Proc. Natl. Acad. Sci. USAVol. 93, pp. 4251-4256, April 1996Chemistry

A general assay for antibody catalysis using acridone as afluorescent tagJEAN-LOUIS REYMOND*, THOMAS KOCH, JOSEF SCHROER, AND EMILY TIERNEY

Department of Molecular Biology, The Scripps Research Institute, 10666 North Torrey Pines Road, La Jolla, CA 92037

Communicated by Richard A. Lerner, The Scripps Research Institute, La Jolla, CA, December 4, 1995 (received for review October 10, 1995)

ABSTRACT A simple and highly sensitive catalysis assayis demonstrated based on analyzing reactions with acridone-tagged compounds by thin-layer chromatography. As little as1 pmol of product is readily visualized by its blue fluorescenceunder UV illumination and identified by its retention factor(Rf). Each assay requires only 10 p1 of solution. The methodis reliable, inexpensive, versatile, and immediately applicablein repetitive format for screening catalytic antibody libraries.Three examples are presented: (i) the epoxidation of acridonelabeled (S)-citronellol. The pair of stereoisomeric epoxidesformed is resolved on the plate, which provides a directselection method for enantioselective epoxidation catalysts.(ii) Oxidation of acridone-labeled 1-hexanol to 1-hexanal. Theactivity of horse liver alcohol dehydrogenase is detected. (iii)Indirect product labeling of released aldehyde groups byhydrazone formation with an acridone-labeled hydrazide.Activity of catalytic antibodies for hydrolysis of enol ethers isdetected.

An enormous variety of new catalytic activities are found byscreening catalytic antibody libraries generated by immuniza-tion with transition state analogs of chemical reactions (1).This approach requires an assay for catalysis applicable repet-itively and reliably on a very small scale. Several methods havebeen reported that use a substrate bound to a solid support andrely on selective tagging of the reaction product to generate asignal equivalent to that of the classical ELISA for bindingaffinity. This is accomplished by using a product-specificantibody (cat-ELISA) (2, 3), a biotin tag revealed by an avidinreagent (4), or a DNA-tag revealed by PCR (5).

In essence these methods measure product formation as atest for catalysis. If the amount of product formed is higherthan in a control sample consisting of the same medium underthe same conditions (solvent, pH, temperature), a strongersignal is obtained, indicating that a catalyst may be present. Toapply these methods for a single reaction, however, one mustdevelop a set of specific and often quite expensive reagents andgo through a lengthy optimization process. Here we show anequivalent yet much simpler and readily applicable catalysisassay based on analyzing reactions of substrates tagged withacridone by thin-layer chromatography (TLC). As little as 1pmol of product is readily visualized by its intense bluefluorescence under illumination at 254 nm with a standardlaboratory UV lamp. This corresponds to 0.1% conversion ofthe substrate present at 100 ,tM concentration. This sensitivityshould be sufficient to reveal a reasonable catalyst duringscreening of antibody or peptide libraries. In addition, sub-strate and product are separated by elution on the plate and arereliably identified by their retention factor (Rf). Direct sub-strate tagging and indirect product tagging using hydrazoneformation are demonstrated. Besides its bright fluorescence,acridone is photochemically and chemically stable, and itsderivatives can be handled without special protection from

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

light or air. The method is simple, inexpensive, extremelyversatile, and readily applicable. These qualities should makeit the method of choice for assays in catalysis.

MATERIALS AND METHODS

(3'S)10-[(3',7')-Dimethyloct-6'-en-l'-yl]-9(10H)-acridone(Compound 1). Two hundred and fifty milligrams of (6.25mmol) of 60% NaH in oil was added to a suspension of9(10H)-acridone (780 mg; 4.0 mmol) in 5.5 ml of dry dimeth-ylformamide to form a yellow fluorescent solution. (S)-Citronellyl bromide (0.80 ml; 4.0 mmol) was then added. Thereaction mixture was heated at 80°C until the yellow fluores-cence had disappeared (5 hr). Aqueous workup (water/ethylacetate) and purification by flash chromatography on silicagel(gradient, 10-15% diethyl ether in hexane) yielded olefin 1(700 mg; 52%).'H NMR (300 MHz, C2HCl3): 8.61 (dd, J = 8, 2 Hz, 2H);

7.75 (ddd, J = 9, 7, 2 Hz, 2H); 7.52 (br d, J = 9 Hz, 2H); 7.31(ddd, J = 8, 7, 1 Hz, 2H); 5.15 (t x quint, J = 7, 1.5 Hz, 1H);4.40 (m, 2H); 2.2-1.3 (m, 7H); 1.73 (s, 3H); 1.64 (s, 3H); 1.17(d, J = 7 Hz, 3H).

