Structural, Kinetic, and Pharmacodynamic Mechanisms of d-Amino

Structural, Kinetic, and Pharmacodynamic Mechanisms of d-Amino...

0 downloads 60 Views 3MB Size

Article pubs.acs.org/jmc

Structural, Kinetic, and Pharmacodynamic Mechanisms of D‑Amino Acid Oxidase Inhibition by Small Molecules Seth C. Hopkins,*,† Michele L. R. Heﬀernan,† Lakshmi D. Saraswat,† Carrie A. Bowen,† Laurence Melnick,† Larry W. Hardy,† Michael A. Orsini,† Michael S. Allen,⊥ Patrick Koch,† Kerry L. Spear,† Robert J. Foglesong,# Mustapha Soukri,# Milan Chytil,† Q. Kevin Fang,† Steven W. Jones,† Mark A. Varney,† Aude Panatier,‡ Stephane H. R. Oliet,‡ Loredano Pollegioni,§,∥ Luciano Piubelli,§,∥ Gianluca Molla,§,∥ Marco Nardini,∞ and Thomas H. Large† †

Sunovion Pharmaceuticals Inc., Marlborough, Massachusetts 01752, United States Neurocentre Magendie, Inserm U862 and Université de Bordeaux, Bordeaux, F-33077, France § Dipartimento di Biotecnologie e Scienze della Vita, Università degli Studi dell’Insubria, Via J. H. Dunant 3, 21100 Varese, Italy ∥ The Protein Factory, Politecnico di Milano, ICRM-CNR and Università degli Studi dell’Insubria, Via Mancinelli 7, 20131 Milano, Italy ⊥ Tandem Labs, Durham, North Carolina, United States # Scynexis, Durham, North Carolina, United States ∞ Department of Biosciences, University of Milan, I-20133 Milano, Italy ‡

S Supporting Information *

ABSTRACT: We characterized the mechanism and pharmacodynamics of ﬁve structurally distinct inhibitors of D-amino acid oxidase. All inhibitors bound the oxidized form of human enzyme with aﬃnity slightly higher than that of benzoate (Kd ≈ 2−4 μM). Stopped-ﬂow experiments showed that pyrrole-based inhibitors possessed high aﬃnity (Kd ≈ 100−200 nM) and slow release kinetics (k < 0.01 s−1) in the presence of substrate, while inhibitors with pendent aromatic groups altered conformations of the active site lid, as evidenced by X-ray crystallography, and showed slower kinetics of association. Rigid bioisosteres of benzoic acid induced a closed-lid conformation, had slower release in the presence of substrate, and were more potent than benzoate. Steady-state D-serine concentrations were described in a PK/PD model, and competition for D-serine sites on NMDA receptors was demonstrated in vivo. DAAO inhibition increased the spatiotemporal inﬂuence of glial-derived D-serine, suggesting localized eﬀects on neuronal circuits where DAAO can exert a neuromodulatory role.

■

serine can inﬂuence NMDA neurotransmission.4−10 It was recently demonstrated that D-serine gates preferentially synaptic NMDARs, while glycine favorably acts on the extrasynaptic ones, this matching their respective aﬃnities.11 The details of Dserine’s spatiotemporal distribution and its dynamic control in neuronal function are still uncertain. D-Amino acid oxidase (EC 1.4.3.3, DAAO) is a prototypical ﬂavin-dependent oxidoreductase that has been the subject of over 70 years of biochemical investigations.12,13 DAAO is capable of remarkably stereoselective oxidation of neutral Damino acids and has found industrial applications in the selective oxidation of D-isomers of racemic mixtures.13 First thought to be important for oxidizing D-amino acids of dietary and bacterial origin, its presence in mammalian brains was long

INTRODUCTION

N-Methyl-D-aspartate receptors (NMDARs) participate in synaptic transmission to mediate the neuronal plasticity important for regulation of memory, nociceptive behaviors, mood, addiction, motor function, and neuronal development.1−3 Unique among ligand-gated ion channels, activation requires the simultaneous binding of diﬀerent agonists at two distinct sites on the channel. Glutamate (or exogenously applied NMDA) binds the NR2 subunit, while glycine or Dserine binds the NR1 subunit. In addition, NMDAR activation requires membrane depolarization to relieve the Mg2+ ion block of the channel pore. Under normal conditions, the coincidence of these three disparate signals activates NMDARs to allow Ca2+ entry, thus initiating the long-term changes associated with synaptic plasticity. Together these molecular signals exhibit properties of coincident detection important for associative learning. Accumulated evidence suggests that D© XXXX American Chemical Society

Received: February 19, 2013

A

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Table 1. Static and kinetic constants of hDAAO binding and inhibition by small molecules

A

Kd and SE values obtained from nonlinear regression of classical saturation of absorbance changes. BIncreases in absorbance above 500 nm indicates the formation of a charge-transfer (CT) complex between the aromatic nitrogen of the inhibitor and the ﬂavin isoalloxazine ring of FAD. C Kinetic measurements derived from stopped-ﬂow apparatus, with ±SE values (from at least ﬁve determinations) based on goodness of ﬁt to exponentials. DCoupled-enzyme assay in the presence of 5 mM D-serine substrate, with ±SD (number of independent determinations). EOxygen detection at either 5 or 50 mM substrate, with SE values of nonlinear regression of inhibition curves. FKi values from Figure 1, with SE values based on global ﬁtting Michaelis−Menten curves.

evaluating the in vivo potency of inhibition and for understanding its relationship to steady-state D-serine concentrations, we characterized the dynamics of brain D-serine concentrations with a PK/PD model of D-serine production and degradation by DAAO activity. We demonstrated that DAAO inhibition was suﬃcient to reduce falling behavior following adminstration of D-serine-site NMDAR antagonist (L-701,324). Lastly we presented evidence that NMDAR currents increased by a DAAO inhibition occurred via glial-derived D-serine. The chemical structures of the ﬁve inhibitors are shown in Table 1.

considered vestigial (for review about the physiological role of DAAO see ref 12) until the combined discoveries of D-serine14 and serine racemase15 in the brains of mammals. A mouse strain lacking active DAAO has elevated cerebellar D-serine,16 altered NMDAR function,17−19 and sensitivity to MK-801,20 aﬀecting behaviors related to cognition21 and anxiety.22 These observations suggest that DAAO activity plays an important role in regulation of NMDAR neurotransmission. DAAO activity may physiologically limit NMDAR activity by decreasing the local concentrations of D-serine within the astrocytic environment of neurons and synapses in deﬁned brain regions.23,24 An inverse relationship between DAAO activity and D-serine levels emerges in neonatal rat cerebellum at 3 weeks,25,26 highlighting the role of Bergman glia-derived Dserine in guiding the migration of granule cells via activation of NMDARs.27,28 This ontological sequence may also be relevant in humans.29 The therapeutic potential for small molecule inhibitors to treat disorders like schizophrenia or neuropathic pain requires a better understanding of pharmacodynamics and mechanism of DAAO inhibition and of its physiological role in NMDAR function. Known human DAAO (hDAAO) inhibitors have recently been reviewed.30 Toward this end, we examined structurally distinct inhibitors of DAAO to outline their molecular and kinetic mechanisms in vitro and in vivo. In recent years hDAAO was biochemically and structurally characterized.31,32 Here, we compared ﬁve inhibitors of DAAO for binding, inhibition, and kinetics of association and dissociation under comparable assay conditions and using highly pure human enzyme, as well as detailed their binding mode by X-ray crystallography. Important for

■

RESULTS In Vitro Inhibition and Interaction with hDAAO. Pyrrole-2-carboxylic acid33 and indole-2-carboxylic acid34 have been described as DAAO inhibitors with moderate potency (Kd of 26 μM and Ki of 3.4 μM, respectively). A pyrrole-2carboxylic acid containing a fused thiophene moiety 1 (SEP641)35 was identiﬁed in the course of in silico modeling and in vitro testing to have improved potency (pIC50 = 8.27 M, pIC50 = −log IC50 in M) and to be competitive with substrate (Ki = 3.5 ± 0.17 nM) (Figure 1). The inhibition of hDAAO was demonstrated by both a coupled enzyme assay and a direct measurement of oxygen consumption. As ﬁrst described for benzoic acid,36 all inhibitors tested were competitive with substrate as evident by the increase in IC50 values with increased substrate concentrations (Table 1). Ki values were determined using the coupled enzyme assay and global ﬁtting of a classical Michaelis−Menten competitive inhibition model (Figure 1, global R2 of 0.99 and above). For comparison, the eﬀect of the classical inhibitor benzoate was also investigated. B

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 1. Inhibition of recombinant hDAAO by tested compounds demonstrates competition with D-serine substrate. DAAO activity was measured in the presence of increasing concentrations of inhibitor and substrate and using a coupled assay. A Michaelis−Menten model of competitive inhibition was globally ﬁt according to an adjustable parameter Ki (see eq 1a and eq 1b). Compound numbers and structures are indicated, along with the estimated Ki and Km values in each panel.

active site size,38 and R. gracilis DAAO where Tyr238 plays the same role.37 To take advantage of this plasticity, a 4-substituted pyrrole-2-carboxylic acid was proposed by computer docking to occupy this region, above the plane of the FAD, occupied by the indole side chain in the iminotryptophan−pkDAAO complex. The synthesized compound 2 (SEP137)40 was found to be a potent inhibitor (pIC50 = 7.94 M) competitive with substrate (Ki = 7.2 ± 0.42 nM). The crystal structure of hDAAO in complex with compound 2 (2.3 Å resolution, Figure 2A, Supporting Information Table S1) showed that the pyrrole2-carboxylate scaﬀold was superimposed with that of compound 1 (see comparison in Figure 2B) and that the ligand side chain, pointing toward the entrance of the active site, is located in a pocket lined by Leu51, Gln53, Leu56, Ile215, His217, Tyr224, Tyr228, and Ile230 side chains. Noteworthy, the electron density data for this complex were in agreement with at least two (similar) conformations of the ligand side chain. The presence of two ligand conformations may relate to the two slightly diﬀerent aﬃnities described below. The same phenethyl substitution of 2 was introduced on a scaﬀold similar to 1, obtaining a relatively weaker (pIC50 = 6.66 M) inhibitor compound 3 (SEP371).41 Examination of the Xray structure of compound 3 bound to hDAAO (2.75 Å resolution, Figure 2A, Supporting Information Table S1)

