Errata: Corrigendum Volume 280, Issue 23, 6274, Article first published online: 20 November 2013
A. Mozzarelli, Department of Biochemistry and Molecular Biology, University of Parma, Viale GP Usberti 23/A, 43100 Parma, Italy Fax: +39 0521 905151 Tel: +39 0521 905138 E-mail: firstname.lastname@example.org
Kynurenine aminotransferase (KAT) is a pyridoxal 5′-phosphate-dependent enzyme that catalyzes the conversion of kynurenine, an intermediate of the tryptophan degradation pathway, into kynurenic acid, an endogenous antagonist of ionotropic excitatory amino acid receptors in the central nervous system. KATII is the prevalent isoform in mammalian brain and a drug target for the treatment of schizophrenia. We have carried out a spectroscopic and functional characterization of both the human wild-type KATII and a variant carrying the active site mutation Tyr142→Phe. The transamination and the β-lytic activity of KATII towards the substrates kynurenine and α-aminoadipate, the substrate analog β-chloroalanine and the inhibitors (R)-2-amino-4-(4-(ethylsulfonyl))-4-oxobutanoic acid and cysteine sulfinate were investigated with both conventional assays and a novel continuous spectrophotometric assay. Furthermore, for high-throughput KATII inhibitor screenings, an endpoint assay suitable for 96-well plates was also developed and tested. The availability of these assays and spectroscopic analyses demonstrated that (R)-2-amino-4-(4-(ethylsulfonyl))-4-oxobutanoic acid and cysteine sulfinate, reported to be KATII inhibitors, are poor substrates that undergo slow transamination.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Kynurenine aminotransferase (KAT, EC 18.104.22.168) is a pyridoxal 5′-phosphate (PLP)-dependent enzyme catalyzing the irreversible transamination of l-kynurenine (KYN) to produce kynurenic acid (KYNA). KYN is the central metabolite in the kynurenine pathway (Scheme 1), the main catabolic process of tryptophan in most living organisms . Kynurenine pathway enzymes and metabolites (kynurenines) affect biological functions of the immune and nervous systems [2–6]. In particular, KYNA acts as a broad-spectrum endogenous antagonist of all three ionotropic excitatory amino acid receptors in the central nervous system (CNS), the ligand-gated ion channel receptors N-methyl-d-aspartate (IC50 ≅ 8 μm) , and the α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid and kainate receptors . It has been reported that KYNA also acts as a noncompetitive inhibitor of the α7-nicotinic acetylcholine receptor [9–12] and is an endogenous ligand of an orphan G-protein-coupled receptor (GPR35) that is predominantly expressed in immune cells . The activation of glutamate receptors is responsible for basal excitatory synaptic transmission and for mechanisms that underlie learning and memory, such as long-term potentiation and long-term depression [14,15]. Any event that causes overactivation of glutamate receptors leads to a rise in intracellular Ca2+ levels that promotes neuronal cell damage by both activating destructive enzymes and increasing the formation of reactive oxygen species [16,17]. Consequently, mechanisms capable of preventing glutamate receptors from being overstimulated seem to be essential for maintaining the normal physiological condition in the CNS. KYNA is considered to be an antiexcitotoxic agent, limiting neurotoxicity arising from N-methyl-d-aspartate receptor overstimulation . Pharmacologically induced increases in KYNA provide neuronal protection against ischemic damage and anticonvulsive action [18–20]. However, an increase in the endogenous levels of KYNA is associated with reduced glutamate release (glutamatergic hypofunction) and, consequently, decreased extracellular dopamine levels , leading to impaired cognitive capacity  and schizophrenia [23–26]. Furthermore, KYNA levels are abnormally high in the brain and cerebrospinal fluids of Alzheimer’s disease patients  and in the frontal and temporal cortices of Down’s syndrome patients .
[ KYN pathway in mammalian cells. ]
On the basis of the intimate relationships between abnormally high brain KYNA concentrations and neurodegenerative diseases and psychotic disorders, the enzymes involved in KYNA synthesis have been considered as potential targets for the development of compounds with inhibitory activity [2–4,6,18,28–32]. It is well established that KYN transamination to produce KYNA in the CNS of mammals is carried out by at least four distinct enzymes, constituting the KAT family [33–39]: (a) KATI/glutamine transaminase K/cysteine conjugate β-lyase 1; (b) KATII/α-aminoadipate (AAD) aminotransferase; (c) KATIII/cysteine conjugate β-lyase 2; (d) KATIV/glutamic-oxaloacetic transaminase 2/mitochondrial aspartate aminotransferase (AspAT).
KYNA does not cross the blood–brain barrier, and is thus produced in the CNS . Although all four isoforms are present in the mammalian brain, to different extents, only KATI and KATII have been thoroughly characterized with respect to their role in cerebral KYNA synthesis [35,41]. These two isoforms differ by substrate specificity, with KATI showing lower KYN specificity than KATII . The intrinsic catalytic promiscuity of KATI is enhanced by a β-lyase activity [42,43]. Therefore, KATII has been considered to be the principal isoform responsible for the synthesis of KYNA in the rodent and human brain [34,35,41]. Crystallographic studies of KATs from different organisms, including humans, indicate that this enzyme belongs to the α-family of PLP-dependent enzymes  and to the fold type I group [45–49]. However, mammalian KATII and homologs from yeast and thermophilic bacteria do not belong to any of the seven subgroups of fold type I aminotransferases , but rather form a distinct subfamily [47,50,51]. Furthermore, human KATII shows intriguing structural determinants , such as the conformation adopted by the N-terminal region and the presence of Tyr142 above the cofactor molecule. These features are typical of PLP-dependent β-lyases [46,53], and hint at additional PLP-dependent reactions catalyzed by KATII .