13C NMR (125 MHz, C2HC13): 177.9, 141.6, 133.8, 131.7,127.9, 124.1, 122.4, 121.1, 114.3, 44.6, 36.8, 33.6, 30.9, 25.7, 25.4,19.5, 17.7.HRMS: C23H27NO, (M+H+) calculated 334.2171, found

334.2160.(3'S,6'S) 10- [(3 ',7')-Dimethyl-(6',7')-epoxyoct-1 '-yl] -9-

(10H)-acridone (Compound 2a) and (3'S,6'R)10-[(3',7')-dimethyl-(6',7')-epoxyoct-1'-yl] -9(1OH)-acridone (Compound2b). Sodium bicarbonate (100 mg) and m-chloroperbenzoicacid (250 mg of 50-60% suspension in water) were added toa solution of olefin 1 (210 mg; 0.63 mmol) in dichloromethane(6 ml). After 2 hr at 20°C, the solution was washed withsaturated aqueous NaHCO3. The stereoisomeric epoxideswere separated in two successive flash chromatographies onsilicagel (100 g) by elution with a gradient of 10-15% ethylacetate in hexane, yielding isomer 2a (106 mg; 0.3 mmol; 48%)as a yellow solid and isomer 2b (84 mg; 0.24 mmol; 38%) as ayellow oil.

Isomer 2a: IH NMR (300 MHz, C2HC13): 8.60 (dd, J = 8,2 Hz, 2H); 7.75 (ddd, J = 9, 7, 2 Hz, 2H); 7.53 (br d, J = 9Hz, 2H); 7.32 (ddd, J = 8, 7, 1 Hz, 2H); 4.41 (m, 2H); 2.77 (m,1H); 2.1-1.5 (m, 6H); 1.36 (s, 3H); 1.31 (s, 3H); 1.21 (d, J =7 Hz, 3H).

13C NMR (125 MHz, C2HC13): 178.0, 141.7, 133.9, 128.0,122.5, 121.2, 114.4, 64.4, 58.3, 44.7, 33.8, 33.2, 31.0, 26.1, 24.9,19.8, 18.8.HRMS: C23H27NO2, (M+Cs+) calculated 482.1096, found

482.1084.Isomer 2b: 1H NMR (300 MHz, C2HC13): 8.60 (dd, J = 8,

2 Hz, 2H); 7.75 (ddd, J = 9, 7, 2 Hz, 2H); 7.50 (br d, J = 9 Hz,2H); 7.32 (br t, J = 8 Hz, 2H); 4.41 (m, 2H); 2.78 (dd, J = 7.5,5 Hz, 1H); 2.1-1.3 (m, 6H); 1.35 (s, 3H); 1.30 (s, 3H); 1.20 (d,J = 7 Hz, 3H).

*To whom reprint requests should be addressed.

4251

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Proc. Natl. Acad. Sci. USA 93 (1996)

13C NMR (125 MHz, C2HCl3): 178.0, 141.6, 133.9, 128.0,122.5, 121.2, 114.3, 64.5, 58.3, 44.5, 33.9, 33.8, 31.4, 26.6, 24.9,19.4, 18.7.HRMS: C23H27NO2, (M+Cs+) calculated 482.1096, found

482.1085.10-(6'-Hydroxy-hex-l'-yl)-9(10H)-acridone (Compound 3).

Alkylation of acridone (1.0 g; 5.1 mmol) as described aboveusing 6-bromohexyl-(tert-butyldimethylsilyl)ether (2.6 g; 8.8mmol) followed by treatment with 0.2 M HCI in 50% aqueousethanol overnight, aqueous workup, and chromatography(hexane/acetone, 2:1) gave alcohol 3 (1.05 g; 69%), which wasrecrystallized from ethyl acetate (mp 113-114°C).1HNMR (C2HC13, 300 MHz): 8.60 (dd,J = 8.0, 1.7 Hz, 2H),

7.74 (ddd, J = 8.7, 7.0, 1.7 Hz, 2H), 7.50 (br d, J = 8.7 Hz,2H), 7.30 (dt, J = 7.5, 0.6 Hz, 2H), 4.35 (t, J = 8.3 Hz, 2H),3.72 (t,J = 6.2 Hz, 2H), 2.01-1.91 (m, 2H), 1.70-1.50 (m, 7H).MS (Electrospray+): m/z (%): 318 (100) [M+Na]+, 296

(91) [M+H]+, 196 (35).10-(6'-Oxo-hex-l'-yl)-9(10H)-acridone (Compound 4). Re-

action of alcohol 3 (0.23 g; 0.80 mmol) with N-methylmor-pholine N-oxide (140 mg; 1.20 mmol) and tetrapropyl ammo-nium perruthenate (14 mg; 40 ,tmol) in acetonitrile/N,N-dimethylformamide (1:1) (2 ml) with 0.4 g of powderedmolecular sieves 4 A at 20°C for 1 hr, followed by purificationby chromatography (hexane/ethyl acetate, 2:1) gave aldehyde4 (0.140 g; 60%) (mp 116-117°C).1H NMR (C2HC13, 300 MHz): 9.80 (t, J = 1.4 Hz, 1H), 8.55

(dd,J = 8.0, 1.7 Hz, 2H), 7.70 (ddd,J = 8.7, 7.0, 1.7 Hz, 2H),7.44 (br d,J = 8.7 Hz, 2H), 7.27 (dt, J = 7.5, 0.6 Hz, 2H), 4.30(t, J = 8.3 Hz, 2H), 2.53 (dt, J = 7.1, 1.3 Hz, 2H), 2.00-1.87(m, 2H), 1.82-1.72 (m, 2H), 1.63-1.53 (m, 2H).MS (Electrospray+), m/z (%): 316 (100) [M+Na]+, 294

