Biochemml Pharmocolog,v, Vol.40. No. 12. pp. 2587-2596. 1990. lloO6-2952/90$3.00+ 0.00 PrintedinGreat Britain. Q 1990Pergamon Press plc

CHARACTERIZATION OF NEUROPATHY TARGET ESTERASE USING TRIFLUOROMETHYL KETONES

THOMAS C.THOMAS,* ANDRASSZ~KACS,~SSCOTTKOJAS,* BRUCED.HAMMOCK,~~ BARRY W. WILSON/]and MARK G.McNAMEE*~ Departments of *Biochemistry and Biophysics, //Avian Sciences, fEntomology, and liEnvironmenta1 Toxicology, University of California, Davis, CA 95616, U.S.A.

(Receioed 8 Janunry 1990; accepted 8 June 1990)

Abstract-Neuropathy target esterase (NTE) is a membrane-bound carboxylesterase activity which is proposed as the target site in nerve tissue for initiation of organophosphate-induced delayed neuropathy. This activity is identified as phenyl valerate hydrolysis which is resistant to treatment with paraoxon and sensitive to co-incubation with paraoxon and mipafox. NTE preparations were obtained, which did not contain paraoxon-sensitive or mipafox-resistant hydrolases, by selective reconstitution of detergent-solubilized NTE from chick embryo brain into asolectin vesicles during gel filtration. The topography of the catalytic site of NTE was then examined by investigating the inhibition of NTE by a series of 3-alkylthio- and 3-arylthio-l,l,l-trifluoro-propan-2-ones. These trifluoromethyl ketones were found to be rapidly reversible, competitive inhibitors of NTE with I,, values from 1.3 X 1Om4 M to 4.9 x lo-” M. Correlation of Is<,values with octanol/water partition coefficients (Pi. in the range of log P = 1.5 to 5.9, indicated that the optimal lipophilicity for NTE substrates and inhibitors is in the range of log P = 3.0 to 3.4. Electrophilic substitution at the meta position of aromatic rings increased the inhibitory capacity of these inhibitors, whereas substitution at the orrho position reduced inhibitory capacity. These results indicate both that a large hydrophobic pocket is closely associated with the catalytic residue of NTE, and that affinity for the active site is affected by steric and electronic parameters.

Neuropathy target esterase (NTE,** also known as inhibitors has been examined in two previous studies neurotoxic &erase) is an integral membrane protein [ll, 12]. In excess of 150 different phosphates, found predominantly in central and peripheral nerve phosphonates, phosphinates, carbamates, and sul- tissue [l-3]. NTE is proposed as the target site at fates were examined. Analysis of the results involved which certain organophosphates (OPs) such as the examination of a homologous series of diisopropyl phosphoro~uoridate (DFP) and di- compounds, but attempts were not made to isopropylphosphorodiamidic Auoride (mipafox) act quantitatively relate observed trends to the physico- to initiate organophosphate-induced delayed neur- chemical parameters of the compounds. Although a opathy (OPIDN) [4-71. The symptoms of this strong correlation between structure and activity was potentially irreversible neuropathy appear 2-3 weeks not identified, the results indicated that both after a single acute exposure to OPs and are hydrophobic and steric parameters were important. accompanied by a distal degeneration of the long, It has been reported previously that trifluoromethyl large diameter peripheral nerves and the distal ends ketones are potent inhibitors of a large number of of the ascending and descending tracts of the spinal serine active site hydrolases 113-151, including human cord [&lo]. Determination of the role of NTE in brain NTE (Talcott, Ketterman and Lotti, personal the proper function of the nerve axon will require communication to B.D.H.). The substitution of the both the purification of NTE and the characterization hydrogen atoms LYto the carbonyl group with of its physical and catalytic properties. fluorines enhances the reactivity of the carbonyl with The relationship between the structures and the serine in the catalytic site [ 131. and it is postulated inhibitory potencies of a wide range of NTE that these compounds behave as transition state analogs [16]. In the present study we investigated the interaction of NTE with a series of twenty-two $ On leave from the Plant Protection Institute of the aliphatic and aromatic trifluoromethyl ketone Hungarian Academy of Sciences, H-1525 Budapest POB sulfides. The Iso values for inhibition of NTE by 102, Hungary. each of these compounds were determined, and an 1 Author to whom all correspondence should be addres- analysis of quantitative structure-activity relation- sed. ships (QSAR) was performed based on the ** Abbreviations: NTE, neuropathy target esterase; physicochemical properties of each inhibitor. The OPIDN, organophosphate-induced delayed neuropathy; purpose of this analysis was 3-fold: (1) to characterize OTFP, 3-octyIthio-l,l,l-t~fluoropropan-2-one; TFK; tri- fluoromethyl ketone; mipafox, N,N’-diisopropylphospho- the active site of NTE, (2) to provide information rodiamidic fluoride; DFP, diisopropyl phosphorofluori- for the synthesis of more specific substrates and date; paraoxon, phosphoric acid diethyl 4nitrophenyl inhibitors, and (3) to identify appropriate ligands ester; P, octanol/water partition coefficient; and EGTA, for affinity chromatography of NTE. ethyleneglycoibis(aminoethylether)tetra-acetate. In addition, we report a method for obtaining an 2587 2588 ‘I-. c. ‘rHOMAS et al