Although KATII is considered to be interesting drug target in the treatment of schizophrenia and other neurological disorders [54,55], only a few inhibitors have so far been developed [55–61]. They are depicted in Scheme 2. From the point of view of drug development, the existence in the human brain of at least four KYNA-synthesizing enzymes, combined with the need for fine-tuning of KYNA levels to avoid the potentially harmful effects caused by a deficiency of this metabolite in the CNS, requires the design of isozyme-specific inhibitors . The isozyme specificity of a KATII inhibitor, 1-methyl-4-phenylpyridinium (MPP+) , has been reported, and might be the starting point for the development of potent and specific inhibitors of the synthesis of KYNA in the brain. Recently, the three-dimensional structure of the complex between KATII and a fluoroquinolone derivative, BFF-122, has been solved at 2.1-Å resolution, allowing, in combination with spectroscopic and inhibition studies, ascertainment of the mechanism of action of this inhibitor . BFF-122 forms a hydrazone adduct with PLP, and is thus an irreversible inhibitor, like the majority of the pharmacologically relevant inhibitors of PLP-dependent enzymes .
[ KATII natural substrates (KYN and AAD) and inhibitors: ESBA [55,57], CSA , MPP+, 3-NPA , OPS , and BFF-122 [56,62]. ]
In this study, we have characterized (a) the absorption and fluorescence properties and (b) the transamination and the β-elimination in the presence of substrates and substrate analogs of recombinant human KATII and a variant carrying the Tyr142→Phe mutation, which is expected, on the basis of structural evaluations, to exhibit a decreased propensity for β-elimination , a side reaction common to transaminases. During this investigation, two efficient and rapid assays were developed to screen KATII inhibitors: (a) a continuous assay based on the absorbance of the natural substrate KYN; and (b) an endpoint assay, suitable for 96-well plates, based on the coupling of KAT activity to reporter reactions catalyzed by glutamate oxidase and peroxidase. The latter assay is well suited for high-throughput screening of KATII inhibitors.
Spectroscopic characterization of KATII and Tyr142→Phe KATII
The absorption spectrum of human KATII (Fig. 1A) at pH 7.5 exhibited, in addition to the band centered at 278 nm, a band at 360 nm that is typical of a deprotonated internal aldimine. The A280 nm/A360 nm ratio was 5. The extinction coefficient calculated by the method of Peterson  was found to be 9510 m−1·cm−1. The absorption spectrum exhibited a shoulder at about 420 nm that might be attributable to the protonated internal aldimine (see below). KATII instability at pH values lower than 6 precluded the determination of the pH dependence of the protonation of the internal aldimine. In the presence of the natural, nonchromophoric substrate AAD  (Scheme 2), the band at 360 nm disappeared and a species absorbing maximally at 325 nm accumulated, probably the pyridoxamine form of the cofactor (Fig. 1A; Scheme 3, species 5). The shoulder at 420 nm remained unmodified, suggesting the presence of an inactive PLP enzyme species.
[ General reaction mechanism of aminotransferases, including tautomeric and protonation equilibria. The absorption maxima for the catalytic intermediates are reported. The β-elimination side reaction is boxed. Adapted from [44,74,118]. ]
Tyr142→Phe KATII exhibited an absorption spectrum that was almost superimposable on that of the wild-type enzyme, with an invariant A280 nm/A360 nm ratio (data not shown). Because the extinction coefficient of the cofactor at 360 nm was found to be 9400 m−1·cm−1, this invariant ratio can be explained by a concomitant decrease in the extinction coefficient at 280 nm resulting from the Tyr→Phe substitution. The addition of AAD to the mutant caused spectroscopic changes similar to those observed for the wild-type enzyme, with a less intense peak at 325 nm (the wild-type/Tyr142→Phe ratio at 325 nm was 1.12; data not shown).
KATII contains three tryptophans. The emission spectrum upon excitation at 298 nm showed a band centered at 345 nm, indicative of tryptophans being predominantly exposed to solvent. No energy transfer occurred between tryptophan and PLP, as indicated by the absence of peaks centered at either 420 or 500 nm (Fig. 1B), in contrast to observations on fold type II enzymes, such as tryptophan synthase  and O-acetylserine sulfhydrylase . Direct excitation of PLP at 360 nm gave a structured emission with a maximum at 417 nm and a shoulder at 520 nm (Fig. 1C). The emission at 417 nm is typical of the enolimine tautomer of the internal aldimine, whereas the emission at 520 nm is typical of the ketoenamine tautomer [67,68]. The fluorescence emission spectrum of Tyr142→Phe KATII was indistinguishable from that of wild-type KATII.