(95) [M+H]+, 196 (63).6-[10'-9'(10'H)-Acridonyl]hexanoic hydrazide (Compound

7). Alkylation of 9(10H)-acridone (200 mg; 1.0 mmol) in 4 mlof dimethylformamide with NaH (50 mg of 60% oil suspen-sion) and ethyl 6-bromohexanoate (450 mg; 2.0 mmol) asdescribed above and purification by chromatography (hexane/diethyl ether, 2:1) gave ethyl 6-[10'-9'(10'H)-acridonyl] hex-anoate (300 mg; 0.89 mmol; 89%) as a yellow oil. Reaction withhydrazine hydrate (1.5 ml) in ethanol (3 ml) at 20°C for 3 hr,followed by recrystallization from ethanol and chromatogra-phy on silicagel (1-5% methanol in dichloromethane), gavehydrazide 7 (240 mg; 0.74 mmol; 83%) as yellow crystals (mp169°C).1H NMR: (C2HCl3:C2H302H 15:1, 300 MHz): 8.51 (dd, J =

8.0, 1.7 Hz, 2H), 7.67 (ddd, = 8.7, 7.0, 1.7 Hz, 2H), 7.43 (d,J = 8.7 Hz, 2H), 7.35 (dt, J = 7.5, 0.6 Hz, 2H), 4.29 (t, J =

8.3 Hz, 2H), 2.15 (t,J = 7.2 Hz, 2H), 1.90 (quint,J = 8.1 Hz,2H), 1.75 (quint, J = 7.8 Hz, 2H), 1.52 (quint, J = 8.0 Hz, 2H).

13C NMR (125 MHz, C2HC13 + 15% C2H302H): 178.4,173.6, 141.5, 134.1, 127.6, 122.0, 121.3, 114.6, 45.8, 33.7, 26.8,26.2, 25.0.HRMS: C19H21N302, (M+H+) calculated 324.1712, found

324.1705.Hydrazone with Aldehyde 6 (Compound 8). Reaction of

hydrazide 7 (26 mg; 0.08 mmol) in 2 ml of methanol with 20mg of aldehyde 6 and 15 ,tl of acetic acid for 1 hr at 20°C,followed by chromatography on silica gel (10:1 dichlorometh-ane/methanol) gave 45 mg (0.08 mmol; 100%) of hydrazone 8as a yellow solid.1HNMR (300 MHz, 15:1 C2HCl3/C2H302H): 3:1 mixture of

isomers. Major isomer (minor isomer): 8.50 (dd, J = 9, 2 Hz,2H); 7.73 (7.70) (ddd, J = 9, 7, 2 Hz, 2H); 7.64 (br d, J = 8.5,2H); 7.52 (7.48) (br d, J = 9 Hz, 2H); 7.40 (br t, J = 6 Hz, 1H);7.27 (ddd, J = 9, 8, 1 Hz, 2H); 7.07 (7.16) (m, 2H); 4.33 (m,2H); 3.77 (m, 2H); 3.50 (m, 2H); 3.15 (br s, 2H); 2.82 (2.69) (m,1H); 2.60 (2.55) (m, 2H); 1.94, 1.75, 1.69 (3 m, 3 x 2H); 1.00(1.05) (d, J = 7 Hz, 3H).

HRMS: C32H36N404, (M+Cs+) calculated 673.1791, found673.1780.Hydrazone with Butanal (Compound 9). Reaction of hy-

drazide 7 (65 mg; 0.2 mmol) in 2 ml of methanol with 40 ,tl ofbutanal and 15 ,tl of acetic acid for 1 hr at 20°C, followed bychromatography on silica gel (ethyl acetate) gave 72 mg (0.19mmol, 95%) of hydrazone 9 as a yellow solid.1H NMR [300 MHz, C2HC13 + 3% (vol/vol) C2H3OD]:

mixture of 2 isomers in 4:1 ratio. Major isomer (minor isomer):8.53 (8.52) (dd, J = 8, 2 Hz, 2H); 7.73 (7.70) (ddd, J = 9, 7,2 Hz, 2H); 7.52 (7.49) (br d, J = 9 Hz, 2H); 7.29 (7.28) (br t,J = 8 Hz, 2H); 7.12 (7.38) (t, J = 5.5 Hz, 1H); 4.35 (m, 2H);2.68 (t, J = 7 Hz, 2H); 2.20 (2.28) (td, J = 7.5, 5.5 Hz, 2H);1.95 (m, 2H); 1.80 (m, 2H); 1.62 (m, 2H); 1.50 (1.52) (sext, J= 7.5 Hz, 2H); 0.90 (0.94) (t, J = 7.5 Hz, 3H).