NTE preparation which is free from contamination the following protocol. Samples (lC-50 pL) were by non-NT&type phenyl valerate hydrolases. incubated for 20 min at 37” in an appropriate volume Operationally, NTE is identified as that portion of of 50 mM Tris-HCl, 0.2 mM EDTA, pH 8.0, and an total phenyi valerate hydrolyzing activity which is inhibitor solution such that the finai volume was resistant to pretreatment with non-neuropathic 0.74mL. The inhibitor solution was one of the paraoxon (40 kcM), but sensitive to pretreatment following: (a) 5OpL of above buffer; (b) 5OpL of with paraoxon and neuropathic mipafox (50(*M) 0.6 mM paraoxon in 50 mM Tris-citrate, pH 6.0; (c) [17]. Therefore, to determine NTE activity, pairs of 50 PL of 0.75 mM mipafox in 50 mM Tris-citrate, differentially inhibited samples must be assayed. pH 6.0; or (d) 50 PL each of the above paraoxon Protocols to characterize the interaction of NTE and mipafox solutions. Ten microliters of phenyl with other inhibitory compounds must confront two valerate (12.5 mg PV/mL in dimethylformamide) potential problems. First, the catalytic sites of NTE was added and tubes were incubated for 30min at may be less than 50% saturated with substrate during 37”. Enzymatic hydrolysis was stopped by adding assay [18, 191, in which case, nonbound paraoxon 0.5 mL of 1% (w/v) SDS, 0.25% (w/v) 4- and mipafox may compete with the substrate or the aminoantipyrine, 50 mM Tris-HCl, 0.2 mM EDTA, inhibitor of interest for interaction with hydrolytic pH 8.0. Then the colored complex was developed sites. Second, if nonbound paraoxon and mipafox by adding 0.25 mL of 0.4% (w/v) K3Fe(CN)h. The are removed prior to addition of test compounds absorbance of each assay tube was determined in and substrate, then reversibly inhibited hydrolases triplicate using a Bio-Tek 96-well plate reader may be reactivated. Evidence identifying the (Winooski, VT) by addition of 3OOpL from each existence of such a hydrolase in chick embryo brain tube to each of three wells of a Falcon 96-well plate membrane fractions is presented. The problems (Becton-Dickinson, Oxnard, CA). Absorbance was associated with these methods were avoided in the determined at 490 nm and the concentration of present study by screening inhibitors against an phenol was calculated using an absorption coefficient NTE-type phenyl valerate hydrolase preparation equal to l6,812M-‘cm-‘. Units of activity are from which all paraoxon-sensitive and mipafox- reported as micromoles of phenol produced per resistant activities had been removed. minute (I.U.). Protein determinations. Protein was quantitated using the method of Lowry et al. [25] with bovine MATERIALS AND METHODS albumin (Fraction V, Sigma, St. Louis, MO) as a standard. To avoid the interference caused by C~e~~c~~.~. Ahphatic and aromatic 3-substituted precipitation of Triton X-100, 0.1 mL of 24% (w/v) thio-l,l,l-tri~uoropropanones (TFKs) used in this sodium dodecyl sulfate (SDS) was added to samples study were synthesized previously by the reaction containing Triton X-100 (final assay vol. 1.3 mL) of the appropriate thiol with 3-bromo-l,l,l- WI. trifluoropropan-2-one, purified, and analyzed Preparation of microsomal membranes from [l&20]. Mipafox (N,N’-diisopropylphosphoro- chicken embryo brains. Brains from day 19 chicken diamidic fluoride) and phenyl valerate were embryos were homogenized for 25 set in 10 mL of synthesized as previously described [21] according ice-cold Buffer A (50 mM Tris-HCl, 0.5 M NaCl, to the methods of Johnson [17]. The purity of 2 mM EDTA, 2 mM EGTA, pH 7.2, at 21”) per g mipafox was monitored by determination of its of tissue with a Polytron homogenizer on setting 7 melting point (60.5”, white crystals), and the (Brinkmann Instruments, Westbury, NY). Homo- structures of mipafox and phenyl valerate were genates were centrifuged in an SS34 rotor (Sorvall, analyzed by NMR and i.r. Paraoxon (phosphoric Wilmington, DE) for 10 min at 4” and 1100~ (rav acid diethyl 4-nitrophenyl ester, Aldrich Chemical 8.26 cm). The upper layer of foam was aspirated and Co., Milwaukee, WI) was analyzed