The crystal structure of compound 1 bound to hDAAO was solved at high resolution (1.9 Å, Figure 2, Supporting Information Table S1). Analysis of the 3D structure showed a hydrogen bond between the pyrrole nitrogen and the backbone carbonyl oxygen of Gly313, the ion-pairing of the carboxylate with the guanidinium group of Arg283 and the hydrogen bond with the OH group of Tyr228. The furan moiety, containing a sulfur atom, is placed in a hydrophobic pocket of the active site in close contact (3.5−4.0 Å) with the side chains of several apolar amino acids (Leu51, Leu215, and Ile230), and it is positioned approximately parallel to the plane of the isoalloxazine ring of FAD; the two groups interact via πstacking interactions. Tyr224 covers the carboxylate moiety of compound 1 and loosely π-stacks with the solvent exposed face of the inhibitor (Figure 2A). A peculiar feature of DAAO is the plasticity of its active site.37−39 Despite the relatively small size of the ligand binding pocket that limits the chemical features available for the design of candidate inhibitors, the region of the active site that accommodates side chains of neutral D-amino acid substrates possesses a degree of plasticity suﬃcient to allow the enzyme to accommodate bulky side chains. This feature has previously been demonstrated in pig kidney DAAO (pkDAAO) in complex with imino-tryptophan where the active site lid (and in particular Tyr224) changed its conformation to augment the C

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 2. Inhibitors bound in hDAAO active site. (A) Comparison of the active site of hDAAO in the free form (PDB code 2E48) in complex with benzoate (PDB code 2DU8) and selected inhibitors (compounds 1, 2, 3, and 5). Relevant residues lining the inhibitor binding pocket are indicated and shown in stick representation (green). For visual clarity, some residues have been omitted (i.e., Gln53, Leu56, His217). The inhibitor molecules and the FAD cofactor are shown in cyan and yellow, respectively. H-bonds are indicated by dashed lines. (B) Overlay of the positions occupied by the inhibitors in the hDAAO active site, highlighting the alternative conformations occupied by Tyr224 side chain.

inhibitor isoxazole ring. More systematic searches for bioisosteres were conducted in silico using commercial compound libraries. Rapid overlays of chemical structure (ROCS) were followed by calculations of surface electrostatics (EON) enriched for candidate bioisosteres of carboxylic acid. Compound 5 (SEP064)43 was identiﬁed as a potent (Ki = 13 ± 1.4 nM, pIC50 = 7.46 M) inhibitor containing two exocyclic oxygens. The corresponding 3D structure (2.4 Å resolution, Figure 2A, Supporting Information Table S1) shows compound 5 bound in a sandwich between the FAD isoalloxazine ring and the Tyr224 side chain, in which its exocyclic oxygen atoms Hbonded to the guanidyl nitrogens of Arg283 and the Tyr228OH group. In Vitro Interaction with hDAAO. The binding aﬃnity of each inhibitor was estimated by titration of the oxidized form of the enzyme and by following the ligand-induced changes in the visible absorption spectrum of the FAD cofactor, in the absence

showed that the short ethylene linker atoms occupied disparate regions of the active site in the two hDAAO complexes, even though the aromatic moieties of the two ligands occupy the same site (Figure 2B). This adjustment preserved the compound 3 ligand’s pyrrole hydrogen bond with Gly313 (2.78 Å) similar to that found for compound 2 (2.76 Å). To address the perceived liabilities of a carboxylic acid on brain penetration, we sought an inhibitor lacking the carboxylic acid moiety. A bioisosteric replacement of a carboxylate with a benzo[d]isoxazole-3-ol was identiﬁed by searching for structures capable of electrostatic interactions with the guanidinium group of Arg283. We identiﬁed compound 4 (SEP342,42 pIC50 = 8.51 M, Table 1) to be competitive with substrate (Ki = 1.7 ± 0.059 nM). In the absence of the crystal structure of the hDAAO−4 complex, molecular docking considerations suggest the presence of an ionic interaction between Arg283 and the nitrogen and exocyclic oxygen of the D

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 3. Binding and dissociation of diﬀerent inhibitors to hDAAO as determined following the changes in the ﬂavin absorbance spectrum. (A, B) Modiﬁcation of the absorption spectrum of hDAAO following the addition of increasing amounts of diﬀerent inhibitors. (A) Absorption spectra of hDAAO following addition of saturating concentrations of compound 2 (thick continuous line), compound 1 (thin dashed line), and compound 4 (thick dashed line). As comparison, the spectrum of free hDAAO is also reported (thin continuous line). (B) Diﬀerential absorption spectra observed upon addition of increasing concentrations of compound 2 (top, spectrum 1, 0.04 μM; spectrum 2, 1 μM; spectrum 3, 3.8 μM; spectrum 4, 7.3 μM; spectrum 5, 47 μM), compound 1 (center, spectrum 1, 0.04 μM; spectrum 2, 0.4 μM; spectrum 3, 3 μM; spectrum 4, 12 μM; spectrum 5, 94 μM), and compound 4 (bottom, spectrum 1, 0.5 μM; spectrum 2, 2 μM; spectrum 3, 3 μM; spectrum 4, 5.7 μM; spectrum 5, 7.5 μM; spectrum 6, 12 μM). (C) Determination of the dissociation constants of the selected inhibitors to hDAAO, as reported in panels A and B: plots of absorbance E

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 3. continued changes at diﬀerent wavelengths versus inhibitor concentrations (absolute absorbance values). (D) Investigation of the binding and dissociation process of inhibitors to hDAAO as determined by stopped-ﬂow analysis. Left: time courses of the absorbance intensity (at diﬀerent wavelengths, see text) after mixing hDAAO (∼10 μM) with a comparable concentration (8−10 μM) of diﬀerent inhibitors. For a better comparison, absorbance changes are reported as percentage of the total change observed: trace 1, benzoate (at 491 nm); trace 2, compound 5 (at 495 nm); trace 3, compound 4 (at 497 nm); trace 4, compound 1 (at 521 nm); trace 5, compound 2 (at 491 nm); trace 6, compound 3 (at 493 nm). Inset: plot of kobs versus inhibitor concentration for compound 3 (open circles) and compound 4 (closed circles). The rate constants were determined by ﬁtting the experimental traces as reported in the main panel according to a monoexponential process. Right: time courses of the absorbance at 450 nm after 1:1 vol/vol anaerobic mixing of hDAAO−inhibitor complex ([hDAAO] = 18−20 μM and [inhibitor] = 40 μM) with 200 mM D-serine. Absorbance changes are reported as percentage of the residual oxidized enzyme. Numbering of the traces is the same as in left panel. For comparison, trace obtained in the same conditions without any inhibitor added is also reported (trace 7) which shows the conversion of oxidized enzyme into the corresponding reduced form in the dead time of mixing.

therefore deﬁned a k release parameter sensitive to the dissociation kinetics of the ligand from the hDAAO−inhibitor complex in the presence of substrate. Here the oxidized enzyme, premixed with a saturating concentration of inhibitor, was rapidly reacted with an excess of substrate D-serine under anaerobic conditions. Under these conditions, only the free holoenzyme form (such as that obtained following the dissociation of inhibitor) can be reduced by the substrate. This process was followed at 450 nm as a reduction of the ﬂavin cofactor. As shown in Figure 3D right panel, benzoate was the fastest dissociating inhibitor (trace 1), followed by its bioisostere compound 5. The compounds with occupancy of the side chain binding pocket (compounds 2 and 3) were next slowest, and the smallest compounds had no measurable dissociation during the time course of the experiment (compounds 1 and 4). In Vivo Competitive Inhibition. The baseline, steady-state concentrations of D-serine were measured across various brain regions obtained from rats and mice using a chiral LC/MS/MS method. This direct method eluted the D-enantiomer of serine prior to the more abundant L-enantiomer. The cerebellum concentration of D-serine in mice and rats was 2.3 ± 0.7 and 5.1 ± 1.3, respectively (nmol/g tissue wet weight ± sd, each n > 100). D-Serine concentrations did not vary according to time of day between naive animals and vehicle-treated animals, with type of vehicle, or with route of administration (data not shown). D-Serine concentration was much higher in forebrain regions (whole brains with brainstem and cerebellum dissected away) of rats (175 ± 20, n = 59) and mice (157 ± 13, n = 9) relative to cerebellum or brainstem concentrations. Plasma concentrations of D-serine were 1.72 ± 0.45 in rats (n = 160) and 1.37 ± 0.39 in mice (n = 310, average nmol/mL plasma ± sd). A simple model of regulation of D-serine concentration based on production (via serine racemase) and degradation (via DAAO) would predict increases in D-serine concentration with decreases in DAAO activity (inhibition). We therefore measured the time course of D-serine concentrations in cerebellum of mice after doses of each inhibitor. As evident in Figure 4A for compounds 1, 2, and 4 (12, 30, and 125 mg/kg ip, respectively), D-serine concentration increases at a similar initial rate of ∼2 nmol g−1 h−1, indicative of maximal inhibition at these early time points, and likely reﬂects the endogenous rate of D-serine synthesis. Higher doses of any inhibitor did not further increase the initial rate. We related the observed dynamics of D-serine concentration to inhibitor pharmacokinetics by also measuring brain concentrations of each inhibitor in the same samples (Figure 4B). In particular, of the three inhibitors tested the concentrations of compound 4 declined