13CNMR (125 MHz, C2HC13 + 10% C2H3OD): 178.2, 178.1,175.4, 169.5, 151.9, 148.1, 141.5, 141.4, 134.0, 127.5, 127.4,121.9, 121.8, 121.2, 114.6, 114.5, 45.9, 45.8,34.2,34.1,33.9,32.0,26.8, 26.6, 26.3, 26.0, 25.0, 24.1, 19.8, 19.5, 13.5, 13.4.HRMS: C23H27N302, (M+Cs+) calculated 510.1158, found

510.1144.Assay Setup. Substrates and reference products were used as

25 mM stock solutions in 2-propanol/water 2:1 (1), 2-propanol(2a and 2b), water/acetonitrile 1:1 (3-7). Reactions in volumesfrom 20 to 200 g,l were set up by mixing the appropriate buffer,reagents, and substrates in 0.5 ml-polypropylene Eppendorftubes or in individual wells of 96-well half-area tissue cultureplates. V-bottom-shaped plates were used for volumes below20 ,tl. After the necessary incubation time, the reactionmixtures were analyzed 8-12 at a time by transferring 10 tpl ofeach solution from the 96-well plate to the TLC plate using amultichannel pipetter.

Analysis of Direct Tagging. Standard nonfluorescent silicagel 60 glass plates (0.25 mm thick) cut to the appropriate size(10 x 10 cm or 6.7 x 13.3 cm) were used. A row of 10 or 12samples of 10 ,tl was applied with the multichannel pipetter ina line 1.5 cm above one of the edges. After drying for 10 minunder vacuum, the plate was eluted on 2.5 cm with a polarsolvent (4:1, dichloromethane/methanol) to affect preconcen-tration of the samples. The plate was dried again under vacuumfor 10 min and then eluted using the separating solvent. Theplate was analyzed in a dark room under illumination with254-nm light from a portable lamp for TLC.

Analysis of Indirect Tagging with Hydrazide 7. Hydrazide 7was first added as 5 ,tl of a 550 AM solution in 1.0 M Mes bufferat pH 5.5 to each 50-,tl reaction sample and the resultingsolutions were incubated for 3 hr at 20°C. Samples (10 Aul) werethen transferred in row of 10 to a 10 x 10 cm HP-TLC plate(silica gel 60, 0.2 mm thick) with a built-in preconcentrationzone. Loading on the HP-TLC plate required extreme care dueto the mechanical fragility of the preconcentration zone butwas necessary because hydrazide 7 and its derivatives did notelute if applied directly on the silica gel surface as aqueoussolution. The same procedure as described above was thenfollowed except that a preconcentration run was left out.HP-TLC plates and standard TLC plates performed equally interms of sensitivity and resolution.

Control Experiments with Dansylamide and Fluorescein.Stock solutions (10 mM) of either dimethylaminonaphthalenesulfonamide (dansylamide) or fluorescein in 1:1 acetonitrile/water were diluted in down to 40, 20, 10, 5, 2.5, 1, 0.5, 0.25, and0.125 ,tM and 10 tul of each was analyzed on an HP-TLC plate.Dansylamide eluted with 1.5% methanol in dichloromethane(Rf = 0.30) and fluorescein with 5% methanol in dichlo-romethane (Rf = 0.25). Both compounds gave yellow fluores-cent bands under UV illumination at either 254 nm or 366 nmvisible down to 10 ,tM with dansylamide and 0.5 ,tM withfluorescein.

4252 Chmsr:Ryodeal

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Proc. Natl. Acad. Sci. USA 93 (1996) 4253

RESULTSDirect Substrate Tagging: Stereoselective Epoxidation. We

demonstrate the principle of direct tagging with acridone at theexample of the epoxidation of olefin 1, derived from (S)-citronellol, to stereoisomeric epoxides 2a and 2b (Fig. 1). Anegative picture of the TLC plate taken under illuminationwith a portable UV 254-nm lamp is shown in Fig. 2. Epoxides2a (lane 4) and 2b (lane 5) are readily distinguished from eachother and from olefin 1 (lane 3) upon elution with 85:15hexane/ethyl acetate. The detection limit is reached near 1pmol of epoxide (lanes 6-10). Using this method, we havefollowed the epoxidation reaction in aqueous buffer (50 mMphosphate/50 mM sulfate, pH 6.5 or 7.5) with 30% (vol/vol)2-propanol cosolvent, using either acetonitrile/H202 (lane 1)(6) or formamide/H202 (lane 2) (7).

Oxidation of Primary Alcohols by a Dehydrogenase. Directtagging is easily adapted for a simple test for the dehydroge-nase activity of enzymes (Fig. 3). As shown on the negativepicture in Fig. 4, alcohol 3 (lane 4) is easily separated from itsoxidation product aldehyde 4 (lane 5) by elution with 2:3 ethylacetate/hexane. The detection limit is reached near 1 pmol(lanes 6-10). The oxidation of alcohol 3 to aldehyde 4 is readilyobserved in the presence of horse liver alcohol dehydrogenase(HL-ADH), with added nicotinamide dinucleotide (NAD+),and flavin mononucleotide (FMN) as oxidant (lane 1) (8). Noreaction takes place in the absence ofNAD+ (lane 2), or in theabsence of enzyme (lane 3).