A new continuous spectrophotometric assay for KATII activity
To overcome the limitations of the discontinuous KAT assay [41,56–58,69,70], an assay for the continuous monitoring of KYN transamination was developed. The absorption spectrum of a solution containing 900 μm KYN, 10 mmα-ketoglutarate (KG) and 50 μm PLP (pH 7.5, 37 °C) exhibited a maximum at 361 nm, typical of KYN at neutral pH. The spectrum obtained upon addition of KATII to the reaction mixture and equilibration exhibited a band at 332 nm and a shoulder at 344 nm (Fig. 2A), typical of KYNA . The difference spectrum (Fig. 2A, inset) showed a positive peak at about 340 nm and a negative peak at 360 nm. Thus, at wavelengths lower than 352 nm, the accumulation of KYNA could be monitored with good sensitivity. Nonetheless, the extinction coefficient of KYN at 340 nm was too high (3290 m−1·cm−1) to allow for initial velocity determinations at KYN concentrations higher than 8 mm, the published Km for KYN being about 5 mm . Therefore, assays were carried out with monitoring of the reaction at 310 nm, a wavelength that represents a compromise between high sensitivity and an extinction coefficient for KYN that is low enough to monitor time courses with KYN concentrations up to about three-fold the expected Km. The extinction coefficients at 310 nm for KYN and KYNA were calculated to be 1049 m−1·cm−1 and 4674 m−1·cm−1, respectively, with a Δε at 310 nm of 3625 m−1·cm−1. Under these conditions, reactions were carried out as a function of KYN concentration between 2 and 23 mm, in the presence of 10 mm KG (Fig. 2B). This KG concentration was assumed to be saturating on the basis of the previously determined Km for KG of 1.2 mm . Control measurements showed that V0 values were independent of KG concentration down to 0.2 mm (Fig. 2B, inset). At the lowest KG concentration (0.02 mm) and at high KYN concentrations, the rate of the reverse reaction from PMP to PLP became rate-limiting (Fig. 2B, inset). Data points, reported in the typical Michaelis–Menten plot (Fig. 2B), were fitted to K′ values of 10 ± 1 mm, V′max values of 0.022 ± 0.001 mm·min−1, and a kcat of 25 min−1. In order to directly compare the rate determined with the continuous assay with the rate reported for the discontinuous assay, which was carried out in the presence of 200 mm potassium phosphate, 5 mm KG, and 0.04 mm PLP (pH 7.5, 45 °C) (116), we assayed the enzyme under the same experimental conditions. The continuous assay gave a Km (mm) of 2.09 and a kcat (min−1) of 110. The discontinuous assay gave a Km (mm) of 0.96 and a kcat (min−1) of 186. The kcat difference is mostly attributable to the method used for evaluation of the protein concentration. In fact, for the published data (116), the protein concentration was determined with the Bradford method, whereas we measured the bound PLP concentration with the alkali method. We determined that the PLP method led to a 1.38-fold higher value for active site concentration than the Bradford method. Thus, the actual kcat (min−1) for the discontinuous assay was 134, only 1.21-fold higher than the value determined with the continuous assay.
β-Lyase activity of KATII and Tyr142→Phe KATII
It is well established that transaminases, owing to the chemistry of the catalyzed reaction, are prone to β-elimination as a side reaction when the substrate contains a good β-leaving group [43,72–74]. In fact, the quinonoid intermediate formed upon α-proton removal can follow two pathways (Scheme 3): (a) protonation on the imine nitrogen to form the ketimine (transamination pathway); and (c) elimination of the β-substituent to form the α-aminoacrylate Schiff base (β-elimination pathway), which spontaneously and irreversibly hydrolyzes to pyruvate and ammonia. In turn, these products may inhibit or inactivate the enzyme.
First, we analyzed the reaction of KATII in the presence of 5 mmβ-chloroalanine (BCA), a substrate that contains chloride as a good β-leaving group . Transamination of BCA in the presence of 10 mm KG was found to be negligible, as measured by the glucose oxidase (GOX)-coupled assay, which monitors the formation of glutamate (see Experimental procedures, and below). In contrast, a series of spectra recorded as a function of time exhibited the accumulation of a species with maximum absorbance at 330 nm (Fig. 3A), which progressively shifted to about 315–320 nm. Upon reaction completion, the concentration of pyruvate was estimated on the basis of the absorbance at 315 nm, and was found to be 3.7 mm. The concentration of ammonia, determined by Nessler’s assay, was 3.8 mm. This indicates that a significant amount of BCA had undergone a β-elimination reaction, with formation of α-aminoacrylate, which decomposes to pyruvate and ammonia. The same assay carried out on Tyr142→Phe KATII indicated a reduced efficiency of the mutant in the β-elimination of chloride in the presence of BCA. In fact, only ∼ 1.8 mm ammonia was produced from 5 mm BCA, under the same conditions. The initial rate of pyruvate formation catalyzed by KATII and Tyr142→Phe KATII in the presence of BCA (Fig. 3B) allowed determination of specific activities of 5 nmol·μg−1·min−1 and 0.22 nmol·μg−1·min−1, respectively. The formation of pyruvate was characterized by a fast linear phase (Fig. 3B), followed by a slow phase. The deviation from linearity in the reaction occurred at a concentration of pyruvate that was less than 1% of the total substrate concentration. This deviation is not generated by the lack of adherence to steady-state conditions, is strongly suggestive of an inactivation process taking place as a consequence of the β-elimination reaction. Two possible mechanisms can be invoked to explain enzyme inhibition: covalent modification of the enzyme, and product inhibition. In the latter case, removal of the products from the reaction mixture should lead to the recovery of enzymatic activity, whereas covalent modification causes permanent inactivation of the enzyme. It is known that, during β-lytic reactions, some aminotransferases become covalently inactivated by a syncatalytic mechanism involving the cofactor and a basic residue in the active site [72,76] (see also Scheme 4). To determine whether this is the case for KATII, the residual activity of the enzyme was measured upon reaction with BCA. KAT II (174 μm), incubated with 50 mm BCA for 20 min at 25 °C, was assayed upon 200-fold dilution, using 10 mm KYN and 10 mm KG. The activity was found to be only 3%, indicating that a significant amount of the enzyme was inactivated as a consequence of the occurrence of the β-elimination reaction.