for interfering the supernatant was isolated. The low speed contaminants by thin-layer chromatography and gas supernatant was centrifuged in a Type 60 Ti rotor phase chromatography, and by determination of its (Beckman, Palo Alto, CA) for 20 min at 4” and Isa to NTE (600pM) as recommended by Johnson 100,OOOg (rsv 6.15 cm) to obtain a crude microsomal [22]. Stock solutions of mipafox (10-l M) in 50 mM membrane fraction. Membrane pellets were resus- Tris-citrate, pH 6.0 and paraoxon (10-l M in acetone) pended in Buffer B (50mM Tris-HCl, 0.2mM were prepared and stored in a desiccator at -25”. EDTA, pH 8.0, at 21’) and stored in liquid nitrogen Triton X-100 (Surfa~t-Amps X-100) was obtained for future use. from Pierce (Rockford, IL). Asolectin was obtained Inhibition qf membrarzes with paraoxo~. Mem- from Associated Concentrates (Woodside, NY). brane suspensions were treated with 100pM Animai- Fertilized White Leghorn chicken eggs paraoxon for 20min at 37”. Inhibition was stopped were obtained from De Kalb West (Turlock, CA) by lo-fold dilution of suspensions into ice-cold Buffer and incubated by the Department of Avian Sciences, B. Samples were centrifuged in a Type 60 Ti rotor University of California, Davis. White Leghorn at 100,000 g for 60 min at 4”. Pellets were resuspended laying hens were obtained from flocks maintained in the same volume of ice-cold buffer and centrifuged by the Department of Avian Sciences. as before. These pellets were resuspended at ca. Assay for phenyl valerate hydrolysis. Phenyl 20mg protein/ml of buffer and either used valerate hydrolysis was assayed calorimetrically by immediately or stored in liquid nitrogen. the method of Johnson [17] with modifications Detergent solubilization. Two types of samples described previously [23,24]. In addition, this were prepared for gel filtration. In the first case, method has been scaled down 4-fold to yield control membranes (no pretreatment with paraoxon) Characterization of neuropathy target esterase 2589 were extracted at 1 mg protein/ml in ice-cold 0.2% LogP value for the octanol/water solvent system (w/v) Triton X-100, 0.5 M NaCl, 1 mM EDTA, was then calculated using a solvent regression 1 mM EGTA, 1 mM Tris-HCl (pH 7.2 at 21”) for equation from the literature [30,31]. 60min on ice with occasional mixing. The sample was centrifuged in a Type 60 Ti rotor at 100,OOOg RESULTS for 1 hr at 4”. The supernatant was concentrated Gel filtration purification of soluble NTE. Crude overnight at 4” against 5 L of 0.2% Triton X-100 microsomal membrane fractions were solubilized (w/v), 1 mM EDTA, 1 mM EGTA, 0.5 M NaCl, with Triton X-100 and fractionated on a Superose 20mM Tris-HCI (pH7.2 at 21”) in a Micro- 12 gel filtration column. The column had been ProDiCon apparatus with PA-30 dialysis membranes equilibrated in a buffer containing 0.5 M NaCl, (30,000 M, cut-off; Bio-Molecular Dynamics, 0.02% Triton X-100 and 0.02% asolectin. Aliquots Beaverton, OR). This resulted in a 40-fold decrease were removed from fractions and analyzed for in volume. total, paraoxon-resistant, mipafox-resistant and In the second case, paraoxon-treated membranes paraoxon + mipafox-resistant phenyl valerate hydro- were extracted at 5 mg protein/ml in ice-cold 0.3% lase activity (Fig. 1A). Results indicate that these (w/v) Triton X-100, 0.5M NaCl, 1 mM EDTA, activities were resolved into three peaks which 1 mM EGTA, 1 mM Tris-HCl (pH 7.2 at 21”) for differed in their organophosphate sensitivities. Peak 30min on ice with occasional mixing. The sample I (fractions 15-22) contained the paraoxon-resistant, was centrifuged in a Type 75 Ti (Beckman) rotor at but paraoxon + mipafox-sensitive activity known as 100,OOOg for 1 hr at 4”. The supernatant was neuropathy target esterase (NTE) (Fig. 1B). Peak recovered and used without further concentration. II (fractions 21-25) contained both a mipafox- Gelfiltration. A Superose 12 gel filtration column resistant, paraoxon-sensitive activity and a (HR 10/30, Pharmacia) was equilibrated at room paraoxon + mipafox-resistant activity. Peak III temperature in 0.02% (w/v) Triton X-100, 0.02% (fractions 27-31) contained a paraoxon-sensitive and (w/v) asolectin, 0.5 M NaCl, 1 mM EGTA, 1 mM mipafox-sensitive activity. EDTA, 20 mM Tris-HCl, pH 7.2. Samples (500 pL) The NTE peak (Peak I) eluted in the excluded were loaded and eluted at 0.25 mL/min using an volume of the column, indicating an apparent FPLC system (Pharmacia). Elution was monitored molecular weight of greater than 1,500,OOO. Fractions at 280nm, and 0.5-mL fractions were collected. 