Indirect Tagging via Hydrazone Formation. The principle ofindirect tagging is demonstrated for the acid-catalyzed hydro-lysis of enol ether 5 to produce aldehyde 6 (Fig. 5). Thesensitivity limit for detection of these compounds on fluores-cent TLC plates by quenching with the benzamide nucleus isonly 1 mM. In this case, however, the reaction releases analdehyde group that can be chemoselectively tagged by usingan acridone/hydrazide reagent. Thus, reaction of aldehyde 6(100 ,uM) with hydrazide 7 at pH 5.5 for 3 hr results in "70%conversion of the aldehyde to the fluorescent hydrazone 8.Elution on a TLC plate (CH2Cl2/MeOH, 20:1) allows us toseparate hydrazone 8 (lane 2) from hydrazide 7 (lane 1). Theresults are shown in Fig. 6. The upper bands visible in each laneresult from direct decomposition of the hydrazide reagent onthe plate and could not be removed by repeated purification ofthe hydrazide. The sensitivity limit is reached at 2 AM con-centration of aldehyde 6 (lanes 3-5). Using this method, we canidentify the activity of several catalytic antibodies for thereaction (lanes 6-8) (9-12). Test concentrations of othersimple aldehydes at micromolar levels are also easily detect-ed-for example, butanal (lanes 9 and 10).

DISCUSSIONAssay Sensitivity. TLC is an extremely versatile analytical

technique and has been developed to identify every possibleclass of organic and inorganic compounds (13). It is used inevery synthetic laboratory to monitor reactions. Both quali-tative and quantitative analyses, including kinetic analyses, are

Lane 1 2 3 4 5 6 7 8 9 10

FIG. 2. Epoxidation of (S)-citronellol derivative 1 to stereoisomers2a or 2b analyzed by TLC. Standard 10 x 10 cm TLC plate (silica gel60; 0.25 mm thick). Elution: 4:lCH2Cl2/MeOH (2.5 cm, preconcen-tration), then 85:15 hexane/ethyl acetate (twice). Each lane is theanalysis of 10 tal of aqueous 30 mM phosphate/30 mM sulfate bufferwith 30% 2-propanol and the following: epoxidation of 100 ,uM olefin1 for 4 hr at 20°C using 700 mM H202 with 250mM acetonitrile at pH7.5 (lane 1) and with 250 mM formamide at pH 6.5 (lane 2); olefin 1in 10 ,AM concentration (lane 3), epoxide 2a in concentrations of 10tjM (lane 4), 2 ,uM (lane 6), 0.5 ,uM (lane 8), and 0.125 ttM (lane 10);epoxide 2b in concentrations of 10 ,iM (lane 5), 1.0 ,iM (lane 7), and0.5 tLM (lane 9).

possible, so that the technique is in principle adequate forscreening of catalysis. However, reported detection limits arein the range of 0.03 to 1.7 nmol (0.1-5 jig for an averagemolecular weight of 300) depending on the compound classand visualization method (fluorescence or chemical staining)(13). Fluorescence, for instance, has been used for TLCanalysis of amines and ketones using tagging with the dansyl(dimethylamino naphthalene sulfonyl) fluorophore. Dansylchloride derivatization allows for analysis of amino acids andamines by TLC (14), and hydrazone formation with dansylhydrazine can be used to identify keto steroids on TLC (15, 16).Dansyl has also been used as a tag to screen the catalyticactivity of antibodies (17). The dansyl group is a weak fluoro-phore, and its derivatives are photosensitive and are rapidlybleached when adsorbed on the TLC plate. In our hands, thesensitivity for the dansyl group lies at 100 pmol if the plate isanalyzed within 2-3 min, which is within the typical range ofTLC detection methods (Fig. 7).The acridone tag introduced here provides an exceptionally

sensitive fluorescent probe (18-20). At 1 pmol, it is 100 timesmore sensitive than the dansyl group. Indeed, it is even -5 timesmore sensitive than fluorescein, which gives a yellow fluorescencedetectable only down to -5 pmol in our TLC assay.

Applicability for Catalysis Screening. A detection limit of 1pmol should be sufficient to reveal any catalyst within acombinatorial screening protocol. A simple analysis can bedone by counting turnovers per catalytic species. To detect a

0 H202/HCONH2 0 0

H20/2-propanol 7:3 i + riIpH 6.5 -HCCH3 H HCOAH H3CCH3

H3C CH1H3C Ha H3CHsC^^^̂ CHaHC'1"^^ sC1

1 2a 2b

FIG. 1. Epoxidation of (S)-citronellyl acridone derivative 1 to stereoisomeric epoxides 2a and 2b (the relative stereochemistry of epoxides 2aand 2b is unknown and has been attributed arbitrarily to facilitate the discussion).