[ Proposed mechanism for the reaction of KATII with BCA and the syncatalytic inactivation at the stage of the α-aminoacrylate intermediate. X is a nucleophilic amino acid in the active site of the enzyme. Adapted from . ]
BCA is considered to be the best substrate to test for β-elimination reactions. However, KATII β-elimination activity was also evaluated with S-phenylcysteine (SPC). SPC was chosen because cysteine S-conjugates are good substrates for the β-lytic activity of the related enzyme KATI [43,77]. Cysteine S-conjugate β-lyase side reactions can have both negative and positive physiological consequences. Adverse effects may occur as a result of cysteine S-conjugate β-lyases catalyzing reactions that generate toxic sulfur-containing fragments, whereas possible beneficial consequences of cysteine S-conjugate β-lyases activity include pharmacological applications in cancer therapy via the bioactivation of prodrugs into antiproliferative and proapoptotic agents [42,43,78–80]. It was found that the reaction of KATII in the presence of 3 mm SPC produced 210 μm ammonia, whereas the β-lytic activity of Tyr142→Phe KATII was undetectable. We also evaluated whether the natural substrate KYN underwent β-elimination by KATII. The specific activity measured with 20 mm KYN was 9 × 10−6μmol·min−1·μg−1, which is four orders of magnitude lower than that measured with BCA.
Reactivity of KATII with cysteine sulfinate (CSA) and (R)-2-amino-4-(4-(ethylsulfonyl))-4-oxo-butanoic acid (ESBA)
In vivo experiments have indicated that both CSA and ESBA are inhibitors of KATII [55,57,59]. However, their structures (Scheme 3) suggest that they might be substrates for transamination or/and β-elimination. Indeed, CSA is a known substrate of AspAT that is able to catalyze both its transamination [81,82] and β-elimination, with production of sulfinate .
The spectra of KATII (Fig. 4A) and Tyr142→Phe KATII (data not shown) in the presence of CSA exhibited a decrease in the intensity of the band at 360 nm, with the concomitant accumulation of a species absorbing at 330 nm, probably pyridoxamine (PMP). In the presence of KG, CSA transaminated to β-sulfinylpyruvate , as demonstrated by the GOX-coupled assay (data not shown). To further investigate the CSA mechanism of action, KATII activity assays were carried out at 4 mm and 20 mm CSA (Fig. 4B). It was found that CSA inhibited the KATII transamination reaction. Data points were fitted to Eqn (4) with an apparent Vmax of 0.025 ± 0.001 mm·min−1, an apparent Km of 12.3 ± 1.5 mm, and a Kii of 17.2 ± 3.5 mm. The corresponding Ki, calculated from Eqn (5), was 13 μm. The IC50 value reported from in vivo experiments on rats  is ∼ 2 μm.
ESBA is an aromatic compound (Scheme 3) that is structurally analogous to KYN. ESBA absorbed at 287 nm with an extinction coefficient of 2050 m−1 cm−1 (Fig. 5A). ESBA might be either a pure inhibitor, as previously proposed , or, more likely, a substrate analog. We evaluated both the transamination and the β-lytic activities of KATII on ESBA, in the absence and presence of oxoacids, with monitoring of the reaction products, including ammonia. The reaction of ESBA with KATII, in the absence of 2-oxoacids, led to marked changes in the absorption spectrum, with an intensity increase at 283 nm and at ∼ 330 nm (Fig. 5A). In the presence of 10 mm KG, a species absorbing maximally at 338 nm accumulated (Fig. 5A). The amount of ESBA transaminated by KATII at equilibrium was assessed by the GOX-coupled assay, and found to be about 90%. Thus, the main product of the reaction was 4-[4-(ethylsulfonyl)phenyl]-2,4-dioxobutanoic acid, which is characterized by an extinction coefficient at 338 nm of 15 400 m−1·cm−1. Kinetic parameters for the reaction of ESBA with KATII were determined by monitoring the change in absorbance at 338 nm, caused by 4-[4-(ethylsulfonyl)phenyl]-2,4-dioxobutanoic acid accumulation, as a function of time, at different ESBA concentrations. Data were fitted to the Michaelis–Menten equation with K′m = 4.5 ± 0.9 mm and V′max = 7.8 ± 0.6 μm·min−1 (Fig. 5B). The kcat value for the reaction of KATII with ESBA was 9 min−1, only about 2.5-fold lower than the value of 25 min−1 for the reaction with KYN.
Τhe rate of β-elimination was determined by monitoring the formation of ammonia as a function of time for a solution containing KATII, 8 mm ESBA, and 12 mm KG. The reaction was linear within 180 min, with a slope of 2.5 μm·min−1 ammonia (e.g. the specific activity was 25 pmol·μg−1·min−1). This rate is expected to be a lower limit, because, for substrates with poor leaving groups, the transamination reaction, in the presence of 2-oxo acids, is favored with respect to the β-elimination reaction. As a comparison, the reaction of KATII with 5 mm BCA gave a specific activity of 5 nmol·μg−1·min−1, indicating that ESBA is a poor substrate for β-elimination.
We also determined whether ESBA or its reaction products inactivated KAT II, as was observed with BCA. A solution of KAT II (174 μm) was incubated with 8 mm ESBA for 60 min at 25 °C. The reaction was diluted 200-fold in an assay solution containing 10 mm KYN and 10 mm KG. KATII reacted with ESBA was found to be two-fold less active than the unreacted enzyme, suggesting that β-lytic activity of ESBA leads to partial syncatalytic inactivation of the enzyme. The mechanism of inhibition of ESBA on KATII could not be determined, owing to the interference of the ESBA spectrum with the spectroscopic signals used to monitor KATII activity. However, inhibition parameters were further evaluated by an endpoint assay (see below).