15-20 were combined and determined to have a Fractions were placed on ice as they were collected. specific activity of 0.212 I.U. of NTE/mg protein. When paraoxon-treated membranes were detergent This is a 2.7-fold purification over the starting solubilized and fractionated by gel filtration, NTE- fraction, and a71% recovery of NTE activity. Similar containing fractions were combined (see Fig. 2 for results were obtained with a Sephacryl S300 column range) and stored in liquid nitrogen. (1.5 x lOOcm, lOmL/hr, 4”, sample vol. 1.8mL). Determination of I,,, values and partition coef- When an identical sample was eluted in a buffer ficients. All I,,, determinations with the 3-substituted which did not contain asolectin, NTE eluted as a thio-l,l ,l-trifluoropropan-2-ones were performed symmetrical peak with an apparent molecular weight with gel filtration purified NTE from paraoxon- of 850,000 (results not shown). treated membranes. Rates of phenyl valerate In an attempt to eliminate the paraoxon-sensitive hydrolysis were determined as described above, activity (Peak II) which overlaps with NTE, except that samples were not preincubated for 20 min membrane fractions were pretreated with paraoxon. at 37” with mipafox and paraoxon. Phenyl valerate These membranes were then solubilized with Triton and NTE were added simultaneously to tubes X-100 and fractionated on a Superose 12 column containing appropriate concentrations of inhibitor. (Fig. 2). This process of treating the membranes Assay tubes were incubated at 37” for 30 min prior with paraoxon successfully removed the paraoxon- to addition of SDS to stop substrate hydrolysis. sensitive activity in Peak II. Unexpectedly, the Inhibitors were added in a volume of 10 PL of paraoxon-sensitive and mipafox-sensitive activity in ethanol. Ethanol controls were included in each Peak III was not eliminated by this treatment. After assay. To determine I,,, values, a broad titration was washing paraoxon-treated membranes to remove first performed with inhibitor concentrations in the nonbound paraoxon, this peak of paraoxon-sensitive range of 1O-4-1O-y M. This was followed by a more activity was still present in the activity profile. When complete titration curve in the linear region of a this fractionation was performed using membranes percent activity vs -log [inhibitor] plot. All inhibitor treated simultaneously with both paraoxon and concentrations were assayed in triplicate. Iso values mipafox, Peak III was still present (data not shown). were determined by linear regression analysis of Preliminary experiments, however, appear to narrow titrations in which a minimum of two points indicate that the degree of reactivation was not were above 50% inhibition and two points were equivalent to that seen when membranes were below 50% inhibition. Reported Iso values are the treated with paraoxon alone. average of two such narrow titrations. Fractions 16-21 containing the NTE activity were Octanol/water partition coefficients were cal- combined and assayed. In the three preparations culated by the FRAGMENT method of Hansch and used in the TKF inhibition experiments, Leo [27,28]. The partition coefficient for OTFP paraoxon + mipafox-resistant activity represented was measured [29,30] at room temperature in an average of 4.5% (SD,_ 1 = 1.3%) of the total cyclohexane, using an HP-5890A gas chromatograph activity. This compares to 17% paraoxon + mipafox- with a DB-17 megabore column (F & W Scientific, resistant activity in brain homogenates. These Folsom, CA) and an electron capturing detector. combined fractions were stored as 1-mL aliquots in 2590 T. C. THOMASet al.

