Analytical Biochemistry 409 (2011) 183–188

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Development of a microplate fluorescence assay for kynurenine aminotransferase

Jacky Wong, William J. Ray, Anna Y. Kornilova ⇑Neurology Department, Merck, West Point, PA 19486, USA

a r t i c l e i n f o

Article history:Received 24 May 2010Received in revised form 29 October 2010Accepted 29 October 2010Available online 6 November 2010

Keywords:FluorescenceKynurenine aminotransferaseHTSEnzymatic assayInhibitors

0003-2697/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.ab.2010.10.037

⇑ Corresponding author. Fax: +1 215 652 2075.E-mail address: anna_kornilova@merck.com (A.Y.

1 Abbreviations used: KYNA, kynurenic acid; NMDAkynurenine hydroxylase; L-KYN, L-kynurenine; KAT, khuKAT, human KAT; HPLC, high-performance liquidthroughput screening; GTK, glutamine transaminase Kb-lyase; L-Phe, L-phenylalanine; AMP, 2-amino-2-meth50-phosphate hydrate; Na pyruvate, sodium pyruvate;sodium acetate; AEBSF, 4-(2-aminoethyl) benzenesulKRB buffer, Krebs–Ringer bicarbonate buffer; cDNA,dimethyl sulfoxide; S/B ratio, signal-to-background raratio; SD, standard deviation; DMEM, Dulbecco’s modifibovine serum.

a b s t r a c t

Inhibition of kynurenine aminotransferases (KATs) is a strategy to therapeutically reduce levels ofkynurenic acid (KYNA), an endogenous antagonist of glutamatergic N-methyl-D-aspartate (NMDA) andcholinergic a7 nicotinic receptors. Several methods of measuring KAT activity in vitro have been devel-oped, but none is well-suited to high throughput and automation. In this article, we describe a modifica-tion of existing high-performance liquid chromatography (HPLC)-based methods that enables thedevelopment of a 96-well microplate assay in both enzyme- and cell-based formats using human KATI as an example. KYNA enzymatically produced from L-kynurenine is measured directly in a reaction mix-ture fluorimetrically.

� 2010 Elsevier Inc. All rights reserved.

Dysregulation of the kynurenine pathway of tryptophan metab-olism in the brain has been linked to multiple neurodegenerativeand psychiatric diseases. In these disorders, either increased or de-creased levels of kynurenic acid (KYNA)1 are correlated with vari-ous disease states. Because KYNA is an antagonist of both the a7

nicotinic acetylcholine receptor and the N-methyl-D-aspartate(NMDA) glutamate receptor [1], altered KYNA levels could signaldysregulation of neuronal excitability, synaptic plasticity, or neuro-protection. Therefore, modulating this pathway using small mole-cule inhibitors is an emerging strategy for the treatment of avariety of conditions. Existing strategies, such as kynurenine hydrox-ylase (KMO) inhibition, are focused on increasing KYNA as a poten-tial neuroprotective approach for diseases such as Huntington’sdisease, Parkinson’s disease, ischemia, and epilepsy [2]. However,chronically increased levels of KYNA can potentially lead to hypo-function of glutamatergic and cholinergic neurotransmission, ashas been proposed to occur in Alzheimer’s disease [3], schizophrenia

ll rights reserved.

Kornilova)., N-methyl-D-aspartate; KMO,ynurenine aminotransferase;chromatography; HTS, high-; CCBL1, cysteine S-conjugateyl-1-propanol; PLP, pyridoxal-

ZnOAc, zinc acetate; NaOAc,fonyl fluoride hydrochloride;complementary DNA; DMSO,tio; S/N ratio, signal-to-noiseed Eagle’s medium; FBS, fetal

[4,5], and Down syndrome [6]. Thus, it is important to develop andunderstand the properties of inhibitors that reduce KYNA.