A 96-well plate assay for high-throughput screening of KATII inhibitors
Because KATII is a potential target for schizophrenia and other neurological disorders, a high-throughput screening assay was developed to identify KATII inhibitors, and implemented on a 96-well plate format. The assay is based on the determination of the endpoint absorbance intensity at 500 nm, generated from the coupled enzymatic reactions of glutamate oxidase and peroxidase in the presence of o-dianisidine, acting on glutamate produced in the transamination of AAD or other substrates in the presence of KG. This assay is well suited to monitor the transamination of potential substrates and the inhibition caused by the screened compounds. The results of a typical assay are shown in Fig. 6. Incubation of a mixture containing 2.2 μm KATII and 10 mm KYN for 30 min led to the formation of 810 ± 91.9 μm glutamate; that is, 8.1 ± 0.9% of KYN was transaminated within the incubation time. When the reaction was carried out in solution and the transamination was determined directly by the absorption intensity of KYNA (see above), the same degree of KYN transamination was measured. The transamination reaction in the presence of 10 mm AAD (Fig. 6) generated a higher amount of glutamate (1.1 ± 0.0997 mm), owing to the higher catalytic efficiency of KATII towards AAD than to KYN . A mixture of 1 mm ESBA, 200 mm CSA and 2 mmO-phosphoserine (OPS) gave measurable levels of transamination, which were approximately 8 ± 0.78%, 1.4 ± 0.07% and 41 ± 4.2%, respectively, of the level of transamination with AAD (Fig. 6B). Transamination in the presence of either 5 mm BCA or 50 mm 3-nitropropionic acid (3-NPA) was found to be negligible (Fig. 6B). Furthermore, the assay allows identification of compounds that inhibit KATII activity. It was found that the presence of either 1 mm or 100 μm ESBA led to 71 ± 4.9% and 63 ± 1.4% KATII activity inhibition, respectively (Fig. 6B), in good agreement with data previously obtained (64% inhibition at 1 mm ESBA) . CSA, BCA, 3-NPA and OPS inhibition of KATII was also measured (Fig. 6B), and found to be in good agreement with data reported in the literature, showing an IC50 value of approximately 2 μm for CSA , and inhibition of 24% and 38% with 5 mm 3-NPA  and 1 mm OPS , respectively.
Spectroscopic properties of KATII and the Tyr142→Phe mutant
The absorption spectra of KATII and Tyr142→Phe KATII show a band at 360 nm that, on the basis of previous studies on aminotransferases [83–85], is attributed to a Schiff base of the active site lysine (in KATII, Lys263) with a deprotonated imine nitrogen (Fig. 1A; Scheme 3, species 1a). A previous study on bovine and rat KAT II reported absorption spectra with two main peaks at 320–330 nm and 400 nm . Furthermore, KATI shows an absorption spectrum with two bands centered at 335 nm and 422 nm, indicative of a mixture of the PLP and PMP forms of the enzyme or of enolimine and ketoenamine tautomers of the internal aldimine of PLP . PLP is a probe of the active site environment, and the tautomeric distribution therefore reflects the active site polarity. The deprotonated form of the internal aldimine of PLP is typical of aminotransferases, e,g, AspAT . It is well established that the deprotonated form of PLP in transaminases is stabilized by a hydrogen bond with the hydroxyl group of a tyrosine that lowers the pKa of the imine nitrogen by two orders of magnitude [87–89]. On the basis of the KATII structure , Tyr233 is the conserved tyrosine that plays this stabilizing role. In aminotransferases, the deprotonated species is in equilibrium with a protonated form that can be present as either the enolimine tautomer, absorbing at ∼ 330 nm, or the ketoenamine tautomer, absorbing at 420 nm (Scheme 3, species 1b). The pKa of this equilibrium is 6.3 for AspAT . The pKa of the protonation equilibrium of the internal aldimine of KATII was not determined, owing to enzyme instability, but was estimated to be lower than 6.
PLP is an intrinsic fluorescent probe that is often exploited in vitamin B6-dependent enzyme spectroscopy, because of its sensitivity to the environment surrounding the cofactor, and thus to changes in the conformation and ligation state of the active site [90–92]. Several studies have investigated the function and dynamics of aminotransferases with fluorescence techniques [67,68,93–100]. Unlike those of other PLP-dependent enzymes , the emission spectra of KATII and Tyr142→Phe KATII do not show any energy transfer between tryptophans and the cofactor (Fig. 1B). Analysis of the KATII three-dimensional structure  reveals that three tryptophans are within 30 Å of the cofactor, so the absence of energy transfer is an indication of an unfavorable orientation between tryptophan rings and PLP. The deprotonated internal aldimine, absorbing at 360 nm, is fluorescent, with a structured emission showing a maximum at 420 nm and a shoulder at about 500 nm (Fig. 2C). This is in agreement with findings on model PLP Schiff bases and AspAT . Only the emission at 420 nm is directly attributed to the fluorescence emission of the deprotonated internal aldimine; the band at 500 nm might result from direct excitation of the band absorbing at 420 nm. This mechanism is confirmed by the emission spectrum upon excitation at 420 nm, which is centered at 500 nm. Excitation at 330 nm gives a structured emission with a shoulder at about 380 nm, which might originate from the excitation of residual PMP , and a band centered at about 420 nm, which originates from the excitation of the deprotonated internal aldimine band.