of substrate. The spectra obtained with compounds lacking a pyrrole nitrogen (benzoic acid, compounds 4 and 5) do not induce changes in absorption above 500 nm, while compounds containing a pyrrole nitrogen (compounds 1, 2, and 3) induce a perturbation above 500 nm indicative of a charge-transfer complex formation (see Figure 3A and B). The absorption changes saturate with increasing ligand concentration to deﬁne Kd values (see Table 1 and Figure 3C). Compounds 2 and 1 were distinguished by apparently two binding processes of diﬀering aﬃnities (Figure 3B and Figure 3C), as evidenced by the absorption changes at 490 nm (high aﬃnity) vs the changes at 520 nm (low aﬃnity). The other compound with a pyrrole nitrogen, compound 3, also gives a spectral perturbation at 490 nm (not shown), but this aﬃnity is weaker (3.7 μM) than its pyrrole-based analogues (see Table 1). Subsequently, the kinetics of binding was monitored using a stopped-ﬂow apparatus. By rapid mixing of the inhibitor and hDAAO, the time course of complex formation was followed at the same wavelengths reported above in static titration experiments. The time courses were ﬁt satisfactorily according to a monoexponential change. A normalized comparison of the observed absorbance changes during binding is shown in Figure 3D left, at ∼10 μM inhibitor concentration. For compound 1, the time course at 490 nm did not ﬁt satisfactorily because of the relatively low changes in absorption and high scattering at this wavelength. The observed rate constants (kobs = kon + koff) plotted vs inhibitor concentration yield a linear (pseudo-ﬁrstorder) behavior, as shown in the inset of Figure 3D. The slope of the straight line corresponds to the rate constant for ligand binding (kon), and the y-intercept yields the rate for dissociation (koff).44 The rates of inhibitor binding ranged from 2 × 105 to 8 × 105 M−1 s−1 for all compounds (Table 1) with the notable exception of compound 3 which binds relatively slowly (kon < 1 × 105 M−1s −1). This slower kon value is compensated by a slower koff value of 0.48 ± 0.22 s−1, thus yielding a Kd value (∼6 μM, as calculated by the koff/kon ratio) similar in magnitude to that estimated by static titration (3.7 ± 0.5 μM). These kinetic experiments further indicate that the binding of compound 2 to hDAAO is at least a two-step process. The binding at higher aﬃnity for compound 2, evident from static titration at 490 nm, is largely due to a slower oﬀ-rate (koff = 0.18 s−1) relative to the oﬀ-rate obtained from the lower aﬃnity absorption change at 520 nm (koff = 1.1 ± 0.5 s−1). In consideration of the pharmacodynamics of DAAO inhibition in vivo, where inhibitor concentrations decline according to its pharmacokinetics, we sought a parameter that would reﬂect each inhibitor’s potential for long-lasting inhibition of hDAAO and that might last beyond the time when tissue concentrations decline below the IC50 values. We F

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Menten constant of DAAO for D-serine, which we here call invivoKi. To examine the predictions of this simple model, we dosed each inhibitor to rats and sought doses and time points to approximate steady states in D-serine concentration. In rats, a single injection of compound 2 (125 mg/kg ip) caused a prompt and steady rise in D-serine concentration that reached a plateau by 12 h after administration (Figure 5A). Interanimal variability in exposure of compound 2 resulted in a range in Dserine concentrations reached 18, 24, and 48 h after compound 2 administration (arrows in Figure 5A and Figure 5B). As expected from the model of competitive inhibition, D-serine

Figure 4. D-Serine (A) and inhibitor (B) concentrations in cerebellum from mice sacriﬁced at diﬀerent times after administration of DAAO inhibitors. Triangles represent a single administration of compound 1 (12 mg/kg ip); squares, additional administrations every 2 h to maintain full inhibition (12 mg/kg ip at 2, 4, and 6 h); diamonds, compound 2 (30 mg/kg ip); circles, compound 4 (125 mg/kg ip). Initial rates of increase in D-serine concentrations (slopes of curves between 0 and 2 h) are ∼2 nmol g−1 h−1. By use of this value, red lines represent simulations of D-serine and inhibitor concentrations according to a PK/PD model of competitive inhibition with [Dserine] (Km = 5 mM), invivoKi values of 4 and 0.1 μM for compound 2 and compound 1, respectively, and ﬁrst-order pharmacokinetics (see eq 6 in Supporting Information). Repeated doses of compound 1 were simulated as steady concentrations of 2 μM inhibitor and produced the steady increases shown for [D-serine]. The weaker invivoKi value of 2 allows steady-state [D-serine] to be achieved by 6 h, with a return to baseline by 24 h (data not shown).

most rapidly, and after 3 h its concentration fell below 1 μM. This occurred at a time point when D-serine abruptly stopped increasing. D-Serine rapidly returned to baseline steady-state levels as compound 4 is eliminated (compare panels A and B in Figure 4). Administration of 12 mg/kg compound 1 increased D-serine constantly for 2 h. Then the inhibitor concentration fell below 1 μM, whereby D-serine started to decline toward baseline. A second, third, and fourth dose of compound 1 (12 mg/kg every 2 h) sustained inhibitor concentrations above 1 μM and thus supported continued increases in D-serine concentration. Lastly, a single dose of compound 2 also increased D-serine concentration but attained a plateau by 6 h (Figure 4A) and was associated with elevated and sustained inhibitor concentrations (>1 μM). We postulated that such plateaus in D-serine concentrations would be expected in a simple model of competitive inhibition in vivo, where the D-serine increases are suﬃcient to compete with inhibitor to establish new steady state levels of D-serine production and degradation. Under this assumption, we applied Michaelis−Menten competitive binding kinetics (eqs 4−6 in Supporting Information) to derive an expected linear relationship between steady-state D-serine and inhibitor concentrations. This analysis requires only one adjustable parameter, related to the eﬃcacy of an inhibitor to “shift” the apparent Michaelis−

Figure 5. Analysis of the inferred linear relationship as expected during plateaus in D-serine pharmacodynamics from a PK/PD model of competitive inhibition in vivo. Mean cerebellum D-serine content changes (top panels) with concentration of inhibitor (bottom panels) for N = 84 rats (≥3 rats per data point). (A) Compound 2 administered at 125 mg/kg ip increased cerebellum D-serine concentration. Single rats sacriﬁced during plateaus in D-serine (arrows) diﬀered from each other in absolute levels of D-serine and inhibitor concentration, but a plot (C) of individual D-serine concentration vs amount of remaining inhibitor yielded a linear relationship (open circles, 48 h; open triangles, 18 h; shaded triangles, 24 h; ﬁlled circles, mice data as determined after 9 h from Figure 4). Linear ﬁts were constrained to go through the untreated baseline Dserine content corresponding to a value of 5.1 nmol/g (rats) or 2.1 nmol/g (mice). The slopes of these lines yielded a 4 μM invivoKi according to the derivation outlined in the Supporting Information. (D) Similar analysis for compound 1 but using a variety of diﬀerent doses and routes of administration (≥4 rats per data point, N = 37; diamonds = 10 mg/kg, po; open circles = 10 mg/kg, ip; triangles = 3 mg/kg, po). In all cases a plateau in D-serine concentration was observed at 4 h after dose (arrows in panels D and E). These animals also demonstrated the expected linear relationship between accumulated D-serine and remaining inhibitor, with a 650 nM invivoKi value obtained from the slope value (F). G

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

content was linearly related to the cerebellum content of compound 2 in individual animals during these plateaus (Figure 5C). The invivoKi, based on this linear relationship, was estimated to be 4 μM (rats). This was similar to the value of 2 μM obtained in mice (using the plateau observed 9 h after administration of 30 mg/kg ip of compound 2 from the experiment in Figure 4). The above analysis was made possible by the metabolic stability and interanimal variability in exposure of compound 2. On the other hand, compound 1 was cleared more quickly than compound 2, and exposures did not vary as much between individual animals (Figure 5D and Figure 5E). Therefore, we varied exposures by adjusting dose and route of administration and identiﬁed a time at which D-serine concentrations stopped increasing; i.e., in rats, D-serine concentrations stopped increasing 4 h after administration of 10 and 3 mg/kg ip. As expected from the competitive inhibition model, the (transient) plateau observed with compound 1 also yielded a linear relationship between in vivo D -serine and compound concentrations (Figure 5F). This relationship was steeper (relative to compound 2) and yielded a more potent invivoKi of 650 nM. A similar analysis for compound 4 after doses of 10 mg/kg ip and po (data not shown) demonstrated linear relationships and invivoKi values of 400 nM (mice) and 1 μM (rats), but the rapid clearance of this compound may have prevented appropriate sampling during the short-lived plateau. These invivoKi values represent intrinsic potencies in vivo and compare favorably to concentrations interpolated at 50% of the maximal increase in D-serine (EC50) 6 h after administrations of diﬀerent doses (ip, see Figure 6) of compounds 2 (5 μM), 1 (0.5 μM), and 4 (ip, 1.5 μM) to rats.

Figure 7. DAAO inhibition increased extracellular D-serine concentrations in rat cerebellum, as determined by microdialysis. At time zero two rats were administered ip with 125 mg/kg compound 2 (open diamonds and shaded circles). Line represents the linear regression of the 0−6 h time interval indicating an increase in D-serine content as was observed for whole tissue content.

increased D-serine binding to the glycine site of the NMDA receptor. To examine if this occurs in vivo, we attempted to reverse the falling behavior produced by the glycine site antagonist 7-chloro-4-hydroxy-3-(3-phenoxy)phenyl-2-(1H)quinolinone (L-701,324) in rats. Administration of L-701,324 (5 mg/kg ip) reliably produced falling behavior that averaged 20 falls per 2 m traversed (Figure 8). Administration of

Figure 6. Relative in vivo potency for three diﬀerent inhibitors of hDAAO as estimated by the half-maximal increases in D-serine accumulation in cerebellum tissue 6 h after ip administration of diﬀerent doses of inhibitors to rats. EC50 values were determined by ﬁtting to experimental data according to a sigmoidal dose−response relationship (curves) of D-serine and inhibitor concentrations (continuous line). The compound with the lowest in vivo EC50 value was compound 1 (0.5 μM), followed by compound 4 (1.5 μM) and compound 2 (5 μM).