KYNA is a product of the irreversible transamination of L-kynu-renine (L-KYN) catalyzed by kynurenine aminotransferases (KATs).Hence, inhibition of KATs is an attractive strategy for normalizationof increased levels of KYNA in the diseased brain. Currently, thereare four KATs in human and rodent brains, with KAT I and KAT IIbeing the most studied [7]. The ability to pharmacologically mod-ulate KATs has been limited, and only a rat-specific KAT II inhibitorwith poor potency had been reported [8] until recently. A differentrat KAT II inhibitor was discovered recently [9], and it was reportedto also have activity against human KAT (huKAT) II [10]. The dis-covery of KAT inhibitors has been hampered by the inability touse existing high-performance liquid chromatography (HPLC)-based KAT assays for high-throughput screening (HTS). KAT I wasinitially identified as glutamine transaminase K (GTK) and laterwas shown to be identical to cysteine S-conjugate b-lyase (CCBL1)(for a review, see Ref. [11]). An absorbance-based KAT I (GTK,CCBL1) assay was developed around this enzyme’s ability totransaminate L-phenylalanine (L-Phe) to a light-absorbing phenyl-pyruvate (322 nm, e = 2.4 � 104 in 3 M NaOH) [12]. The major dis-advantages of this assay are the following: many compoundsabsorb at this wavelength, the assay cannot be used for assayingother KATs that do not transaminate L-Phe, and the assay requireshighly concentrated NaOH for detection. Here, using huKAT I as anexample, we present the development of a simple HTS-compatiblemicroplate fluorescence-based KAT enzymatic assay that enablesthe discovery of potential drugs for neurological diseases and toolsfor further elucidation of the role of the kynurenine pathway inbrain function.

184 Microplate fluorescence assay for KAT / J. Wong et al. / Anal. Biochem. 409 (2011) 183–188

Materials and methods

Chemicals

2-Amino-2-methyl-1-propanol (AMP), pyridoxal-50-phosphatehydrate (PLP), sodium pyruvate (Na pyruvate), L-KYN sulfatesalt, zinc acetate (ZnOAc), sodium acetate (NaOAc), Tween 20,4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF),benzonase, HPLC-grade water and acetonitrile, L-glutamine, andKrebs–Ringer bicarbonate buffer (KRB buffer) were purchased fromSigma (St Louis, MO, USA).

huKAT I expression and purification

To generate KAT I enzyme for biochemical experiments, full-length human complementary DNA (cDNA), encoding an additional6 histidine residues (His6) in frame at the C terminus, was expressedin insect cells and purified by nickel affinity chromatography essen-tially as described by Rossi and coworkers [13]. Briefly, insect Sf21cells were infected with huKAT I-C-His6-carrying baculovirus andused to generate 10 L of cell suspension, from which cells wereresuspended in 150 ml of lysis buffer (25 mM sodium phosphatebuffer [pH 7.4], 0.1 mM PLP, 25 lg/ml benzonase, and 1 mM AEBSF),incubated on ice for 45 min, and lysed using microfluidizer twice.The lysate was centrifuged at 18,000g for 45 min, and the superna-tant was loaded onto a nickel affinity column for His6 pull-down.The fractions containing huKAT I from the nickel affinity columnwere identified by Western blotting using anti-KAT I antibody (Gen-way, product no. 15-288-21989B) and then pooled, and the salt con-centration of the pool was adjusted to 3 M NaCl. The pool waspurified over a phenyl Sepharose column to eliminate impurities,and the huKAT I-containing fractions were concentrated and furtherpurified on a size exclusion column (Superdex 200). The purifiedprotein was concentrated to 1 mg/ml and flash frozen.

huKAT I in vitro activity assay

Reactions were performed in Costar black 96-well untreatedplates (Fisher Scientific). According to final optimized conditions,100 ll of reaction mixture contained 10 ll of compound or solvent(dimethyl sulfoxide [DMSO]), 0.3 ng/ll huKAT I, 90 lM l-KYN,0.001% Tween 20, 1 lM PLP, and 1 mM Na pyruvate in 150 mMAMP buffer at pH 9.5. Compounds were preincubated with reactioncomponents for 15 min before the addition of L-KYN and then wereincubated for 4 h at 37 �C.