Development of activity assays for KATII
KAT activity has usually been assayed by a cumbersome discontinuous method coupled with HPLC determination or liquid scintillation spectrometry of KYNA [41,56–58,69]. The development of a continuous assay for KATII with KYN as substrate has been hampered by its suboptimal spectral properties. In fact, KYN strongly absorbs at 361 nm (ε = 4350 m−1·cm−1) , which does not allow for assays at concentrations higher than 2 mm, the estimated Km of the enzyme being about 5 mm . Our strategy for the development of a continuous assay was based on the observation that only at 310 nm is the absorbance of KYN low enough to allow assays with up to 23 mm KYN in 0.1-cm pathlength cuvettes. At the same time, at 310 nm the absorbance difference between KYN and KYNA is significant, making transamination time courses easily detectable. Kinetic traces collected at 310 nm were linear over the time needed to calculate initial velocities (about 10 min), and showed no lag or burst phases. The extinction coefficient used to calculate V0 was 3625 m−1 cm−1; that is, for each molecule of KYN consumed in the reaction (1049 m−1 cm−1), one molecule of KYNA (4674 m−1 cm−1) is produced. The amount of KYN consumed during the linear portion of the kinetics is less than 5%, thus fulfilling the requirements for steady state. Furthermore, this assay allows for the determination of V0 with a range of substrate concentrations, from 0.3 to more than 2.5 Km, that is acceptable for accurate determination of Vmax and Km . We also checked that the concentration of KG was saturating under our experimental conditions. We estimated KmKG to be lower than 20 μm. At saturating KG concentrations, V′max is equal to Vmax, which has a value of 0.022 mm·min−1, and K′m is equal to αKmKYN, which has a value of 10 mm, where α = k4/k2.
In many cases , the rate-limiting step for transaminases is the α-proton abstraction (Scheme 3); thus, k4 > k2, and K′ represents an upper limit to KmKYN. Consequently, KmKYN is in the millimolar range, in agreement with a previous study on KATII , where a value of ∼ 5 mm was reported. The kcat value, which we found to be about 110 min−1, is in good agreement with the value previously reported, ∼ 134 min−1 . The difference is probably associated with the less precise determination of enzyme active sites with the Bradford method than with cofactor release by alkali denaturation, a method that is accepted to be the most accurate for PLP-dependent enzymes. We should also point out that a continuous assay is usually more precise than a discontinuous assay, especially when the concentration of the product has to be determined via HPLC analysis. In summary, the continuous assay allows the fast determination of enzyme activity, and, more importantly, of inhibition mechanisms (see below). As a further advantage, the amount of enzyme needed for each assay is low, about 9 μg, which is compatible with the low yields of KATII expression.
The high-throughput search for specific and potent enzyme inhibitors takes great advantage of fast, simple, cheap and reliable activity assays. The 96-well plate assay that we have developed for KATII allows fast evaluation of libraries of compounds via an initial visual inspection. The assay can also be exploited for the quantitative comparison of substrate preference and inhibitor selectivity with different isoforms of the KAT superfamily. The assay is very sensitive and is linear in an 80-fold concentration range of glutamate (between 10 and 800 μm). Furthermore, very low amounts of enzyme are used for the assay. The assay was optimized for ∼ 20 μg of KATII per assay and an incubation of 30 min. With increases in the incubation time, the amount of enzyme can be proportionally decreased. Additionally, the assay works with both AAD and KYN, which makes it specific for KATII. In fact, one of the main drawbacks of the discontinuous KATII assay is the lack of isoform specificity . Unlike KYN, which can be used as the amino group donor by all four known KAT isoenzymes, AAD is efficiently used only by KATII [41,103].
Reactivity towards natural and non-natural substrates
KATII was previously indicated to be an ADD aminotransferase. Indeed, the catalytic efficiency of the enzyme towards AAD is slightly higher than that towards KYN . The reaction of KATII with AAD in the absence of ketoacids leads to the transamination of PLP to form PMP (Fig. 1A), which accumulates in the presence of KG. Thus, at equilibrium, the PMP form of KATII is the most stable enzyme form; that is, the rate of its formation is higher than the rate of its consumption. The high absorbance of KYN in the spectral range where PLP and PMP intermediates also absorb hampers performance of the same analysis for the reaction between KYN, KG, and KATII.