Figure 8. The falling behavior of rats induced by glycine site NMDAR antagonist L-701,324 was reversed by treatments that elevated cerebellum D-serine concentration. Six hours before testing, rats were administered vehicle (squares), compound 2 at 125 mg/kg ip (circles), or 250 mg/kg D-serine plus 125 mg/kg compound 2 (diamonds). Falling index (see Experimental Section) was measured every 5 min for each treatment group and represented as mean values ± 95% conﬁdence intervals in the ﬁgure. The reductions in falling behavior associated with both treatments (compound 2 or D-serine plus compound 2) were statistically separated from vehicle treatments at all time points after administration of L-701,324 (two-way repeated measures ANOVA followed by the Holm−Sidak method for pairwise multiple comparisons versus a control group, 0.05). Group sizes were n = 13 rats for vehicle and compound 2 and n = 4 for the combination of D-serine and compound 2 (diamonds). D-Serine concentrations (mean ± SD, nmol/g wet weight) were estimated from a diﬀerent set of animals 6 h after equivalent treatments.

Changes in extracellular concentrations of D-serine were estimated using microdialysis in cerebellum of two rats treated with 125 mg/kg compound 2 ip (Figure 7). D-Serine increased in the dialysate of rats with a similar time course as the tissue concentrations. D-Serine concentrations in hippocampus, already ∼30-fold higher than cerebellum levels, did not increase signiﬁcantly (tissue or extracellular concentrations, data not shown) after doses of 125 mg/kg compound 2 ip to rats. DAAO Inhibition Reverses Glycine Site Antagonism. Therapeutic eﬀects of DAAO inhibition might occur via

compound 2 or of exogenous D-serine (250 mg/kg ip) in combination with compound 2 (125 mg/kg ip) prevented these falling behaviors measured 6 h after treatment. The resulting Dserine concentrations in cerebellum (130 ± 30 nmol/g, n = 14) were likely suﬃcient to block the eﬀects of L-701,324 at the glycine binding site of NMDARs. In fact, pretreatment with compound 2 alone (125 mg/kg ip) was suﬃcient to reverse these falling behaviors, indicating that suﬃcient D-serine (17 ± 6 nmol/g, n = 37) was generated to compete for the eﬀects of L-701,324. H

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

■

Inhibition of DAAO Activity Increases Occupancy by Glial-Derived D-Serine. It has been proposed that D-serine concentrations within the synaptic cleft depend on the glial environment of neurons.7 Here we hypothesized that DAAO activity near synapses might reduce the occupancy of NMDA receptors by locally degrading glial-derived D-serine. In the supraoptic nucleus (SON) of the rat hypothalamus, glial processes withdraw under physiological conditions such as parturition, lactation, or chronic dehydration.45 In lactating rats, this glial remodeling has been associated with reduced D-serine concentrations in the synaptic cleft and, consequently, impaired synaptic NMDA receptor-mediated responses.7 We used this model to assess the consequences of DAAO inhibition on synaptic NMDA receptor activity under diﬀerent conditions of astrocytic coverage of SON neurons and their synapses: in virgin rats where it is maximal and in lactating animals when glial ensheathing is extensively reduced. In whole-cell patch clamp recordings of SON neurons in hypothalamic slices from virgin rats (Figure 9) the AMPA/NMDA ratio of excitatory

Article

DISCUSSION

In this paper we characterized the structural and kinetic mechanism of DAAO inhibition by small molecules of diﬀerent scaﬀolds. The experiments detailed the molecular mechanism of competitive inhibition and outlined the kinetic consequences of diﬀerent pyrrole carboxylic acid analogues and bioisosteric replacements of benzoate. These inhibitors illustrated the simple pharmacodynamic relationship between D-serine production and degradation, as controlled by DAAO. Finally these results provided evidence that DAAO activity, at least in certain brain regions, reduced the inﬂuence of glial-derived Dserine on local NMDA receptors in a manner sensitive to DAAO inhibition by small molecules. Mechanism of Inhibition. The pyrrole-2-carboxylic acids (compounds 1, 2, and 3) and the bioisosteres of benzoate (compounds 4 and 5) all inhibited hDAAO in vitro but with a range of potencies. The mechanisms of inhibition were all competitive with D-serine, the relevant endogenous substrate in mammalian brains. The inhibitor binding modes were investigated by resolving the 3D structure of the inhibitor− hDAAO complexes by X-ray crystallography (with the only exception of compound 4) (Figure 2A). The overall protein conformation was similar in all the hDAAO−ligand complexes (rms deviation in the 0.7−1.3 Å range over the Cα pairs of the full protein) and was in agreement with the 3D structures previously deposited in the PDB (e.g., with the hDAAO in complex with benzoate: rms deviation in the 0.7−1.0 Å range). Ligand interactions with the residue side chains lining the hDAAO active site were divided into two groups. The ﬁrst group of interactions presented pharmacophore features in the ligands’ planar cores (mainly electrostatic and H-bond interactions) to the polar residues of the active site: speciﬁcally, the inhibitor carboxylic groups with the guanidyl side chain of Arg283 in a fashion similar to all other reported ligands in published DAAO structures. Likewise, an H-bond was provided by the inhibitor pyrrole nitrogen to the backbone carbonyl oxygen atom of Gly313 and by the hydroxyl group of Tyr228 (with the exception of compound 3) (Figure 2A). The second group of interactions presented van der Waals contacts with several hydrophobic side chains, namely, Leu51, Gln53, Leu56, Ile215, His217, and Ile230 and the aromatic rings of Tyr224 and Tyr228. In particular, the chlorophenyl moiety of compound 2 occupied a pocket between the Tyr224 side chain and Leu215. Interestingly, Leu215 corresponds to Met213 in RgDAAO, a residue that was demonstrated to be crucial for the determination of substrate speciﬁcity in this enzyme.46 Unfortunately, it was not possible to estimate the relative contribution of these two groups of interactions to the overall binding energy. Plasticity of the active site loop (residues 216−228) and in particular of Tyr224 is likely important for the steric accommodation of various substrate side chains and for shielding the hydride transfer step from solvent accessibility.38 Moreover, the aromatic side chain of Tyr224 moved (up to ∼2.5 Å) to stack over the aromatic ring of ligands such as benzoate, forming an additional π−π interaction. Interestingly, despite the presence of aromaticity in compound 1, this inhibitor does not induce additional π−π interactions from Tyr224 (Figure 2B). This is also evident in the X-ray structure of a close analogue similar to compound 1 (4H-furo[3,2-b]pyrrole-5-carboxylic acid, PDB code 3CUK).47 Only in the complex between hDAAO and compound 5 was the side chain of Tyr224 placed over the ligand planar rings, its

Figure 9. DAAO inhibition increased glial-derived D-serine occupancy of NMDA receptors in the supraoptic nucleus of lactating rats. (A) Representative traces of NMDA excitatory postsynaptic currents, normalized relative to AMPA currents. (B) Slices incubated with 10 μM compound 2 showed increased NMDA currents only in lactating rats (p < 0.05), indicating enhanced D-serine levels within the synaptic cleft. Numbers above bars indicate number of independent recordings.

postsynaptic currents (EPSCs) was smaller (1.12 ± 0.23; n = 10) than that measured in the SON from lactating animals (4.08 ± 0.41; n = 17; p < 0.05), as previously reported.7 Incubation of the slices for at least 45 min in the presence of compound 2 (10 μM) did not modify signiﬁcantly the AMPA/ NMDA ratio in virgin animals (1.46 ± 0.29; n = 7). However, this ratio was largely reduced in the SON of lactating animals (1.71 ± 0.53; n = 6) because of a signiﬁcant increase of the NMDA component of the EPSC (Figure 9). In the presence of compound 2, the AMPA/NMDA ratio measured in lactating rats was no longer diﬀerent from that measured in virgin animals (p > 0.05), suggesting that DAAO inhibition strongly enhanced D-serine levels within the cleft, thereby increasing the level of occupancy of the NMDA receptor glycine binding sites (Figure 9). The lack of action of the same compound in virgin animals was likely due to an occlusion with endogenous Dserine whose synaptic levels under control conditions are suﬃciently high to provide maximal activation of the glycine site as previously reported.7 I

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

ﬁrst an interaction with the ﬂavin ring, followed by an accommodation of the ligand by conformational changes in the active site lid containing Tyr224. In fact, all hDAAO− inhibitor complex crystal structures showed a similar conformation of the active site lid (Figure 2). It is interesting, furthermore, to speculate that inhibitors incorporating additional interactions with the active site lid would not gain potency as a result of these interactions but rather loose potency as a result of slower rates of association. Pharmacokinetic−Pharmacodynamic Relationship. Typically, drug discovery programs seek to optimize potency of inhibitors relative to other pharmaceutical properties by reliance on IC50 values. These measurements are sensitive to assay conditions, reagent quantities, and incubation times. As a result, it is often misleading to translate concentrations achieved in brain tissue into occupancy of drug target and inhibition of enzyme. To estimate enzyme inhibition achieved in vivo, we created a PK/PD model based on the elevated DAAO activity and D-serine turnover in the rodent cerebellum. Following in vivo administration of each inhibitor, there was a prompt and steady rise in cerebellar D-serine concentration that did not exceeded a characteristic rate (∼2 nmol g−1 h−1, linear part of curves in Figure 4A). Higher doses, or more potent compounds, did not further increase this initial rate of D-serine concentration increase. We assumed that this rate was the endogenous rate of D-serine production (by serine racemase or by spillover from forebrain D-serine or both) and the observed maximal increases in cerebellum D-serine reported 100% inhibition of DAAO activity. In support of this, D-serine increases continued until inhibitor concentrations declined below a critical concentration, thus recovering DAAO activity and ultimately returning to a steady-state balance of D-serine synthesis and degradation at baseline. At certain times after administration of 2 or 1, there were periods of relatively stable D-serine concentration that were nevertheless elevated from baseline levels. These plateaus in D-serine concentration are expected based on a competitive mechanism and were experimentally observed during periods of inhibitor concentrations stable enough to establish a new steady state between D-serine production and degradation (see also ref 50). The PK/PD relationships among the compounds were accurately described using a model of competitive inhibition with only one adjustable parameter of inhibitor potency, here termed invivoKi. This parameter corresponds mathematically to a competitive inhibition constant in vitro (Ki), where the apparent Km in the presence of an inhibitor [I] is Km(1 + [I]/ Ki). We performed computer simulations of D-serine synthesis and degradation, where D-serine synthesis was a constant determined experimentally to be the highest observed rate of Dserine increase upon maximal DAAO inhibition (initial slopes 2 nmol/g/h). For DAAO activity the Michaelis−Menten kinetic equation with Km of 5 mM (determined with the in vitro assay conditions used in this study and also in ref 31) and a maximal activity (Vmax) of DAAO was adjusted empirically until a steadystate D-serine concentration was maintained at baseline cerebellum levels (initial conditions 2 nmol/g for mice and 5 nmol/g for rats). It was apparent that constant inhibitor concentrations yielded plateaus in D-serine levels within hours, reﬂecting the new steady-state levels of D-serine synthesis and degradation established by competition between higher Dserine concentrations at steady inhibitor concentrations. The linear relationships between concentrations of D-serine and inhibitor were in fact observed, as expected by the model, to