Measurement of KYNA using HPLC

Enzymatic reactions were quenched with 10 ll of 1 N HCl. Sam-ples then were mixed with mobile phase (50 mM NaOAc and 5%acetonitrile, pH 6.2) at a 1:1 ratio, and 40 ll was injected into thecolumn. The HPLC system consisted of a Waters 2795 Alliance sep-aration module and a Waters 2475 multiwavelength detector (Mil-ford, MA, USA) set at an excitation wavelength of 344 nm and anemission wavelength of 398 nm. The mobile phase was deliveredat a flow rate of 1.5 ml/min, and 500 mM ZnOAc was delivered post-column by an external Shimadzu LC-20AD pump (Columbia, MD,USA) at a flow rate of 0.5 ml/min. The column was a Waters XBridgeShield RP18 column (3.5 lm, 4.6 � 50 mm). A retention time of2.5–3.3 min for KYNA was observed under these conditions. KYNApeaks were processed using Waters Empower Analysis software.All measurements were performed in duplicates.

Measuring KYNA in a microplate

Here 100 ll of the buffer containing 50 mM NaOAc and 350 mMZnOAc at pH 5.47 was added directly to each enzymatic reaction

sample, and fluorescence was measured on a SpectraMax platereader (Molecular Devices, Sunnyvale, CA, USA). All measurementswere done in duplicates. The signal-to-background (S/B) ratio, sig-nal-to-noise (S/N) ratio, and Z0 factor for a 96-well plate formatwere calculated as follows: S/B ratio = (mean signal)/(mean back-ground), S/N ratio = (mean signal �mean background)/standarddeviation (SD) of background, and Z0 factor = [1 � 3 � (SDsignal +SDbackground)]/(mean of signal �mean of background) [14].

Generation of SH-SY5Y cell line stably expressing huKAT I

A plasmid-carrying huKAT I was purchased from Origene (prod-uct no. SC114349), and huKAT I was subcloned into Origene’spCMV-Neo plasmid via NotI restriction digestion. Sequence of thefinal product was confirmed by DNA sequence analysis (GENEWIZ).SH-SY5Y cells were grown in Dulbecco’s modified Eagle’s medium(DMEM) supplemented with 10% fetal bovine serum (FBS), 1% non-essential amino acids, and 1% penicillin/streptomycin. Cells(1 � 106) were transfected (2 lg of plasmid) by electroporationusing an Amaxa nucleofector (Lonza, Switzerland) and were platedinto 6-well plates. Then, 24 h after transfection, 400 lg/ml neomy-cin (Geneticin) was added to select for a pool of cells stablyexpressing huKAT I (7 days). The protein expression level was ana-lyzed by Western blotting using anti-KAT I antibody (Genway,product no. 15-288-21989B).

huKAT I cell-based assay

SH-SY5Y cells stably expressing huKAT I were seeded in a 96-well collagen-coated plate and allowed to grow overnight. The fol-lowing day, growth medium was removed and replaced with100 ll of compound resuspended in KRB buffer (pH 7.0), 2% FBS,and 25 lM L-KYN. Plates were placed in an incubator (37 �C, 5%CO2) overnight. Then 50 ll of the conditioned cell medium wastransferred to a new plate containing 50 ll of 250 mM ZnOAc.Fluorescence was measured on a SpectraMax (Molecular Devices)or Envision (PerkinElmer) plate reader. HPLC measurements ofKYNA in condition media were performed similarly to HPLC mea-surements of KYNA in in vitro enzymatic reactions as describedabove.

Results and discussion

Existing HPLC-based methods for the quantification of KYNA asa product of KAT enzymatic activity are based on chromatographicseparation of KYNA on an HPLC column, followed by its detectionbased on either its electrochemical properties or the fluorescentproperties as a complex with Zn2+ (kex = 344 nm, kem = 398 nm)[15,16]. KYNA quantification by HPLC via fluorescence remainsthe more sensitive method, especially for measuring KYNA levelsin brain and cerebrospinal fluid (CSF), where the levels of KYNAare in the low nanomolar (nM) range [3]. However, we observedthat under the assay conditions for measuring KAT activityin vitro, where low micromolar (lM) amounts of KYNA are beingformed, the only fluorescent species strongly emitting at 398 nmis the KYNA–Zn2+ complex (Fig. 1A and B). Hence, we recognizedthat chromatographic separation of the product is unnecessaryand that fluorescence can be measured directly in the reactionmixture in the microplate.