PLP-dependent enzymes are reported to be quite promiscuous, with respect to both substrate and the types of reaction catalyzed. It is known that many PLP-dependent enzymes, including tryptophan synthase , 3,4-dihydroxyphenylalanine decarboxylase , AspAT , and serine racemase [107,108], can catalyze side reactions, some of which have been suggested to play physiological roles [107,108]. In particular, aminotransferases are known to be (relatively) efficient in catalyzing β-elimination reactions, especially on substrates with good leaving groups, such as BCA and, more interestingly, on S-substituted cysteine derivatives [42,43,72,73,85,109,110]. In the case of KATII, the exploration of β-lyase activity was particularly intriguing, in that structural comparisons suggested the presence of features normally observed in β-lyases, such as the conformation of the N-terminal region and a tyrosine at position 142, which, in most aminotransferases, is occupied by a tryptophan or a phenylalanine . Thus, we investigated the β-lytic activity of KATII and Tyr142→Phe KATII with BCA and SPC. Whereas the Tyr142→Phe mutation does not change the reactivity of KATII towards AAD and KYN in the presence of KG, it heavily influences the β-lytic activity of the enzyme. In fact, both the wild type and the mutant are able to eliminate chloride from BCA, with the production of ammonia and pyruvate, but the Tyr142→Phe mutant is 20-fold less efficient. One possible explanation is that the tyrosine at position 142 plays a role in the balance between β-elimination and transamination. In the conditions tested here, we could not reach saturation in a plot of initial velocity against BCA concentration. This is consistent with the observation that Km values of transaminases for BCA are usually very high, preventing the determination of kinetic parameters. This also hampers the calculation of catalytic efficiency for chloride elimination by KATII and Tyr142→Phe KATII. For this reason, we cannot rule out the possibility that the lower rate of β-elimination observed for Tyr142→Phe KATII is partly attributable to a higher Km value. Although a detailed characterization of the β-lytic activity of this mutant is outside the scope of this work, we checked the β-lytic activity of wild-type KATII and the mutant enzyme in the presence of SPC. Also in this case, we observed β-elimination with production of ammonia by the wild-type enzyme but no activity of the Tyr142→Phe mutant, a further indication of a reduction of β-lytic activity brought about by the mutation. Interestingly, the β-elimination, although with a very low ratio with respect to transamination, also takes place on the natural substrate KYN. This unusual β-elimination reaction should produce o-aminobenzaldehyde, as already reported in a controversial paper on the activity of kynureninase . At present, it is not known whether this reaction has any physiological significance or is regulated by any effector, as is the case, for example, with the mammalian serine racemase [107,108]. However, one should bear in mind that the β-elimination reaction requires the formation of an α-aminoacrylate intermediate that, in the case of KATII, as demonstrated by experiments with BCA, leads to concomitant syncatalytic inactivation of the enzyme. Furthermore, it is unlikely that a very inefficient reaction, when compared to the main one, would have any physiological significance, unless it is tuned by effectors and ligands. The understanding of this aspect of the KATII mechanism of action is beyond the aim of this work, but deserves further attention.
As expected on the basis of the ESBA structure, KATII catalyzes both transamination and β-elimination of this compound. A rough estimate of the catalytic efficiency of the β-elimination reaction, based on specific activity at a fixed substrate concentration, indicates that ESBA, like KYN, is a poor substrate for β-elimination, when compared with BCA. However, both ESBA and BCA are capable of permanently inactivating KATII through a mechanism that probably involves the formation of a covalent adduct between the active site lysine and the α-aminoacrylate intermediate, as already reported for alanine aminotransferase (AlaAT)  and AspAT  (Scheme 4). We were unable to recover a PLP–ESBA derivative, either after ultrafiltration or after gel filtration on microspin columns (data not shown). This hampered the determination of the type of modification by MS. The dependence of the percentage of initial activity of KATII on time of incubation in the presence of 5 mm BCA (data not shown) gives an exponential decay with kobs = 0.7 min−1 (t1/2 ∼ 1 min). In the case of AlaAT, the pseudo-first-order rate constant for inactivation was 0.36 min−1, with t1/2 ∼ 2 min, in the presence of 5 mm BCA . The partition ratio (moles of product per mole of inactivated enzyme) is about 500, a value comparable with those found for AlaAT, 1050  and for kynureninase, 530 .
ESBA and CSA – inhibitors and/or substrates?
Because of the potential role of KATII as a drug target in the treatment of psychiatric disorders, such as schizophrenia, many efforts have been made to design selective inhibitors. However, only a few molecules have been proved to inhibit KATII activity: CSA on brain slices of rats , ESBA in reverse dialysis experiments on rat hippocampus , MPP+ and 3-NPA on both cortical brain slices and partially purified KAT , and BFF-122  (Scheme 2). Although MPP+ was shown to be able to discriminate between KATI and KATII, stimulating the design of isoform-specific inhibitors, the use of this compound triggers Parkinsonian symptoms . Thus, at present, ESBA and BFF-122 are the only available specific and potent KATII inhibitors . Moreover, ESBA was found to be pharmacologically active on rats but almost inactive on humans [55,57]. It has been supposed that this difference in inhibitory activity may arise from the presence in the catalytic site of human KATII of two hydrophobic residues, Leu40 and Pro76, which are replaced by polar serines in the rat enzyme . Recently , a human KATII double mutant harboring the serines characterizing the rat ortholog active site was generated in order to investigate the molecular basis for ESBA species specificity. The site-directed mutagenesis approach did not provide any experimental support to explain the striking difference in ESBA inhibitory efficiency towards rat and human KATII, and underlined the need for more in-depth biochemical investigations aimed at deciphering the mechanism of ESBA inhibition. The mechanisms of ESBA and CSA inhibition have not been previously investigated. We have demonstrated that both CSA and ESBA are substrate analogs, and not purely competitive inhibitors. In the case of ESBA, the mechanism of inhibition could not be determined, owing to its strong interference with almost all of the assays that we used. However, the spectroscopic signal generated by the accumulation of the product of the reaction of ESBA with KAT was exploited to calculate the kinetic parameters and, thus, an approximate affinity of ESBA for the enzyme. Although the identity of the product could not be assessed by MS, it seems likely, from its spectroscopic properties, that it represents the α-ketoacid generated by ESBA transamination. An apparent Km of 4.5 mm is in good agreement with 64% inhibition at 1 mm ESBA .
CSA is a physiological substrate of AspAT [115,116], owing to its structural similarity with aspartate. Inhibition of KATII by CSA was determined to be uncompetitive, a quite unexpected finding, considering that CSA is also a substrate of KATII, converting the PLP form of the enzyme to the PMP form. Uncompetitive inhibition involves the exclusive (or predominant) binding of the inhibitor to the enzyme–substrate complex or to any intermediate downstream of it. In the case of CSA, uncompetitive inhibition arise from preferential binding of CSA to the PMP form of KATII, suggesting that CSA might mimic KG better than KYN or AAD.