original location being occupied by the Tyr55 side chain. In this inhibitor, the two exocyclic oxygen atoms together replaced the geometry and charge of a carboxylate in forming the bidentate interaction with the Arg283 guanidinium group. Compound 5 does not contain a carboxylic acid, but the X-ray structure showed these two exocyclic oxygens participating in a bioisosteric interaction with the side chain of Arg283. Although without an X-ray structure solved, the inhibitor compound 4 also likely binds as a bioisostere of benzoate, with interactions between its isoxazole nitrogen and exocyclic oxygen together replacing the geometry and charge of a carboxylate. The orientation of the exocyclic oxygen in the active site could have two possible directions, generated by ﬂipping compound 4 around its long axis. The energetic preference for these two possible bound states is currently unknown. The closely related inhibitor CBIO (5-chlorobenzo[d]isoxazol-3-ol), whose analogues have been described,48 should bind in the hDAAO active site in a mode similar to that of compound 4. All ﬁve tested inhibitors completely inhibited hDAAO in vitro, with diﬀerent IC50 values, as obtained by two independent assay methods (see Figure 1 and Table 1). Low Ki values (in the nanomolar range) were obtained by global ﬁtting of all substrate and inhibitor concentrations simultaneously and with equal weight by a model of competitive inhibition. The inhibition constants also reported by ref 47 for compounds 1 and 2 were markedly diﬀerent from the values reported here, likely resulting from diﬀerences in assay conditions. For a comparison among recently developed DAAO inhibitors see ref 49. It is interesting to compare the relatively weak binding aﬃnities for several of the compounds reported here, when measured by static titration of the oxidized enzyme in the absence of enzyme turnover, to the often more potent inhibition constants. The notable exception is benzoate whose Kd value (2.2 μM) compares favorably to its Ki value (2.0 μM). The bioisosteric analogue compound 4, on the other hand, has Kd ≈ 4 μM for the oxidized enzyme, but a Ki value of 1.7 nM. This diﬀerence reﬂects the consequence of a very slow release (low krelease value) in the presence of substrate, a parameter that is many orders of magnitude slower than for benzoate (see Figure 3 and Table 1). The other bioisostere of benzoate, compound 5, has appreciably faster dissociation than compound 4 and, as a result, is a less potent inhibitor (Ki value of 13 nM). Nevertheless these bioisosteres had increased inhibitor potency relative to benzoate, which may be attributed to a rigid and coplanar arrangement of their polar atoms interacting with Arg283. The aﬃnity of 3, apparently similar to all the other inhibitors, actually resulted from altered kinetics of both binding and dissociation from the active site of hDAAO. It might be speculated that the phenethyl substitution requires longer times before the appropriate conformation of ligand− protein complex can be achieved. The equilibrium titration experiments were used to measure ligand binding to enzyme in the oxidized state (no substrate turnover). Two diﬀerent spectral changes were evident for compounds 1 and 2, reﬂecting two binding processes with diﬀerent aﬃnities. The weaker aﬃnity interaction was observed at the longer wavelengths, likely resulting from the formation of a charge-transfer complex. While the two binding processes had comparable association rates, the more potent binding mode was likely the result of a slower dissociation (lower koff value), as was observed directly for compound 2. We suggest that the binding of all pyrrole-based inhibitors proceeded in two steps: J

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

this ﬂavooxidase. The substrate stereospeciﬁcity of DAAO for neutral D-amino acids, the ﬂavin cofactor, and the active site plasticity deﬁne the pharmacophore for active-site DAAO inhibitors and determine their range of kinetic parameters. In the brain, the PK/PD relationships between inhibitor and Dserine concentrations were well described by a simple model of production and degradation of brain D-serine. Modeling the PK/PD relationships between inhibitors yielded a single parameter suitable for optimizing in vivo potency. A physiological role of DAAO inhibition was demonstrated as a speciﬁc regulator of D-serine on certain aspects of NMDARmediated neurotransmission, especially during physiological conditions where the glycine sites of NMDARs are not saturated. Further investigations with DAAO inhibitors on NMDAR-regulated neuronal circuits will be important for uncovering the potential therapeutic applications of DAAO inhibitors in human disease.

occur as compound concentrations approached steady state or to occur transiently between the rise and fall of a D-serine time course. Steeper slopes of D-serine versus compound concentrations during these plateau conditions thus deﬁned more potent in vivo inhibition constant (analytical derivation of invivoKi is outlined in Supporting Information). The PK/PD relationships of DAAO inhibitors were well characterized by this model and suggest that constant increased D-serine concentrations in the brain may be obtained therapeutically by maintaining steady-state concentrations of a DAAO inhibitor. DAAO Inhibition Increases D-Serine at Glycine Site of NMDAR and NMDAR Currents. The eﬀects of DAAO inhibition were conveniently monitored using adult cerebellum D-serine concentrations because of the relatively high activity of DAAO compared to forebrain regions (e.g., frontal cortex and hippocampus). Indeed elevation of D-serine in the cerebellum likely aﬀected NMDA receptors in cerebellar circuitry. This was examined in experiments with the glycine-site antagonist L701,324, whose eﬀects on cerebellar NMDA circuitry we speculated to be more pronounced given the lower levels of tissue D-serine concentrations. The cerebellum’s role in coordinating motor activity likely contributed to the eﬀects observed after antagonist administration. Encouraged by the literature reports of ataxic side eﬀects of glycine-site antagonists,51 we examined the behavior of rats treated with L-701,324 and developed a behavioral paradigm (as described in Experimental Section) to quantify the falls of rats traversing a standard distance. Treatment of rats with exogenous D-serine (250 mg/kg ip) in combination with DAAO inhibitor compound 2 (125 mg/kg ip) was suﬃcient to raise cerebellum D-serine content to levels observed in forebrain regions. Consistent with ddY/DAO− mice,17 increased cerebellum Dserine completely reversed the falling behavior induced by the antagonist. In fact, robust reversal of falling behavior was evident without exogenous D-serine and was achieved simply by inhibiting DAAO for 6 h. These data suggest that DAAO inhibition increases NMDA receptor glycine-site occupancy by D-serine in vivo. D-Serine increases observed in total tissue content were also observed extracellularly by microdialysis. The eﬀects of DAAO inhibitors outside the cerebellar circuitry may be limited by the localized expression of DAAO. Activity of DAAO inhibitors in animal models of schizophrenia, for example, has been reported to require coadministration of exogenous D-serine52,53 to demonstrate eﬃcacy. On the other hand, eﬀects of DAAO inhibitors have been described in animal models of chronic pain,54,55 neuropathic pain,56,57 memory,58 hippocampal LTP,58 and EEG.50 NMDAR-mediated changes in synaptic plasticity require coincidence between its activating neurotransmitter and depolarization of its postsynaptic membrane. In synapses characterized by astrocytic processes closely juxtaposed,59 this dependence could be shifted according to the extent of D-serine released from glial cells as demonstrated in the supraoptic nucleus7 or in prefrontal cortex.60 Here we demonstrated that inhibiting DAAO restored the synaptic D-serine concentrations, thus eﬀectively increasing glial control of NMDARs in this excitatory synapse and possibly in other synapses such as those in deﬁned hippocampal networks.

■

EXPERIMENTAL SECTION

Chemicals. D- and L-Serine, Tris-HCl, and EDTA were purchased from Sigma-Aldrich (St. Louis, MO). Synthetic methods were as described in patent examples for compounds 2,40 3,41 4,42 and 1.35 Compounds 1 and 2 were tested in vivo as a sodium salts at >98% purity. Compound 543 and benzoic acid were purchased from SigmaAldrich. Compounds were dissolved in 50 mM phosphate buﬀer and administered orally (po) or intraperitoneally (ip) at a dose volume of 1 or 5 mL/kg for rats or mice, respectively. Control animals were similarly treated without the addition of inhibitors. Animals. Male Sprague−Dawley rats (Charles River Labs, Wilmington, MA; Harlan Laboratories Inc., Indianapolis, IN) weighing between 180 and 250 g at the time of experiment were group-housed and used after at least 2 days of habituation in a temperature- and humidity-controlled environment under a 12 h light−dark cycle (lights on at 7 a.m.). Alternatively, male C57BL/6 mice (Charles River Labs, Wilmington, MA) weighing between 20 and 30 g were used. Animals were allowed food and water ad libitum. Experiments were performed during the light period. Animals were maintained in accordance with the guidelines of an Institutional Animal Care and Use Committee and in accordance with the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, National Research Council, National Academy of Sciences, 1996). DAAO Activity by Coupled Enzyme Assay. The eﬀects of small molecule inhibitors of DAAO were evaluated in vitro by measuring the activity of puriﬁed human recombinant DAAO at various concentrations of D-serine and inhibitors. The activity of DAAO was estimated by the oxygen consumption assay31 or by the production of hydrogen peroxide as monitored via a coupled enzyme assay and commercially available reagent Amplex Red (Invitrogen Molecular Probes). Horseradish peroxidase (HRP) was used to convert the nonﬂuorescent HRP substrate Amplex Red into a ﬂuorescent product resoruﬁn. The reaction was initiated by addition of DAAO, followed by incubation at room temperature for 1 h. Fluorescence was monitored on a FlexStation microplate reader (Molecular Devices Corporation) at the excitation and emission wavelengths of 530 and 590 nm, respectively. Controls by additions of H2O2 (for a 16 μM ﬁnal concentration) were performed on each plate. Reaction conditions were 50 mM sodium phosphate, pH 7.4, 4 units/mL HRP, 50 μM Amplex Red, 1−250 mM D-serine, 1.5% DMSO. The Ki values were estimated by a global ﬁtting routine in GraphPad Prism software using a model of competitive inhibition and Michaelis−Menten kinetics, where

rate =

■

CONCLUSIONS The structural and kinetic mechanisms of DAAO inhibitors described here relate directly to the structure and function of