0.18 ,

0.14 0.12 - 0.10 - 0.08 - 0.06 -

0.04 -

z 14 18 22 26 30 34 38

0.10

0.09

0.08 0.07 0.06

0.05 0.04 0.03 0.02 0.01 0.00 14 18 22 26 30 34 38 Fraction

Fig. 1. Gel filtration of Triton X-lOO/NaCl solubihzed phenyl valerate hydrolases from a microsomal preparation of chick embryo brain. Proteins were solubilized and fractionated on a Superose 12 column using a Pharmacia FPLC system as described in Materials and Methods. (A) Individual aliquots from each fraction were incubated with either buffer (0), paraoxon (0). mipafox (A), or paraoxon + mipafox (A) prior to addition of substrate. Peaks have been labeled I, II and III to correspond with text. (B) NTE activity is plotted as the difference between paraoxon-resistant and paraoxon + mipafox-resistant activities.

liquid nitrogen. No significant loss of activity was The activity of samples was inhibited an average of observed in samples stored up to 13 months. 44% at this inhibitor concentration regardless of the Velocity us substrate concentration. The relation- length of incubation prior to addition of substrate. ship of velocity to substrate concentration was In a related experiment, a sample was incubated analyzed using fractions 15-20 from the gel filtration with 2.5 x lo-’ M OTPF at 37” for 6 min (sample D fractionation profiled in Fig. 1. The maximum in Table 1). This sample was then diluted 25-fold velocity obtained using substrate concentrations up and assayed for phenyl valerate hydrolysis at a final to 10 mM was 1.1 x 10m3 I.U. V,,,,, appeared to be inhibitor concentration of 1 x 10e8M OTFP. Con- reached at substrate concentrations between 1 and trol samples, which did not contain inhibitor, were 1.5 mM which is in the region of maximum substrate incubated and then diluted as above. These were solubility. A Lineweaver-Burk reciprocal plot of assayed in the presence of either 2.5 x lo-' M OTFP eight substrate concentrations from 0.2 to 0.9 mM (sample B) or 1 x 10es OTFP (sample C). Upon 25- indicated that the K,,, of phenyl valerate for NTE fold dilution in assay buffer, sample D, which had was 5.33mM (r = 0.999) and the V,,, was been incubated at 2.5 x lo-’ M OTFP, behaved as 5.1 x lo-s1.u. if it were being inhibited at the post-dilution con- Time-dependent incubation with trifluoroketones. centration of 1 x 10d8 M (Table 1). The same results Superose 12 fractions 16-21 obtained from paraoxon- were achieved when samples were incubated at 4” pretreated membranes were incubated with for 20 hr. In an additional experiment, it was 7 x 10-s M 3-octylthio-l,l,l-trifluoropropan-2-one observed that the inhibition of samples with (OTFP) for 0, 5, 10 and 20 min prior to addition of 1 x 10m3 M OTFP could be reversed rapidly and substrate, and hydrolysis was stopped after 30 min. completely by gel filtration on a desalting column. Characterization of neuropathy target esterase 2591

2 0.10-A III 0 .$x 0.09 - ‘Z 0.08 - 2 0.07 -

25 30 35 40 Fraction

Fig. 2. Gel filtration of Triton X-lOO/NaCl solubilized phenyl valerate hydrolases from a paraoxon- treated membrane fraction. Membrane fraction from chick embryo brains was treated with paraoxon prior to treatment with Triton/NaCl extraction buffer. Methods and symbols are as described in Fig. 1. The arrowed line in panel B indicates the region of the NTE peak which was combined and used to screen inhibitors.