In conclusion, although both CSA and ESBA show good inhibitory properties on KATII, with, at least for CSA, inhibition constants in the micromolar range, these molecules are actually substrates. New hints for the development of a KATII-specific inhibitor come from the recent observation that large molecules with bulky substituents specifically bind to the II-isoform of the enzyme [61,62], probably as a consequence of the higher mobility of the N-terminal domain of KATII than those of other aminotransferases . We believe that these observations, together with the tools developed here for the high-throughput screening of KATII-specific inhibitors, will aid in the development of anti-schizophrenic and precognitive drugs.
All chemicals were purchased from Sigma–Aldrich (St Louis, MO, USA), and were used as received. ESBA was synthesized as previously described [55,57]. Experiments, unless otherwise specified, were carried out in 50 mm Hepes buffer (pH 7.5) at 25 °C.
Protein expression and purification
Human KATII and the mutant Tyr142→Phe were expressed and purified as previously described . The enzyme was fully saturated with PLP by addition of a 10-fold molar excess of cofactor, followed by extensive dialysis against a solution containing 20 mm Hepes and 50 mm NaCl (pH 8.0), and stored in small aliquots at −80 °C.
Absorption spectra were collected with a Cary 400 spectrophotometer (Varian, Cary, NC, USA). Spectra were corrected for buffer contributions. The extinction coefficients of wild-type KATII and Tyr142→Phe KATII were calculated from the amount of PLP released upon alkali denaturation, by the method of Peterson .
Fluorescence emission spectra were collected with a Spex Fluoromax-2 fluorimeter (Jobin-Yvon, North Edison, NJ, USA) by exciting the emission of tryptophans at 298 nm or the emission of the cofactor at 330, 360 and 420 nm. Excitation and emission slits were set at 1.5 nm (λex = 298 nm), 3.5 nm (λex = 330 nm and λex = 360 nm), and 4 nm (λex = 420 nm). Spectra were corrected for buffer contributions.
A continuous assay for the measurement of transamination of KYN by KATII with KG as the acceptor was developed (see Results). KYN and KYNA concentrations were calculated from the absorbance in the UV–visible range, using published extinction coefficients  (for KYN, ε257 nm = 6750 m−1·cm−1, and ε361 nm = 4350 m−1·cm−1; for KYNA, ε332 nm = 9800 m−1·cm−1, and ε344 nm = 7920 m−1·cm−1).
β-Elimination reactions carried out by aminotransferases involve formation of an α-aminoacrylate intermediate that rapidly decomposes to pyruvate and ammonia. Pyruvate accumulation was measured at 220 nm. Time courses were monitored at 25 °C in 0.1-cm pathlength cuvettes. The assay was carried out in 100 mm phosphate buffer (pH 7.5), to minimize the background signal at 220 nm.
The ammonia concentration formed in reaction mixtures was determined with Nessler’s assay , using a ready-to-use solution (Fluka, Buchs, Switzerland; code 72190). At pH 7.5, 98% of ammonia is in the NH4+ form, and the solubility of NH3 in water at 25 °C is about 50% (w/w); thus, no significant loss of ammonia by evaporation was expected under these experimental conditions. A calibration curve was constructed with ammonium sulfate concentrations in the range 25 μm to 4 mm. Forty microliters of a solution containing ammonia was diluted with 860 μL of water. One hundred microliters of Nessler’s reagent were added to the mixture, and the absorbance of the solution was immediately recorded at 436 nm.
KAT activity measured with a GOX-coupled assay
A reaction mixture containing 10 mm AAD or 10 mm KYN, 10 mm KG, 40 μm PLP, 2.3 μm KATII and 50 mm Hepes (pH 7.5) was incubated at 25 °C for 30 min. The reaction was stopped by addition of 14 mm phosphoric acid. Then, 0.015 units of GOX (GOX-Sigma G5921, St. Louis, MO, USA), 2.25 units of peroxidase (perox-Sigma P8375) and 0.75 mmo-dianisidine were added to the solution. The mixture was incubated at 37 °C for 90 min. Sulfuric acid (3.36 mm) was added, and the absorbance was measured at 530 nm. This assay was adapted to a 96-well plate format, with absorbance intensities being measured at 500 nm with a home-made plate reader. Blanks contained the same reagents as for the assay, except that KATII solution was replaced by the same volume of buffer. A calibration curve was constructed with known concentrations of glutamate, ranging from 10 to 800 μm, in the presence of 10 mm KG, 40 μm PLP, and 14 mm phosphoric acid. Inhibition assays were set up, with either CSA, BCA, ESBA, 3-NPA or OPS being added to the mixture. CSA, CSA, 3-NPA, OPS and ESBA at the same concentrations used in the inhibition assay were also tested for transamination activity in the presence of 10 mm KG.
Data fitting was carried out with sigma plot software, release 9.0. Initial velocities as a function of KYN concentration at a constant KG concentration (10 mm) were fitted to a hyperbolic equation:
where V′max and K′ are apparent Vmax and Km, α is k4/k2 (see reaction mechanism in Discussion), and KmKYN and KmKG are the Michaelis constants for KYN and KG, respectively .
The parameters for the reaction of KATII in the presence of CSA were determined by globally fitting data to Eqn (4):
where Km app and Vmax app are apparent Km and apparent Vmax, and Kii is the dissociation constant for dissociation of the inhibitor from the enzyme–substrate complex. Because the KYN concentration used in the assay is close to Km, an approximated value of Ki can be obtained from the following:
This work was supported by a grant from MIUR (COFIN to A. Mozzarelli).