Vmax[D‐serine] K m obs + [D‐serine]

⎛ [inhibitor] ⎞ K m obs = K m ⎜1 + ⎟ Ki ⎝ ⎠ K

(1a)

(1b)

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

mm, 5 μm, Regis Technologies) to separate the D- and L-enantiomers of serine, with the mobile phase (formic acid and methanol) chosen to elute the D-enantiomer prior to the more abundant L-enantiomer. The detection limit at a 3:1 signal-to-noise ratio was estimated to be ∼400 pg/mL. The internal standard was isotopically labeled D,L-serine-d3 (Cambridge Isotope Labs). Calibration standards (5.0−500 ng/mL) were linear and analyzed before and after every set of experimental tissue samples and generally had less than 15% CV. Computer Docking of Inhibitors. Computer docking procedures utilized OMEGA conformers in the FRED docking program followed by minimization in the SZYBKI program (Openeye Scientiﬁc, Santa Fe, NM). Virtual screens for bioisosteres of carboxylic acids were performed by rapid overlay of chemical structure (ROCS by OpenEye Scientiﬁc) followed by evaluation of electrostatic potentials (EON by OpenEye Scientiﬁc). Commercial sources of small molecular mass compounds were compiled by ChemNavigator (San Diego, CA) and conformation databases calculated by OMEGA. Crystal structures and docked orientations of ligands were prepared by and visualized in MOE (Chemical Computing Group, Montreal, Canada). Figures were prepared in DSViewer (Accelerys Software, Inc.) or by VMD software.62 X-ray Crystallography. hDAAO protein containing amino acids 1−340 was expressed using an N-terminal 6-His tag in BL21(DE3) E. coli cells and puriﬁed by elution on a nickel column followed by size exclusion chromatography.31 Crystallization trials were performed using the holoenzyme form of hDAAO concentrated to approximately 10 mg protein/mL at 1 mM inhibitor ﬁnal concentration. Structure data were obtained from both a laboratory X-ray source (Rigaku FR-E generator with a Saturn detector) and synchrotron sources. Raw diﬀraction data were processed with either the d*TREK63 or the HKL64 software. Statistics for each data collection are reported in detail in Supporting Information Table S1. The hDAAO−inhibitor complex structures were solved by molecular replacement based on Molrep,65 using the hDAAO coordinates (PDB code 2DU8) as the search model for the protein part of the complex. Crystallographic reﬁnement was performed using the program BUSTER (version 2.9, Global Phasing Ltd., Cambridge, U.K.), after model building and inspection using the program COOT.66 TLS reﬁnement was performed on models for crystals containing compounds 2, 3, and 5 to compensate for anisotropy in the diﬀraction data. Inspection of the residual diﬀerence electron density, after the initial round of reﬁnement, allowed us to unambiguously identify and build the bound inhibitors. The relevant reﬁnement statistics are reported in Supporting Information Table S1. The program Procheck67 was used to assess the stereochemical quality of the protein structures. D-Serine Simulations. The pharmacodynamics of D-serine and inhibitor concentrations were simulated using the software package Stella, version 8.1.4 (ISEE Systems). DAAO activity was expressed as a function of enzyme concentration, maximum velocity, inhibitor concentration, and an adjustable parameter here termed invivoKi (see Supporting Information). Compound pharmacokinetics was expressed as ﬁrst-order absorption and clearance rates. D-Serine production was set constant at 2 nmol g−1 h−1. An empirical value of DAAO activity was derived from steady state D-serine concentration in the absence of inhibitor. The values of absorption, clearance, and invivoKi were adjusted manually until D-serine and inhibitor concentrations approximated the time courses for the administrations of compounds 1 and 5 represented in Figure 2. Behavior Assay for NMDAR Glycine Site Antagonism. The falling behavior of rats treated with the glycine-site NMDA receptor antagonist L-701,324 (Tocris Bioscience) was scored manually. Brieﬂy, a wireframe platform 2 m long and 0.5 m wide was installed with a boxed enclosure at one end. Prior to the experiment six rats were housed in one cage for 24 h. On the morning of the experiment, each cage of rats was handled and dosed with DAAO inhibitor 6 h before the start of the measurements. Thirty minutes before measurements began, all six rats were handled, placed on the platform, and allowed to explore for 30 min. Reliably within 20 min, all six rats had ﬁnished exploratory activity and returned to the enclosure. Ten minutes before

The parameters Vmax (maximum enzyme velocity), Km (Michaelis− Menten constant), and Ki (inhibition constant) were adjusted globally to ﬁt all DAAO activity values measured across the various inhibitor and substrate concentrations. Equilibrium Binding Constants by Absorption Spectroscopy. The binding of the inhibitors was investigated by adding increasing concentrations of inhibitor to a ﬁxed amount of hDAAO (10 μM, 0.4 mg/mL) in the presence of ∼6 μM free FAD cofactor in 50 mM potassium phosphate buﬀer, pH 7.5, at 15 °C.31,61 Absorption spectra were collected by using a Jasco V-560 spectrophotometer (300−800 nm). To obtain absorbance changes at speciﬁc wavelengths, the diﬀerence spectra were calculated by using the Jasco software. The dissociation constant (Kd) values for the enzyme−ligand complexes were determined according to a saturation binding model:61

ΔAbs = ΔAbsmax X /(Kd + X )

(2)

where X is the ligand concentration added and ΔAbs is the change in absorbance at a speciﬁc wavelength. Kinetic Constants by Stopped Flow Spectrophotometry. Under the same experimental conditions, human DAAO (0.9 mg/mL) in the presence of 4 μM free FAD was rapidly mixed in a SFM-300 Biologic apparatus with a similar volume of increasing concentrations of inhibitor.61 The time course of spectral changes was recorded in the 300−700 nm wavelength range, and the absorbance vs time data sets were extrapolated at wavelength(s) identiﬁed by equilibrium titrations. Observed rate constants (kobs) at increasing inhibitor concentrations were determined from the time courses by nonlinear regression using single exponential equations. The rate constants of inhibitor association and dissociation were obtained by linear regression of the equation kobs vs inhibitor concentration according to the following equation:44 kobs = koff + konX

(3)

where X is the inhibitor concentration. The rate of inhibitor dissociation was also estimated in the presence of substrate. Here, the oxidized hDAAO was mixed with 40 μM inhibitor under anaerobic conditions with an excess of the substrate Dserine (200 mM). Under these experimental conditions only the free holoenzyme form (following dissociation of inhibitor) could be reduced by the substrate, and the rate of inhibitor dissociation was followed by ﬂavin reduction as a decrease in absorbance at 450 nm. Flavin reduction proceeded very quickly (>100 s−1) following inhibitor release,31 and in the presence of excess substrate it was not ratelimiting. Single or double exponential equations were used to ﬁt to the time courses of absorbance at 450 nm to obtain rates of inhibitor dissociation. Microdialysis. Lightly tethered rats weighing between 290 and 310 g were housed in a microdialysis bowl on an interactive caging system (BASi, Indianapolis, IN) with full range of motion. A 4 mm microdialysis probe (BASi) was implanted in cerebellum to maximize exposure of the dialysis membrane to the vermis (A, − 2.0 mm; V, +3.0 mm; L, 0.0 mm relative to the λ). A jugular vein catheter was also implanted for collection of plasma and was automatically microﬂushed at 12 min intervals to maintain patency. Dialysates samples of 60 μL each were collected every 20 min at a ﬂow rate of 3 μL/min and analyzed by LC/MS/MS for D-serine and inhibitor concentrations. D-Serine Measurements by LC/MS/MS. The pharmacodynamics of DAAO inhibitors was evaluated by measuring the whole tissue content D -serine at deﬁned intervals on individual animals postsacriﬁce. Brieﬂy, animals were sacriﬁced by guillotine and trunk blood was collected in EDTA-coated vials to measure plasma concentrations of inhibitor and D-serine. Whole brain was rapidly removed, and the cerebellum, brainstem, and remaining forebrain regions were rapidly separated. Tissues were immediately frozen in Eppendorf tubes on dry ice and stored at −80 °C until further analysis. Hippocampus tissue was dissected under a stereomicroscope. Tissue samples were sonicated in reagent alcohol (10 times volume of tissue), and supernatants were injected via an autosampler (Leap Technologies) connected to a tandem MS/MS (Applied Biosystems/ MDS Sciex API 5000) and a Chirosil RCA(+) column (150 mm × 4.6 L