Chemistry: Reymond et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Proc. Natl. Acad. Sci. USA 93 (1996)

° HL-ADH, NAD+, FMN20 mM borate pH 8.8

-XCHCH2OH

3

O C

--^^*CHO

OCH3 1) cat. Ab, pH 5.5,

NH N3 37 °C, 15h

HO-N CH3____ H

0

5

4

FIG. 3. Oxidation of hexanol acridone derivative 3 to aldehyde 4 byhorse liver alcohol dehydrogenase.

single catalytic turnover, 1 pmol of catalyst should be presentin the 10-,tl sample transferred to the plate. In the case ofcatalytic antibodies, the assay is done on the hybridoma cellculture supernatant, where the antibody concentration is '15,tg/ml, which corresponds to 100 nM or 1 pmol in 10 ,tl. Thus,a single catalytic turnover would be sufficient to produce adetectable amount of product. For a catalyst of a molecularweight of 5000 (for example a 50-mer peptide), 10 gul of a 1mg/ml solution contains 2 nmol. At that concentration, amixture of 2000 different candidates could be similarly ana-

lyzed for carrying out a single turnover.A more detailed analysis can be done, including preequi-

librium substrate binding, in the form of the Michaelis-Mentenconstant Km [M], a first-order catalytic rate constant kcat [s-1]for the catalyst-substrate complex, and a competing first-orderbackground reaction for the substrate alone with rate constantkuncat [s-1]. On the basis of reported substrate binding con-stants for enzymes and catalytic antibodies, one can assumethat at substrate concentration [S] = 100 tLM, [S] < Km formost catalysts. Under these conditions, the catalytic rate isgiven by Vcat = [S] x [cat] x (kcat/Km), where [cat] is theconcentration of catalyst in the sample. The reaction rate in areference sample is given by Vuncat = [S] X kuncat, and the ratein the test sample containing catalyst by Vobs = Vcat + Vuncat.An apparent doubling of the reaction rate, which can bedetected on the TLC plate by visual inspection, requires thatVObs > 2 Vuncat; hence, Vcat > Vuncat. Using the transitionstate-catalyst dissociation constant KTs = Km/(kcat/kuncat) asa description for catalysis (21-23), Vcat > Vuncat requires that[cat] > KTS. Since it is also required that Km > [S], the

Lane 1 2 3 4 5 6 7 8 9 10

2) 50 iM 7, pH 5.

0

Oo

HH2NN

0

7

0

H

OH1O0-__ CH

0

6

0

.5, 20 °C, 3h H

H [N'N NNN

HO- N CH3 0

0 8

0

butanal, pH 5.5, 20 °C, 3h ISNi

HN'

0

9

FIG. 5. Formation of fluorescent hydrazones detects aldehyde(S)-6 formed by hydrolysis of enol ether 5 with catalytic antibodies as

hydrazone 8 and butanal as hydrazone 9.

combined conditions yield Km/KTs = kcat/kuncat > [S]/[cat].For example, for [S] = 100 ,iM, a catalytic antibody present at[cat] = 10-7 M concentration in the test sample should havea rate enhancement kcat/kuncat > 103 to give rise to detectablecatalysis, a number well in the range of the reported efficien-cies of catalytic antibodies and probably within reach ofcatalytic peptides (24). It is also evident from the detectionlimit kcat/kuncat > [S]/[cat] that the threshold for detection canbe lowered by further reducing the substrate concentration.

Convenience of the Acridone Tag. Acridone offers highsensitivity for fluorescence detection yet, in contrast to otherfluorescent probes like fluorescein, possesses a remarkablysimple chemical structure. Acridone is a vinilogous aromatic

Lane 1 2 3 4 5 6 7 8 9 10

9

'~~..'t.; . .: ·'....'?.. -:. ''- ' .. .... "fg.'~.; ,,

_ .:4,~,,. " :" .'=~i..........".?"........:? '',?....;': ":: ,...='".~=:.' ,~ ~

·~t.i. - - ,, j _ ..4......., ..~~ O p

~8~~...,._::-_ :_... . ...

FIG. 4. Detection of alcohol dehydrogenase activity by TLC usingthe oxidation of alcohol 3 to aldehyde 4. HP-TLC plate 10 x 10 cm(silica gel 60; 0.2 mm thickness, with preconcentration zone). Elution:2:3 ethyl acetate/hexane (once, 10 cm). Each lane is the analysis of 10pil of aqueous 20mM borate buffer at pH 8.8 containing the following:reaction mixture of 100 ptM alcohol 3 in the presence of horse liverdehydrogenase (20 tug/ml), 300 pIM flavin mononucleotide, and 20tAM nicotinamide dinucleotide (NAD+) (lane 1); NAD+ omitted (lane2); enzyme omitted (lane 3); alcohol 3 in concentrations of 10 ptM(lane 4), 2 pM (lane 6), 0.5 puM (lane 8), and 0.125 p.M (lane 10);aldehyde 4 in concentrations of 10 puM (lane 5), 1 ,pM (lane 7), and 0.25taM (lane 9).

.:..:'..¥.'.ii*, i'.;;i. :-, .. ¢!.... a..:....ii

FIG. 6. Indirect detection of micromolar concentration of alde-hydes by hydrazone formation. All samples were treated for 3 hr at20°C, pH 5.5, with hydrazide 7 at 50 ,tM concentration before analysis.HP-TLC plate 10 x 10 cm (silica gel 60; 0.2 mm thick, with precon-centration zone). Elution: 5% methanol in dichloromethane. Eachlane is the analysis of 10 tal containing blank reference (lane 1);aldehyde 6 at a concentration of 100 ,tM (lane 2), 10 piM (lane 3), 5ptM (lane 4), and 2 tuM (lane 5); incubation of 250 ptM enol ether 5at 37°C, pH 5.5, 15 hr, with catalytic antibodies 14D9 (lane 6), 22H25(lane 7), and 19C9 (lane 8) at 1 mg/ml. No aldehyde was detectedunder these conditions without catalytic antibody (not shown); butanalis readily detected in concentrations of 100 pAM (lane 9) and 10 tpM(lane 10).