Here, using KAT I as an example, we describe the developmentof a microplate fluorescence KAT assay for a 96-well format,although the assay is also suitable for miniaturization to 384-and 3456-well formats for HTS application. One of the criteria forHTS assay is to provide the best acceptable S/B ratio while usingthe minimal amount of reagents for which supply is limited (e.g.,

Fig.1. HPLC traces showing that KYNA–Zn2+ complex is the major fluorescent species in KAT enzymatic reaction mixture and in cell-based assay. (A, B) KYNA is produced in areaction mixture containing 0.3 ng/ll huKAT I, 90 lM L-KYN, 1 lM PLP, and 1 mM Na pyruvate in AMP buffer at pH 9.5 and 37 �C for 4 h in the absence (A) or presence (B) of5 mM L-glutamine inhibitor. (C, D) KYNA is produced in SH-SY5Y cells stably expressing huKAT I in the presence of 25 lM L-KYN in KRB buffer at pH 7.0 for 24 h in the absence(C) or presence (D) of 5 mM L-glutamine inhibitor. Data are representative of repeat experiments. EU, emission units.

Microplate fluorescence assay for KAT / J. Wong et al. / Anal. Biochem. 409 (2011) 183–188 185

enzyme). Thus, we chose to test KAT I activity at pH 9.5 becauseKAT I exhibits its highest activity at this pH [17,18], warrantingthe use of the minimal amount of enzyme. The enzymatic reactionwas performed at this pH using AMP buffer. We avoided using Trisbuffer due to the reports that Tris interferes with KAT I activity[13,17]. We used purified recombinant huKAT I as a source of theenzyme. Besides the substrate L-KYN, other components of thereaction mixture were the KAT cofactor PLP and Na pyruvate,which is an oxoacid required for PLP regeneration. Although pyru-vate has been shown to be only one of the oxoacids supporting KATI activity [17], we chose it based on published assay conditions[18]. However, other oxoacids may be used following additionaloptimization. Finally, we discovered that the addition of 0.001%Tween 20 to the reaction mixture improved the signal 4-fold.The addition of a low concentration of detergent is often used inHTS because it is thought to reduce the amount of enzyme stickingto the walls of the well and, in turn, improves the availability of ac-tive enzyme.

Zn2+ buffer pH optimization

The key step in enabling direct KYNA detection in a microplatewas the adjustment of the pH of the Zn2+-containing buffer, whichwas added to the enzymatic reaction mixture at the end of thereaction for KYNA detection. In the conventional HPLC method,the enzymatic reaction is usually stopped by acidification andthe samples are subsequently analyzed on HPLC using the Zn2+-containing mobile phase at pH 6.2. In a microplate assay, theZn2+-containing reagent is added directly to the enzymatic reac-

tion. On this addition, the final pH cannot be higher than 7.0 dueto Zn(OH)2 precipitation at basic pH, but at the same time theKYNA–Zn2+ complex fluorescence decreases as pH becomes moreacidic. Thus, if the pH is not adjusted appropriately, signal will belost either to Zn(OH)2 precipitation or to decreased KYNA–Zn2+

complex fluorescence. To bring the pH 9.5 of the AMP reactionmixture to pH 6.0 on 1:1 mixing with Zn2+-containing buffer, thelatter needs to be at pH 5.4. When measured in a microplate, theKYNA–Zn2+ fluorescence signal was linearly proportional for awide range of KYNA concentrations (Fig. 2), thereby fulfilling therequirement for detection system linearity.

Optimization of PLP and L-KYN concentrations

We initially started the assay development using conditions,such as 40–80 lM PLP and 3–5 mM L-KYN, from previously re-ported assays [17–19]. Interestingly, under these conditions, wehave observed a substantial spontaneous KYNA formation in no-enzyme control samples, suggesting that those conditions werenot optimized for minimizing spontaneous KYNA formation[17,19]. This spontaneous activity is potentially an important con-founding reaction because the nonenzymatic production of KYNAwill be insensitive to KAT inhibitors and would affect kinetic anal-yses of enzyme activity. The general phenomenon of nonenzymatictransamination reactions in the presence of PLP and oxoacids wasreported half a century ago [20], and we observed a steep depen-dence of spontaneous KYNA formation on PLP concentration atpH 7.5 (Fig. 3), the pH at which KATs are usually assayed in biolog-ical samples. Spontaneous KYNA formation was pH sensitive and