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

(7) Panatier, A.; Theodosis, D. T.; Mothet, J. P.; Touquet, B.; Pollegioni, L.; Poulain, D. A.; Oliet, S. H. Glia-derived D-serine controls NMDA receptor activity and synaptic memory. Cell 2006, 125, 775−784. (8) Schell, M. J. The N-methyl D-aspartate receptor glycine site and D-serine metabolism: an evolutionary perspective. Philos. Trans. R. Soc. London, Ser. B 2004, 359, 943−964. (9) Stevens, E. R.; Esguerra, M.; Kim, P. M.; Newman, E. A.; Snyder, S. H.; Zahs, K. R.; Miller, R. F. D-Serine and serine racemase are present in the vertebrate retina and contribute to the physiological activation of NMDA receptors. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6789−6794. (10) Yang, S.; Qiao, H.; Wen, L.; Zhou, W.; Zhang, Y. D-Serine enhances impaired long-term potentiation in CA1 subfield of hippocampal slices from aged senescence-accelerated mouse prone/ 8. Neurosci. Lett. 2005, 379, 7−12. (11) Papouin, T.; Ladepeche, L.; Ruel, J.; Sacchi, S.; Labasque, M.; Hanini, M.; Groc, L.; Pollegioni, L.; Mothet, J. P.; Oliet, S. H. Synaptic and extrasynaptic NMDA receptors are gated by different endogenous coagonists. Cell 2012, 150, 633−646. (12) Pollegioni, L.; Piubelli, L.; Sacchi, S.; Pilone, M. S.; Molla, G. Physiological functions of D-amino acid oxidases: from yeast to humans. Cell. Mol. Life Sci. 2007, 64, 1373−1394. (13) Pollegioni, L.; Molla, G.; Sacchi, S.; Rosini, E.; Verga, R.; Pilone, M. S. Properties and applications of microbial D-amino acid oxidases: current state and perspectives. Appl. Microbiol. Biotechnol. 2008, 78, 1− 16. (14) Hashimoto, A.; Nishikawa, T.; Hayashi, T.; Fujii, N.; Harada, K.; Oka, T.; Takahashi, K. The presence of free D-serine in rat brain. FEBS Lett. 1992, 296, 33−36. (15) Wolosker, H.; Sheth, K. N.; Takahashi, M.; Mothet, J. P.; Brady, R. O., Jr.; Ferris, C. D.; Snyder, S. H. Purification of serine racemase: biosynthesis of the neuromodulator D-serine. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 721−725. (16) Morikawa, A.; Hamase, K.; Inoue, T.; Konno, R.; Niwa, A.; Zaitsu, K. Determination of free D-aspartic acid, D-serine and D-alanine in the brain of mutant mice lacking D-amino acid oxidase activity. J. Chromatogr., B: Biomed. Sci. Appl. 2001, 757, 119−125. (17) Almond, S. L.; Fradley, R. L.; Armstrong, E. J.; Heavens, R. B.; Rutter, A. R.; Newman, R. J.; Chiu, C. S.; Konno, R.; Hutson, P. H.; Brandon, N. J. Behavioral and biochemical characterization of a mutant mouse strain lacking D-amino acid oxidase activity and its implications for schizophrenia. Mol. Cell. Neurosci. 2006, 32, 324−334. (18) Maekawa, M.; Watanabe, M.; Yamaguchi, S.; Konno, R.; Hori, Y. Spatial learning and long-term potentiation of mutant mice lacking D-amino-acid oxidase. Neurosci. Res. 2005, 53, 34−38. (19) Wake, K.; Yamazaki, H.; Hanzawa, S.; Konno, R.; Sakio, H.; Niwa, A.; Hori, Y. Exaggerated responses to chronic nociceptive stimuli and enhancement of N-methyl-D-aspartate receptor-mediated synaptic transmission in mutant mice lacking D-amino-acid oxidase. Neurosci. Lett. 2001, 297, 25−28. (20) Hashimoto, A.; Yoshikawa, M.; Niwa, A.; Konno, R. Mice lacking D-amino acid oxidase activity display marked attenuation of stereotypy and ataxia induced by MK-801. Brain Res. 2005, 1033, 210−215. (21) Labrie, V.; Duffy, S.; Wang, W.; Barger, S. W.; Baker, G. B.; Roder, J. C. Genetic inactivation of D-amino acid oxidase enhances extinction and reversal learning in mice. Learn. Mem. 2009, 16, 28−37. (22) Labrie, V.; Clapcote, S. J.; Roder, J. C. Mutant mice with reduced NMDA-NR1 glycine affinity or lack of D-amino acid oxidase function exhibit altered anxiety-like behaviors. Pharmacol., Biochem. Behav. 2009, 91, 610−620. (23) Sacchi, S.; Bernasconi, M.; Martineau, M.; Mothet, J. P.; Ruzzene, M.; Pilone, M. S.; Pollegioni, L.; Molla, G. pLG72 modulates intracellular D-serine levels through its interaction with D-amino acid oxidase: effect on schizophrenia susceptibility. J. Biol. Chem. 2008, 283, 22244−22256. (24) Pollegioni, L.; Sacchi, S. Metabolism of the neuromodulator Dserine. Cell. Mol. Life Sci. 2010, 67, 2387−2404.

measurements began, each rat in turn was removed from the enclosure, placed at the opposite end of the platform, and allowed to ﬁnd its way back to the enclosure. This was repeated again at time zero immediately after injection of L-701,324. Reliably and repeatedly, each rat traversed the 2 m platform to return to the enclosure without any falls or foot stumbles. All traverses were recorded on video for scoring after experiment. After injection of 5 mg/kg ip L-701,324, each rat was observed during two sequential trips (total of 4 m) in rapid succession every 5 min for a total of 25 min after injection. Treatment with L-701,324 resulted in an increased number of falls, as deﬁned by a foot misplacement between the wire frames of the platform or a stumble resulting in the rat’s tail hitting the wire frame. The treatment did not aﬀect the overall speed or apparent intent of the rat to return to the enclosure. Falling behavior was analyzed using two-way repeated-measures analysis of variance (RMANOVA; SigmaStat, version 3.5, Systat Software, Inc., Chicago, IL, U.S.) with treatment as the between-groups factor and with time as the within-groups factor.

■

ASSOCIATED CONTENT

S Supporting Information *

Table S1 listing X-ray data and processing statistics; derivation of in vivo D-serine concentration as a function of inhibitor concentration (eqs 4a, 4b, and 5). This material is available free of charge via the Internet at http://pubs.acs.org. Accession Codes

Atomic coordinates and structure factors have been deposited with PDB accession codes 3znn (hDAAO−1 complex), 3zno (hDAAO−2 complex), 3znq (hDAAO−3 complex), and 3znp (hDAAO−5 complex).

■

AUTHOR INFORMATION

Corresponding Author

*Phone: (508) 357-7706. Fax: (508) 490-5454. E-mail: seth. [email protected]. Notes

The authors declare no competing ﬁnancial interest.

■

ABBREVIATIONS USED DAAO, D-amino acid oxidase; hDAAO, human D-amino acid oxidase; NMDAR, N-methyl-D-aspartate receptor; EPSC, excitatory postsynaptic current; SON, supraoptic nucleus

■

REFERENCES

(1) Collingridge, G. L.; Volianskis, A.; Bannister, N.; France, G.; Hanna, L.; Mercier, M.; Tidball, P.; Fang, G.; Irvine, M. W.; Costa, B. M.; Monaghan, D. T.; Bortolotto, Z. A.; Molnar, E.; Lodge, D.; Jane, D. E. The NMDA receptor as a target for cognitive enhancement. Neuropharmacology 2013, 64, 13−26. (2) Zorumski, C. F.; Izumi, Y. NMDA receptors and metaplasticity: mechanisms and possible roles in neuropsychiatric disorders. Neurosci. Biobehav. Rev. 2012, 36, 989−1000. (3) Gonda, X. Basic pharmacology of NMDA receptors. Curr. Pharm. Des 2012, 18, 1558−1567. (4) Gustafson, E. C.; Stevens, E. R.; Wolosker, H.; Miller, R. F. Endogenous D-serine contributes to NMDA-receptor-mediated lightevoked responses in the vertebrate retina. J. Neurophysiol. 2007, 98, 122−130. (5) Junjaud, G.; Rouaud, E.; Turpin, F.; Mothet, J. P.; Billard, J. M. Age-related effects of the neuromodulator D-serine on neurotransmission and synaptic potentiation in the CA1 hippocampal area of the rat. J. Neurochem. 2006, 98, 1159−1166. (6) Mothet, J. P.; Parent, A. T.; Wolosker, H.; Brady, R. O., Jr.; Linden, D. J.; Ferris, C. D.; Rogawski, M. A.; Snyder, S. H. D-Serine is an endogenous ligand for the glycine site of the N-methyl-D-aspartate receptor. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 4926−4931. M