4254 Chmsr:Ryodeal

r,

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Proc. Natl. Acad. Sci. USA 93 (1996) 4255

SO2NH2

N(CH3)2

dansyl amide

(> 100 picomoles)

COOH

HO 0'

fluorescein(> 5 picomoles)

0

H

acridone(> 1 picomole)

FIG. 7. Fluorescent probes and their detection limit by visualanalysis on TLC under illumination at 254 nm.

amide and as such is quite inert chemically. For example, wehave subjected acridone to heat and treatment with concen-trated sulfuric acid or hydrochloric acid, hydrogen peroxide,peracids, sodium hydride, etc., without noticing any decom-position. Acridone should be compatible with a wide variety ofreagents and thus applicable for quite diverse reactions. Be-cause of its lack of reactivity, the tag does not requireprotection and can be incorporated by alkylation on a terminalprimary halide almost at any stage of a substrate synthesis.Furthermore, acridone turns out to be photochemically quitestable, and its derivatives can be handled at light withoutprotection, which makes them convenient for use in synthesis.Most importantly, no bleaching occurs on the TLC plate for atleast 1 week after adsorption, allowing plates to be stored andreanalyzed if needed for comparison purposes.Value of the TLC Analysis. TLC is not only sensitive but also

provides each assay with valuable information on the productformed by the (Rf). Analysis readily shows the overall com-

position of reaction products, which allows us to estimate howclean the chemical reaction is and provides a reliable safeguardagainst artifacts. The Rf information can also provide struc-tural information otherwise difficult to obtain so early in a

screening protocol. This is illustrated by the opportunity todistinguish both stereoisomeric epoxides 2a and 2b derivedfrom olefin 1. The reporter chiral center in 1 is distant enoughfrom the reactive double bond that it should not influence thestereoselectivity of a potential catalyst. Therefore, the stereo-isomeric ratio 2a/2b read from the TLC plate provides a

powerful screening device for enantioselectivity in the epoxi-dation reaction.

Compatibility of the Direct Tagging Procedure with Bioca-talysis. The compatibility of the direct acridone labeling, whichis most sensitive, with biocatalysis, is shown here by oxidationof alcohol 3 to aldehyde 4 by horse liver alcohol dehydroge-nase. It should be noted that solubility in aqueous buffers is nota problem with acridone derivatives at 100 ,uM concentration.Aldehyde 4, epoxides 2a and 2b, and hydrazide 7 are com-

pletely soluble at that concentration. Alcohol 3 requires 2%(vol/vol) dimethylformamide cosolvent for solubility in bufferdue to its high crystallinity. Only olefin 1 required 30%(vol/vol) 2-propanol for solubility. In this case, solubility maybe increased by introducing a water-soluble linker between thetag and the olefin-for example, a polyethylene glycol spacer.Since olefin 1 is perfectly soluble in hexane, biphasic alkane/water conditions could provide an alternative for aqueouscatalysis (25-27). A more serious problem with acridone-tagged substrates could be selection of a catalytic speciesbinding to the substrate at the acridone tag, and this was indeedobserved in the case of a catalytic antibody substrate taggedwith a dansyl group (17). Interaction with the tag is likely withbiocatalysts in an aqueous environment because hydrophobicinteractions dominate substrate binding. Binding to the tagneed not be necessary, however, and this risk can probably beminimized by using a long, hydrophilic spacer arm (polyeth-ylene glycol) and should be completely eliminated by usingpairs of substrates with spacers of different length.

Indirect Tagging Procedure. The indirect tagging procedureeliminates difficulties associated with direct tagging and mighthave broader applicability in biocatalysis. Here we have dem-onstrated chemoselective tagging of an aldehyde group un-covered by hydrolysis of enol ether 5. At a 20-pmol detectionlimit, this analysis is significantly less sensitive than the directtag method. Nevertheless, the activity of catalytic antibodies isreadily visualized. The key advantage of indirect tagging is theability to analyze reactions of substrates lacking the acridonenucleus or any other aromatic chromophore.

Applicability of the Acridone Tag to High ThroughputScreening for Catalysis. High throughput screening requiresan analytical format applicable repetitively on a small scale.We have used 96-well tissue culture plates as reaction vessels,each containing 50 tul of solution. Transfer of tagging reagentsas well as transfer of the 10-,ul probe to the TLC plate is easilycarried out repetitively with multichannel pipetters (eight ortwelve). Assay of a complete 96-well plate requires 8 transferoperations to 6.7 x 13.3 cm plates or 10 transfer operations to10 x 10 cm plates. This protocol could be automated. Thenumber of operations per plate is similar to the cat-ELISAprocedure, where the application of several successive washingsteps and secondary reagents is necessary. Although HP-TLCplates with a preconcentration zone are more convenient sincethe preconcentration run is not necessary, standard glass TLCplates provide equal sensitivity together with better mechan-ical strength, and can also be cut to the 12 sample size. Thesemight therefore be preferable for use in large scale screening.