0 100 200 300 4000

250

500

750

1000

1250

1500

[KYNA] (nM)

Fluo

resc

ence

(AU

)

Fig.2. KYNA–Zn2+ complex fluorescence (in arbitrary units [AU]) is linear over awide range of KYNA concentrations when measured in a microplate. The plot isobtained for various KYNA concentrations in solution containing 150 mM AMPbuffer (pH 9.5), 1 lM PLP, 1 mM Na pyruvate, and 0.001% Tween 20 mixed with50 mM NaOAc/350 mM ZnOAc reagent (1:1, 100 ll/100 ll). Fluorescence wasmeasured in a 96-well plate. Background fluorescence at 0 lM KYNA is due to aplate and the reaction components. All measurements were done in duplicates, anddata are representative of repeat experiments.

spontaneous KYNA formation

pH7.5 pH9.50

10

20

30

40

50

6090 µM L-KYN5 mM L-KYN

3

53

KYN

A (f

old

chan

ge)

Fig.4. Spontaneous formation of KYNA strongly depends on L-KYN concentration atpH 9.5 and less so at pH 7.5. Reactions were performed in AMP buffer in thepresence of 1 lM PLP and 1 mM Na pyruvate at 37 �C for 4 h with either 5 mM or90 lM L-KYN. All measurements were done in duplicates, and data are represen-tative of repeat experiments.

186 Microplate fluorescence assay for KAT / J. Wong et al. / Anal. Biochem. 409 (2011) 183–188

was not observed at pH 9.5 under our standard conditions. How-ever, spontaneous KYNA formation was also sensitive to theL-KYN concentration in the assay, and the L-KYN dependence wasgreatly increased at pH 9.5 (Fig. 4). These findings suggest thatoptimizing KAT-dependent KYNA production requires the carefultitration of PLP and L-KYN at relevant pH values, especially whenlong reaction times are used such as in the case of measuringKAT activity in biological samples. It is also clear that althoughspontaneous KYNA formation can be monitored and subtractedfrom enzyme-dependent signals as appropriate, it should be mini-mized when kinetic analysis is performed. The nonenzymatic for-mation of KYNA has not been uniformly accounted for inpublished studies on KAT kinetics, thereby requiring careful inter-pretation [17,19].

Using our optimized microplate assay, we performed assaycharacterization and kinetic analysis of huKAT I enzyme at pH9.5 at 1 lM PLP and 1 mM Na pyruvate. The production of KYNAwas linear with respect to time (Fig. 5) and substrate concentration(Fig. 6). We estimated Km to be 108 ± 24 lM and kcat to be7.3 ± 0.72 min�1 (Fig. 7). This Km value is 10 times lower than the

0 25 50 75-10

0

10

20

30

40

50

60

90 µM L-KYN pH 7.5

90 µM L-KYN pH 9.5

5 mM L-KYN pH 7.5

5 mM L-KYN pH 9.5

[PLP] (µM)

KYN

A (f

old

chan

ge)

Fig.3. PLP-dependent spontaneous nonenzymatic formation of KYNA. Significantdependence is observed at pH 7.5 but not at pH 9.5. Reactions were performed inAMP buffer in the presence of 1 mM Na pyruvate at 37 �C for 4 h with either 5 mMor 90 lM L-KYN. All measurements were done in duplicates, and data arerepresentative of repeat experiments.

one reported previously for purified huKAT I [17] but is similarto a value reported by another group for partially purified huKATI [18]. Based on these parameters, the concentration of L-KYNwas chosen to be 90 lM and enzyme, which is the costliest reagentin the reaction, was optimized to 0.3 ng/ll so as to provide a 5–8-fold S/B ratio for the 4-h enzymatic reaction.

The quality of the 96-well plate assay was assessed by calculat-ing S/B ratio, S/N ratio, and Z0 factor, a description of assay perfor-mance that measures signal to noise while accounting forvariability at both low and high signals [14]. The mean S/B ratioin a whole plate experiment was 5.5, the S/N was 24, and the meanZ’ factor was 0.65, indicating excellent reproducibility and reliabil-ity. DMSO presence up to 10% did not affect the product formation(data not shown), allowing the testing of inhibitor concentrationsas high as 1 mM.