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

(25) Schell, M. J.; Brady, R. O., Jr.; Molliver, M. E.; Snyder, S. H. DSerine as a neuromodulator: regional and developmental localizations in rat brain glia resemble NMDA receptors. J. Neurosci. 1997, 17, 1604−1615. (26) Hashimoto, A.; Oka, T.; Nishikawa, T. Anatomical distribution and postnatal changes in endogenous free D-aspartate and D-serine in rat brain and periphery. Eur. J. Neurosci. 1995, 7, 1657−1663. (27) Kim, P. M.; Aizawa, H.; Kim, P. S.; Huang, A. S.; Wickramasinghe, S. R.; Kashani, A. H.; Barrow, R. K.; Huganir, R. L.; Ghosh, A.; Snyder, S. H. Serine racemase: activation by glutamate neurotransmission via glutamate receptor interacting protein and mediation of neuronal migration. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 2105−2110. (28) Komuro, H.; Rakic, P. Modulation of neuronal migration by NMDA receptors. Science 1993, 260, 95−97. (29) Fuchs, S. A.; Dorland, L.; de Sain-van der Velden, M. G.; Hendriks, M.; Klomp, L. W.; Berger, R.; de Koning, T. J. D-Serine in the developing human central nervous system. Ann. Neurol. 2006, 60, 476−480. (30) Sacchi, S.; Rosini, E.; Pollegioni, L.; Molla, G. D-Amino acid oxidase inhibitors as a novel class of drugs for schizophrenia therapy. Curr. Pharm. Des 2013, 19, 2499−2511. (31) Molla, G.; Bernasconi, M.; Sacchi, S.; Pilone, M. S.; Pollegioni, L. Expression in Escherichia coli and in vitro refolding of the human protein pLG72. Protein Expression Purif. 2006, 46, 150−155. (32) Kawazoe, T.; Tsuge, H.; Pilone, M. S.; Fukui, K. Crystal structure of human D-amino acid oxidase: context-dependent variability of the backbone conformation of the VAAGL hydrophobic stretch located at the si-face of the flavin ring. Protein Sci. 2006, 15, 2708−2717. (33) Klein, J. R. Inhibition of D-amino acid oxidase by aromatic acids. J. Biol. Chem. 1953, 205, 725−730. (34) Frisell, W. R.; Hellerman, L.; Lowe, H. J. Flavoenzyme catalysis: substrate-competitive inhibition of D-amino acid oxidase. J. Biol. Chem. 1956, 223, 75−83. (35) Dorsey, J.; Hopkins, S.; Foglesong, R.; Heﬀernan, M.; Jones, M.; Perales, J.; Fang, Q.; Ogbu, C.; Jones, S.; Soukri, M.; Varney, M.; Spear, K. Fused Heterocyclic Inhibitors of D-Amino Acid Oxidase. WO2008/005456 A2, 2008. (36) Bartlett, G. R. The inhibition of d-amino acid oxidase by benzoic acid and various monosubstituted benzoic acid derivatives. J. Am. Chem. Soc. 1948, 70, 1010−1011. (37) Pollegioni, L.; Diederichs, K.; Molla, G.; Umhau, S.; Welte, W.; Ghisla, S.; Pilone, M. S. Yeast D-amino acid oxidase: structural basis of its catalytic properties. J. Mol. Biol. 2002, 324, 535−546. (38) Todone, F.; Vanoni, M. A.; Mozzarelli, A.; Bolognesi, M.; Coda, A.; Curti, B.; Mattevi, A. Active site plasticity in D-amino acid oxidase: a crystallographic analysis. Biochemistry 1997, 36, 5853−5860. (39) Sacchi, S.; Caldinelli, L.; Cappelletti, P.; Pollegioni, L.; Molla, G. Structure−function relationships in human D-amino acid oxidase. Amino Acids 2012, 43, 1833−1850. (40) Fang, Q.; Hopkins, S.; Heﬀernan, M.; Chytil, M. Pyrrole and Pyrazole DAAO Inhibitors. US 7,893,098 B2, 2011; Sepracor Inc. (41) Heﬀernan, M. L. R.; Dorsey, J. M.; Fang, Q. K.; Foglesong, R. J.; Hopkins, S. C.; Ogbu, C. O.; Soukri, M.; Spear, K. L. Fused Heterocycles. US 7,579,370 B2, 2009; Sepracor Inc. (42) Fang, Q.; Hopkins, S.; Heﬀernan, M.; Chytil, M.; Wipf, P. Benzo[d]isoxazol-3-ol DAAO Inhibitors. US 7,166,725 B2, 2007; Sepracor Inc. (43) Offe, H.; Jatzkewitz, H. 3-Chromene-4-carboxylic acids. Chem. Ber. 1947, 80, 469−478. (44) Fersht, A. The Basic Equations of Enzyme Kinetics. In Structure and Mechanism in Protein Science. A Guide to Enzyme Catalysis and Protein Folding; W. H. Freeman & Co.: New York, 1999; pp 103−127. (45) Theodosis, D. T. Oxytocin-secreting neurons: a physiological model of morphological neuronal and glial plasticity in the adult hypothalamus. Front. Neuroendocrinol. 2002, 23, 101−135.

(46) Sacchi, S.; Lorenzi, S.; Molla, G.; Pilone, M. S.; Rossetti, C.; Pollegioni, L. Engineering the substrate specificity of D-amino-acid oxidase 2. J. Biol. Chem. 2002, 277, 27510−27516. (47) Sparey, T.; Abeywickrema, P.; Almond, S.; Brandon, N.; Byrne, N.; Campbell, A.; Hutson, P. H.; Jacobson, M.; Jones, B.; Munshi, S.; Pascarella, D.; Pike, A.; Prasad, G. S.; Sachs, N.; Sakatis, M.; Sardana, V.; Venkatraman, S.; Young, M. B. The discovery of fused pyrrole carboxylic acids as novel, potent D-amino acid oxidase (DAO) inhibitors. Bioorg. Med. Chem. Lett. 2008, 18, 3386−3391. (48) Ferraris, D.; Duvall, B.; Ko, Y. S.; Thomas, A. G.; Rojas, C.; Majer, P.; Hashimoto, K.; Tsukamoto, T. Synthesis and biological evaluation of D-amino acid oxidase inhibitors. J. Med. Chem. 2008, 51, 3357−3359. (49) Sacchi, S.; Rosini, E.; Pollegioni, L.; Molla, G. D-Amino acid oxidase inhibitors as a novel class of drugs for schizophrenia therapy. Curr. Pharm. Des. 2013, 19, 2499−2511. (50) Strick, C. A.; Li, C.; Scott, L.; Harvey, B.; Hajos, M.; Steyn, S. J.; Piotrowski, M. A.; James, L. C.; Downs, J. T.; Rago, B.; Becker, S. L.; El-Kattan, A.; Xu, Y.; Ganong, A. H.; Tingley, F. D., III; Ramirez, A. D.; Seymour, P. A.; Guanowsky, V.; Majchrzak, M. J.; Fox, C. B.; Schmidt, C. J.; Duplantier, A. J. Modulation of NMDA receptor function by inhibition of D-amino acid oxidase in rodent brain. Neuropharmacology 2011, 61, 1001−1015. (51) Konieczny, J.; Ossowska, K.; Schulze, G.; Coper, H.; Wolfarth, S. L-701,324, a selective antagonist at the glycine site of the NMDA receptor, counteracts haloperidol-induced muscle rigidity in rats. Psychopharmacology (Berlin) 1999, 143, 235−243. (52) Hashimoto, K.; Fujita, Y.; Horio, M.; Kunitachi, S.; Iyo, M.; Ferraris, D.; Tsukamoto, T. Co-administration of a D-amino acid oxidase inhibitor potentiates the efficacy of D-serine in attenuating prepulse inhibition deficits after administration of dizocilpine. Biol. Psychiatry 2009, 65, 1103−1106. (53) Smith, S. M.; Uslaner, J.; Yao, L.; Mullins, C.; Surles, N.; Huszar, S.; McNaughton, C.; Pascarella, D.; Kandebo, M.; Hinchliffe, R.; Sparey, T.; Brandon, N. J.; Jones, B.; Venkatraman, S.; Young, M. B.; Sachs, N.; Jacobson, M.; Hutson, P. H. The behavioral and neurochemical effects of a novel D-amino acid oxidase inhibitor 4Hthieno[3,2-b]pyrrole-5-carboxylic acid (compound 8) and D-serine. J. Pharmacol. Exp. Ther. 2009, 328, 921−930. (54) Chen, X. L.; Li, X. Y.; Qian, S. B.; Wang, Y. C.; Zhang, P. Z.; Zhou, X. J.; Wang, Y. X. Down-regulation of spinal D-amino acid oxidase expression blocks formalin-induced tonic pain. Biochem. Biophys. Res. Commun. 2012, 421, 501−507. (55) Gong, N.; Gao, Z. Y.; Wang, Y. C.; Li, X. Y.; Huang, J. L.; Hashimoto, K.; Wang, Y. X. A series of D-amino acid oxidase inhibitors specifically prevents and reverses formalin-induced tonic pain in rats. J. Pharmacol. Exp. Ther. 2011, 336, 282−293. (56) Hopkins, S. C.; Zhao, F. Y.; Bowen, C. A.; Fang, X.; Wei, H.; Heﬀernan, M. L.; Spear, K. L.; Spanswick, D. C.; Varney, M. A.; Large, T. H. Pharmacodynamic eﬀects of a D-amino acid oxidase inhibitor indicate a spinal site of action in rat models of neuropathic pain. J. Pharmacol. Exp. Ther. [Online early access]. DOI: 0.1124/ jpet.113.204016. Published Online: Mar 21, 2013. (57) Zhao, W. J.; Gao, Z. Y.; Wei, H.; Nie, H. Z.; Zhao, Q.; Zhou, X. J.; Wang, Y. X. Spinal D-amino acid oxidase contributes to neuropathic pain in rats. J. Pharmacol. Exp. Ther. 2010, 332, 248−254. (58) Hopkins, S. C.; Campbell, U. C.; Heﬀernan, M. L.; Spear, K. L.; Jeggo, R. D.; Spanswick, D. C.; Varney, M. A.; Large, T. H. Eﬀects of D-amino acid oxidase inhibition on memory performance and longterm potentiation in vivo. Submitted for publication, 2013. (59) Haydon, P. G. GLIA: listening and talking to the synapse. Nat. Rev. Neurosci. 2001, 2, 185−193. (60) Fossat, P.; Turpin, F. R.; Sacchi, S.; Dulong, J.; Shi, T.; Rivet, J. M.; Sweedler, J. V.; Pollegioni, L.; Millan, M. J.; Oliet, S. H.; Mothet, J. P. Glial D-serine gates NMDA receptors at excitatory synapses in prefrontal cortex. Cereb. Cortex 2012, 22, 595−606. (61) Harris, C. M.; Molla, G.; Pilone, M. S.; Pollegioni, L. Studies on the reaction mechanism of Rhodotorula gracilis D-amino-acid oxidase. N

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Role of the highly conserved Tyr-223 on substrate binding and catalysis. J. Biol. Chem. 1999, 274, 36233−36240. (62) Humphrey, W.; Dalke, A.; Schulten, K. VMD: visual molecular dynamics. J. Mol. Graphics 1996, 14, 33−38. (63) Pflugrath, J. W. The finer things in X-ray diffraction data collection. Acta Crystallogr., Sect. D: Biol. Crystallogr. 1999, 55, 1718− 1725. (64) Otwinowski, Z.; Minor, W. Processing of X-ray Diﬀraction Data Collected in Oscillation Mode. In Methods in Enzymology; Academic Press: New York, 1997; pp 307−326. (65) Vagin, A.; Teplyakov, A. MOLREP: an automated program for molecular replacement. J. Appl. Crystallogr. 1997, 30, 1022−1102. (66) Emsley, P.; Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 2126−2132. (67) Laskowski, R.; MacArthur, M.; Moss, D.; Thornton, J. PROCHECK, a program to check the stereochemical quality of protein structure. J. Appl. Crystallogr. 1993, 26, 283−291.

O

dx.doi.org/10.1021/jm4002583 | J. Med. Chem. XXXX, XXX, XXX−XXX

Structural, Kinetic, and Pharmacodynamic Mechanisms of d-Amino

Recommend Documents