Reliability and versatility are decisive when several differentreactions must be analyzed. There the TLC format possesseskey advantages over ELISA based assays. First, one is allowedto carry out direct solution chemistry as opposed to a surfacedisplay, which allows us to assay the reaction under its realconditions. Second, product identification is very direct, whichavoids artifacts and allows us to study reactions with multipleproducts. Third, and probably most importantly, the timerequired for setting up a system is minimal because the TLCconditions are usually known from synthesizing referenceproduct and substrate.

CONCLUSIONDirect or indirect covalent tagging with acridone in combina-tion with TLC provides a general assay for catalysis. With adetection limit of 1 pmol, the assay is sensitive enough forapplication in combinatorial screening of antibodies or pep-tides. Multiple sample analysis is carried out with standardlaboratory equipment and requires only a single transferoperation per sample. The technique is simple, reliable, ver-satile, and inexpensive. Future improvements might includeautomation as well as lowering of the detection limit, this byreducing the thickness of the TLC plates or by using a betterlight source in combination with appropriate filters to enhancethe signal/noise ratio of the fluorescence.

We thank Prof. Richard A. Lerner for his help and suggestions andthe Art and Graphics laboratory at Scripps for photography. This workwas supported in part by the Swiss National Science Foundation(T.K.), the Humboldt Stiftung, Germany (J.S.), and the NationalInstitutes of Health (GM 49736 to J.-L.R.).

1. Schultz, P. G. & Lerner, R. A. (1995) Science 269, 1835-1842.2. Tawfik, D. S., Green, B. S., Chap, R., Sela, M. & Eshhar, Z.

(1993) Proc. Natl. Acad. Sci. USA 90, 373-377.3. MacBeath, G. & Hilvert, D. (1994) J. Am. Chem. Soc. 116,

6101-6106.4. Lane, J. W., Hong, X. & Schwabacher, A. W. (1993)J. Am. Chem.

Soc. 115, 2078-2080.5. Fenniri, H., Janda, K. D. & Lerner, R. A. (1995) Proc. Natl. Acad.

Sci. USA 92, 2278-2282.

Chemistry: Reymond et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

4256 Chemistry: Reymond et al.

6. Payne, G. B., Deming, P. H. & Williams, P. H. (1961) J. Org.Chem. 26, 651-663.

7. Chen, Y. & Reymond, J.-L. (1995) Tetrahedron Lett. 36,4015-4018.8. Lok, K. P., Jakovac, I. J. & Jones, J. B. (1985) J. Am. Chem. Soc.

107, 2521-2526.9. Reymond, J.-L., Janda, K. D. & Lerner, R. A. (1992) J. Am.

Chem. Soc. 114, 2257-2258.10. Reymond, J.-L., Jahangiri, G. K., Stoudt, C. & Lerner, R. A.

(1993) J. Am. Chem. Soc. 115, 3909-3917.11. Reymond, J.-L., Reber, J.-L. & Lerner, R. A. (1994) Angew.

Chem. Int. Ed. Engl. 33, 475-477.12. Jahangiri, G. K. & Reymond, J.-L. (1994) J. Am. Chem. Soc. 116,

11264-11274.13. Stahl, E., ed. (1969) Thin-Layer Chromatography (Springer, Berlin).14. Seiler, N. & Wiechmann, M. (1967) J. Chromatogr. 28, 351-362.15. Chayen, R., Dvir, R., Gould, S. & Harell, A. (1971) Anal.

Biochem. 42, 283-286.16. Anderson, J. M. (1986) Anal. Biochem. 152, 146-153.

Proc. Natl. Acad. Sci. USA 93 (1996)

17. Martin, M. T., Angeles, T. S., Sugasawara, R., Aman, N. I.,Napper, A. D., Darsely, M. J., Sanchez, R. I., Booth, P. & Titmas,R. C. (1994) J. Am. Chem. Soc. 116, 6508-6512.

18. Legg, K. D. & Hercules, D. M. (1969) J. Am. Chem. Soc. 91,1902-1907.

19. Rapaport, E., Cass, M. W. & White, E. H. (1972) J. Am. Chem.Soc. 94, 3160-3167.

20. Kawata, H., Shimada, K., Kumagai, T. & Niizuma, S. (1993)J. Chem. Soc. Perkin Trans. 2, 1395-1398.

21. Kurz, J. L. (1963) J. Am. Chem. Soc. 85, 987-991.22. Kurz, J. L. (1972) Acc. Chem. Res. 5, 1-9.23. Kraut, J. (1988) Science 242, 533-540.24. Stewart, J. D. & Benkovic, S. J. (1995) Nature (London) 375,

388-391.25. Chen, C.-S. & Sih, C. J. (1989) Angew. Chem. Int. Ed. Engl. 28,

695-707.26. Klibanov, A. M. (1990) Acc. Chem. Res. 23, 114-120.27. Ashley, J. A. & Janda, K. D. (1992) J. Org. Chem. 57, 6691-6693.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020