HPLC versus microplate IC50 determination

The dose–response curves for the KAT I inhibitor L-glutamine,which competes with L-KYN as a substrate, were obtained for com-parison using both the method described here and a conventionalHPLC fluorescence method described previously [15]. IC50 valuesfor L-glutamine were similar [20–30 lM or Ki = IC50/(1 + S/Km) of10–15 lM] when obtained by both methods (Fig. 8), indicating

0 50 100 150 200 2500

500

1000

1500

2000

2500

3000 0.5 ng/µl KAT I0.25 ng/µl KAT I0.125 ng/µl KAT I

Time (min)

Fluo

resc

ence

(AU

)

Fig.5. KYNA production is linear with respect to time. huKAT I activity wasmeasured at various enzyme concentrations (0.125, 0.25, and 0.5 ng/ll) and 50 lML-KYN in 150 mM AMP buffer (pH 9.5) with 1 lM PLP, 1 mM Na pyruvate, and0.001% Tween 20. All measurements were done in duplicates, and data arerepresentative of repeat experiments. AU, arbitrary units.

0 50 100 150 2000

500

1000

1500

2000

2500

3000

35000.5 ng/µl KAT I0.25 ng/µl KAT I0.125 ng/µl KAT I

L-KYN (µM)

Fluo

resc

ence

(AU

)

Fig.6. KYNA production is linear with respect to substrate concentration. Substratetitrations at various huKAT I enzyme concentrations were obtained in 150 mM AMPbuffer (pH 9.5) with 1 lM PLP, 1 mM Na pyruvate, and 0.001% Tween 20. Reactiontime was 4 h. Background fluorescence at 0 lM L-KYN is due to plate and reactioncomponents but not to KYNA. All measurements were done in duplicates, and dataare representative of repeat experiments. AU, arbitrary units.

0 100 200 300 4000

10

20

30

40

50

60

70 0.5 ng/µl KAT I0.25 ng/µl KAT I0.125 ng/µl KAT I

KYN (µM)

V (n

m K

YNA

/min

)

Fig.7. Kinetic analysis of huKAT I using the microplate assay. KYNA production wasmeasured in 150 mM AMP buffer (pH 9.5) with 1 lM PLP, 1 mM Na pyruvate, and0.001% Tween 20. The kinetic parameters were calculated by fitting the experi-mental data to the Michaelis–Menten equation. Data are representative of repeatexperiments.

1 10 100 1000-20

0

20

40

60

80

100plate

HPLC

log[L-glutamine] (µM)

% K

AT

I inh

ibiti

on

Fig.8. Dose–response curves for L-glutamine inhibition of purified huKAT I at pH9.5. HPLC analysis was performed using samples employed in microplate analysisafter termination of the reaction by the addition of 1 M HCl. KAT I inhibition (%) isestimated in relation to the inhibitor-free reaction mixture (10% DMSO). Allmeasurements were done in duplicates, and data are representative of repeatexperiments.

1 10 100 1000-20

0

20

40

60

80

100

log[L-glutamine] (µM)

log[L-glutamine] (µM)

% K

AT

I inh

ibiti

on

1 10 100 10000

1000

2000

3000

4000

5000

KYN

A (A

U)

Fig.9. Dose–response curve for L-glutamine obtained in microplate cell-basedassay. L-Glutamine was incubated with SH-SY5Y cells stably expressing huKAT I for24 h, and KYNA was measured in a plate after mixing condition media with Zn2+-containing buffer. KAT I inhibition (%) is estimated in relation to the inhibitor-freereaction mixture (1% DMSO). Inset: Same data plotted in KYNA–Zn2+ fluorescence(in arbitrary units [AU]) demonstrating the assay S/B ratio being at 7-fold. Allmeasurements were done in duplicates, and data are representative of repeatexperiments.

Microplate fluorescence assay for KAT / J. Wong et al. / Anal. Biochem. 409 (2011) 183–188 187

that plate format readout produces IC50 estimates comparable tothose determined by HPLC.

