH-Fractions from a series of purification experiments were analyzed for DHO-DH activity using the DCIP assay. In all cases DHO-DH activity and high affinity binding activity co-purified (data not shown).
A77 1726 analogues displayed similar rank order of potency for affinity at the high affinity site (K d ) and IC 50 for enzyme inhibition (correlation coefficient 0.986 for compounds in Table  I with defined IC 50 values).
Involvement of DHO-DH in the Antiproliferative Activity of A77 1726 -The antiproliferative effect of A77 1726 on LPSstimulated mouse spleen cells was examined in the presence and absence of exogenous uridine (30 M). The uridine completely overcame the antiproliferative activity of the compound (Fig. 3).
Involvement of DHO-DH in the in Vivo Activity of the A77 1726 Class of Compounds-The role of DHO-DH inhibition in vivo was directly demonstrated in the acute murine graft versus host reaction. B6C3F1 hybrid mice, from C57BL/6 and C3H/He parental strains, demonstrated marked splenomegaly following intraperitoneal administration of C57BL/6 splenocytes (Table III). The splenomegaly, a hallmark of the response (17), was markedly inhibited by HR325 in the absence but not in the presence of administered uridine (74 and 19% inhibition, respectively).

DISCUSSION
The aim of the current study was to isolate and characterize the target protein that mediates the effects of A77 1726, the active metabolite of leflunomide. A potential target protein was identified in mouse spleen membranes using a radioligand binding approach (Table I). Although the membranes carried at least three binding sites, the high affinity site (K d 12 nM) was of most interest as its binding pharmacology for A77 1726 analogues was qualitatively similar to that for the mouse DTH response in vivo (Table I). The DTH assay was used as a primary screen to detect active compounds; as such, dose ranges were limited, preventing quantitative correlation between binding and potency. However, the qualitative relationship, which extended to over 70 compounds, 2 supported the role of the high affinity site in drug action in vivo.
The purified site was unequivocally identified as DHO-DH by a series of structural and functional criteria. There was high microsequence identity between human enzyme (81%: Ref. 19) or rat enzyme (91%: EMBL data base) and pure peptides derived from the high affinity site (Fig. 2). In addition, the rela-tive molecular weight of 43,000 for the site on SDS-PAGE ( Fig.  1) was close to that of the bovine liver enzyme (M r 42,000; Ref. 20) and that predicted for human DHO-DH (43.0 kDa; Ref. 19). The weak silver staining of the mouse protein is consistent with a protein lacking cysteine residues, as is the case with the human enzyme (19). A77 1726 binding activity and dihydroorotate dehydrogenase activity co-purified through all steps of the purification procedure. Finally, binding affinity of a series of A77 1726 analogues correlated with enzyme inhibitory potency of the compounds.
Dihydroorotate dehydrogenase (EC 1.3.3.1) catalyzes conversion of dihydroorotate to orotate, the fourth step in de novo  2a, 3a, 4a, and 5a). Parallel incubations, in which the photoaffinity ligand was prevented from binding to the high affinity site by 10 M HR325 (lane 2b) or 1 M A77 1726 (lanes 3c, 4c, and 5c), were used to identify the site against a possible background of low affinity sites. After photolysis the non-covalently bound photolabel was removed by dilution with 3 mM HR325 and PD10 desalting. The binding characteristics of the photoaffinity label were examined in preliminary studies carried out in the dark using the non-radioactive ( 127 I) form of the compound. The photoaffinity ligand had a K d of 29.1 nM for the mouse spleen high affinity site in a [ 3 H]A77 1726 competition binding study. The reversible nature of the binding interaction before photolysis was further established in an experiment in which membranes were incubated for 2 h with 235 nM photoaffinity ligand. Although this concentration should occupy Ͼ85% of the high affinity sites, more than 98% of the original binding in the membranes was detected following dilution of the membranes into a [ 3 H]A77 1726 binding assay (5 nM final photoaffinity ligand concentration). Irreversible binding of the radioactive photolabel to mouse spleen membranes was not observed in the absence of photolysis (data not shown). Molecular size markers are indicated in full for 1 and 2, by position for 3 and by position for M r 45,000 and 36,000 markers only for 4 and 5. The M r 43,000 bands in the different gels are aligned. biosynthesis of pyrimidine nucleotides (21). Inhibition of dihydroorotate dehydrogenase activity by A77 1726 and its analogues would clearly have an anti-proliferative effect on a wide range of cells due to depletion of intracellular pools of UTP and CTP, as is the case for other known inhibitors of pyrimidine biosynthesis (22,23). That this is solely responsible for the anti-proliferative effects of the compound is indicated by maintained proliferation in LPS-stimulated mouse spleen cells receiving exogenous uridine during A77 1726 treatment (Fig. 3). Similar effects have been observed with other inhibitors of pyrimidine biosynthesis (22,23). The ability of uridine to protect mice from inhibition of the acute graft versus host response (Table III) directly demonstrates the role of DHO-DH as mediator of the in vivo effects of the A77 1726 class of compounds in this system. At least one other in vivo immunosuppressive effect of the compounds may be mediated by inhibition of de novo pyrimidine biosynthesis, since inhibition of the DTH response in mice shows a qualitative relationship with binding affinity and with inhibition of DHO-DH activity ( Table I).
The possibility that the A77 1726 class of compounds may have additional effects in vivo or in vitro cannot be ruled out by the current studies. However, some of the previously reported effects of long term treatment of cells with the compounds could be secondary to pyrimidine depletion. For example, inhibition of epidermal growth factor-dependent tyrosine kinase activity in intact cells (24) was demonstrated after a 4-day treatment with A77 1726, whereas a 30-min treatment had no effect (7). However, proof that this effect reflects pyrimidine nucleotide depletion awaits uridine reversal studies.
In conclusion, the A77 1726 high affinity binding protein in murine spleen membranes, proposed as a putative target for A77 1726 action, has been isolated and identified as dihydroorotate dehydrogenase. The enzyme catalyzes the fourth step in de novo pyrimidine biosynthesis and its inhibition accounts for the antiproliferative effects of the compounds in vitro and some of the in vivo effects of the compounds.

TABLE III
Effect of HR325 in the presence or absence of uridine on the 10-day acute murine graft versus host response The experiment was carried out as described under "Experimental Procedures." B6C3F1 mice received C57BL/6 splenocytes (day 0), controls received RPMI 1640 vehicle. HR325 was administered daily at 50 mg/kg (days 0 -3), uridine was administered twice daily (at t ϭ 0 and ϩ6 h relative to HR325). Splenomegaly (spleen weight/body weight ratio) was assessed on day 10 (body weight was not significantly affected by any treatment). Values are mean Ϯ S.D. *, p Ͻ 0.005 (Student's t test; n ϭ 7-11). A repeat study indicated 57% inhibition of splenomegaly by HR325, which was reduced to 6% in uridine treated animals.
FIG. 2. Amino acid sequence homology between the mouse spleen high affinity A77 1726 binding site and human and rat dihydroorotate dehydrogenase. Amino acid sequence of peptides isolated from Asp-N digestion of the purified high affinity binding site (see "Experimental Procedures"). Two peptides from the mouse could not be lined up with human or rat sequence as they produced partial matches (3 out of 5 amino acids) with several sequences within the human and rat proteins. ?, no assignment possible; * highlights sequence differences; ‡, T or P; ϩ, L or V.