Microplate cell-based assay development

Testing compounds that are active in enzymatic huKAT I assaysin the cell-based assay is an important way to predict in vivo activ-ity given that the active compound must now pass through the cellmembrane and inhibit huKAT I at physiological pH. Similar to thein vitro enzymatic KAT assay, an HPLC-based method has beenused for quantification of KYNA production and inhibitor testingin cells [21,22]. To quickly test compounds for their ability to inhi-bit KYNA production in cells, we developed a microplate cell-basedassay using SH-SY5Y cells stably expressing huKAT I. The overex-pression of huKAT I in the cell line was more than 20-fold (datanot shown). We observed by HPLC that when these cells are incu-bated with L-KYN, the dominant species that was highly fluores-cent at 398 nm was the KYNA–Zn2+ complex (Fig. 1C and D),indicating that chromatographic separation of the KYNA is unnec-essary and that fluorescence can be measured directly in the reac-tion mixture in the microplate.

Assay optimization resulted in providing a 6–8-fold S/B ratio fora 24-h reaction using 25 lM L-KYN, and L-glutamine titrationyielded an IC50 of 75 ± 6 lM (n = 4) (Fig. 9). KYNA production doesnot reach saturation until after 24 h when 25 lM L-KYN is used(data not shown). A long incubation time harbors significant risksin compound screening, including formation of active breakdownproducts, cellular toxicity, and compound instability. These risksare balanced by the benefits of the longer incubation; it allowsproduct to accumulate to readily detectable levels in the presenceof a relatively low substrate concentration, which in turn helps toavoid saturation of the reaction.

Conclusions

We have described the development of the first microplate fluo-rescence assay for analysis of in vitro KAT I activity. The majoradvantage of the new assay is that it is amenable to HTS due toits simplicity, quick readout, and low cost compared with existingHPLC methods. This assay can be easily miniaturized; we success-fully used it for screening libraries using a 3456-well format. A

188 Microplate fluorescence assay for KAT / J. Wong et al. / Anal. Biochem. 409 (2011) 183–188

shortcoming of the assay is that some compounds that fluoresce atsimilar wavelengths can be lost as false negatives. An advantage ofthe assay is that it could be readily modified for assaying other KATenzymes across a range of physiological pH values via optimizingpH of Zn2+-containing buffer and the amounts of enzyme, oxoacid,and L-KYN according to kinetic parameters of the KAT of interest.Also, we demonstrated the utility of this approach for establishingcellular models for compound prioritization.

Acknowledgments

The authors thank M. Kornienko, R. Smith, S. Sharma, J. ZugayMurphy, and J. Reid for generating recombinant hu KAT I, D. Colussifor contributing to the initial steps of the assay development, andIan Reynolds for providing helpful comments on the manuscript.

References

[1] R. Schwarcz, R. Pellicciari, Manipulation of brain kynurenines: glial targets,neuronal effects, and clinical opportunities, J. Pharmacol. Exp. Ther. 303 (2002)1–10.

[2] E. Vamos, A. Pardutz, P. Klivenyi, J. Toldi, L. Vecsei, The role of kynurenines indisorders of the central nervous system: possibilities for neuroprotection, J.Neurol. Sci. 283 (2009) 21–27.

[3] H. Baran, K. Jellinger, L. Deecke, Kynurenine metabolism in Alzheimer’sdisease, J. Neural Transm. 106 (1999) 165–181.

[4] S. Erhardt, K. Blennow, C. Nordin, E. Skogh, L.H. Lindstrom, G. Engberg,Kynurenic acid levels are elevated in the cerebrospinal fluid of patients withschizophrenia, Neurosci. Lett. 313 (2001) 96–98.

[5] S. Erhardt, S.K. Olsson, G. Engberg, Pharmacological manipulation of kynurenicacid: potential in the treatment of psychiatric disorders, CNS Drugs 23 (2009)91–101.

[6] H. Baran, N. Cairns, B. Lubec, G. Lubec, Increased kynurenic acid levels anddecreased brain kynurenine aminotransferase I in patients with Downsyndrome, Life Sci. 58 (1996) 1891–1899.

[7] Q. Han, T. Cai, D.A. Tagle, J. Li, Structure, expression, and function of kynurenineaminotransferases in human and rodent brains, Cell. Mol. Life Sci. 67 (2009)353–368.

[8] R. Pellicciari, R.C. Rizzo, G. Costantino, M. Marinozzi, L. Amori, P. Guidetti, H.Q.Wu, R. Schwarcz, Modulators of the kynurenine pathway of tryptophanmetabolism: synthesis and preliminary biological evaluation of (S)-4-(ethylsulfonyl)benzoylalanine, a potent and selective kynurenineaminotransferase II (KAT II) inhibitor, ChemMedChem 1 (2006) 528–531.

[9] L. Amori, P. Guidetti, R. Pellicciari, Y. Kajii, R. Schwarcz, On the relationshipbetween the two branches of the kynurenine pathway in the rat brain in vivo, J.Neurochem. 109 (2009) 316–325.

[10] F. Rossi, C. Valentina, S. Garavaglia, K.V. Sathyasaikumar, R. Schwarcz, S.Kojima, K. Okuwaki, S. Ono, Y. Kajii, M. Rizzi, Crystal structure-based selectivetargeting of the pyridoxal 50-phosphate dependent enzyme kynurenineaminotransferase II for cognitive enhancement, J. Med. Chem. 53 (2010)5684–5689.

[11] A.J. Cooper, The role of glutamine transaminase K (GTK) in sulfur and a-ketoacid metabolism in the brain, and in the possible bioactivation ofneurotoxicants, Neurochem. Int. 44 (2004) 557–577.

[12] A.J. Cooper, Purification of soluble and mitochondrial glutamine transaminaseK from rat kidney: use of a sensitive assay involving transamination between l-phenylalanine and a-keto-c-methiolbutyrate, Anal. Biochem. 89 (1978) 451–460.

[13] F. Rossi, Q. Han, J. Li, M. Rizzi, Crystal structure of human kynurenineaminotransferase I, J. Biol. Chem. 279 (2004) 50214–50220.

[14] J.H. Zhang, T.D. Chung, K.R. Oldenburg, A simple statistical parameter for use inevaluation and validation of high throughput screening assays, J. Biomol.Screen. 4 (1999) 67–73.

[15] K.J. Swartz, W.R. Matson, U. MacGarvey, E.A. Ryan, M.F. Beal, Measurement ofkynurenic acid in mammalian brain extracts and cerebrospinal fluid by high-performance liquid chromatography with fluorometric and coulometricelectrode array detection, Anal. Biochem. 185 (1990) 363–376.

[16] K. Shibata, Fluorimetric micro-determination of kynurenic acid, anendogenous blocker of neurotoxicity, by high-performance liquidchromatography, J. Chromatogr. 430 (1988) 376–380.

[17] Q. Han, J. Li, pH dependence, substrate specificity, and inhibition of humankynurenine aminotransferase I, Eur. J. Biochem. 271 (2004) 4804–4814.

[18] W. Schmidt, P. Guidetti, E. Okuno, R. Schwarcz, Characterization of humanbrain kynurenine aminotransferases using [3H]kynurenine as a substrate,Neuroscience 55 (1993) 177–184.

[19] Q. Han, T. Cai, D.A. Tagle, H. Robinson, J. Li, Substrate specificity and structureof human aminoadipate aminotransferase/kynurenine aminotransferase II,Biosci. Rep. 28 (2008) 205–215.

[20] E.E. Snell, The vitamin B6 group: V. The reversible interconversion of pyridoxaland pyridoxamine by transamination reactions, J. Am. Chem. Soc. 67 (1945)194–197.

[21] K. Wejksza, W. Rzeski, E. Okuno, M. Kandefer-Szerszen, J. Albrecht, W.A. Turski,Demonstration of kynurenine aminotransferases I and II and characterizationof kynurenic acid synthesis in oligodendrocyte cell line (OLN-93), Neurochem.Res. 30 (2005) 963–968.

[22] W. Rzeski, T. Kocki, A. Dybel, K. Wejksza, B. Zdzisinska, M. Kandefer-Szerszen,W.A. Turski, E. Okuno, J. Albrecht, Demonstration of kynurenineaminotransferases I and II and characterization of kynurenic acid synthesisin cultured cerebral cortical neurons, J. Neurosci. Res. 80 (2005) 677–682.