Determination of Z5,, values. The inhibitory capa- cities of twenty-two 3-substituted thio-l,l,l-tri- Table 1. Reversible inhibition of neuropathy target esterase fluoropropan-Zones against NTE were determined by 3-octylthio-l,l,l-trifluoropropan-2-one (OTFP) (Table 2). These compounds fall into two groups containing either an n-alkylthio group (N = 4-12) or OTFP concn (M) NTE activity (%) a substituted arylthio group. Titrations were per- formed by simultaneous addition of both sample Sample Incubation Assay 6 min 20 hr and substrate to assay tubes containing appropriate concentrations of inhibitor. Assay tubes were incu- A 0 0 100 100 bated immediately for 30min at 37”, followed by B 0 2.5 x lo-’ 16.7 19.8 determination of released phenol. A broad titration C 0 1.0 x 10-R 72.2 73.6 D 2.5 x lo-’ 1.0 x lo-* 85.0 83.6 curve was obtained for each inhibitor, followed by two narrow titrations in the linear region of the curve Identical samples of gel filtration purified NTE from (Fig. 3). Reported I,, values are the average of the paraoxon-treated membranes were incubated for either two I,,, values obtained by linear regression of each 6 min at 37” or 20 hr at 4”. In each case, sample D contained narrow titration (Table 2). 2.5 X lo-‘M OTFP. Samples were then diluted 25-fold Octanol/water partition coefficient. The octanol/ into assay buffer, and OTFP was added to samples B and C. Samples were incubated at 37” for 7 min prior to addition water partition coefficients for the carbonyl form of of substrate. One hundred percent NTE activity at 6 min each TFK were calculated and are presented as log P was 0.027 I.U./mL and at 20 hr was 0.026 I.U./mL. The (Table 2). In addition log P was calculated for several volume of sample used in the assay was 30 PL. carboxylesterase substrates including phenyl valerate 2592 T. c. TliOMAS et al.

Table 2. Logarithmic octanol/water partition coefficients and inhibitory potencies for 3- substituted thio-l,l,l-trifluoropropan-2-ones

Inhibitor Compound No. (R-) log P Iso CM) (slope)*

1 butyl- 1.54 7.42 x 1O-6 (47) 2 hexyi- 2.62 2.87 x 1O-7 (50) 3 heptyl- 3.16 9.50 x 10-x (53) octyl- 3.70 5.88 x lo-* (56) : decyl- 4.78 7.15 x 10-R (52) 6 undecyl- 5.32 1.09 x lo-.’ (45) I dodecyl- 5.86 9.64 x IO-* (51) 8 cyclohexyl- 2.31 1.59 x lo+ (56) 9 benzyl- 1.59 3.72 x lWh (51) 10 phenyl- I.49 2.37 x lo+’ (55) 11 2-chlorophenyl- 2.20 2.96 x lo-.” (57) 12 3-chlorophenyl- 2.20 1.62 x lo-’ (49) 13 4-chlorophenyl- 2.20 2.01 x Wh (53) 14 3,4-dichlorophenyl- 2.91 1.76 x lo-’ (55) 15 2,6-dichlorophenyl- 2.91 1.26 x 1O-J (63) 16 2,.5-dichlorophenyl- 2.91 6.80 x lo-’ (52) I7 2methoxyphenyl 1.47 4.30 x 10-s (50) 18 3-methoxyphenyl- 1.47 8.28 x lo-’ (54) 19 2-bromophenyl- 2.35 1.25 x Wi (61) 20 3-methyl-4-bromophenyl- 3.01 1.97 x lo-’ (57) 21 3-trifluoromethylphenyl- 2.37 2.90 x lo-’ (51) 22 il-tert-butylphenyl- 3.38 4.93 x Io-“i (53)

phenyl valerate 3.11 phenyl 2-(octylthio)acetate 4.07

* Iso values were determined-by linear regression of titration curves generated as described in the legend of Fig. 3. Reported values are the average of two determinations.

100 90 80

60 .

40

20

; 0 7.8 7

t 1 1 9 8 7 6 5 4 -Log [OTFP] (U)

Fig. 3. Titration of paraoxon-pretreated, Superose 12 purified NTE with OTF’F. A broad titration was performed (full figure) to determine the linear range of the curve. One hundred percent activity is 0.029 I.U./mL. Two separate titrations were performed in this range (inset; 100% activity is 0.032 I.U./ mL). A separate regression line is plotted for each titration (inset). Error bars on the broad titration represent +- one standard deviation (N-l). Full figure labels also apply to inset figure. Characterization of neuropathy target esterase 2593

:: RSCH2CCF3

5 419

Fig. 4. Relationship of inhibitory potency to lipophilicity ( pIscr vs log P). The PI,, and log P values of twenty-two 3-substituted thio-1 ,l,l-trifluoropropan-2-ones were determined as described in Materials and Methods. Substituents were either n-alkyl or substituted phenyl groups. Compounds containing a phenyl ring mono- or di-substituted in the ortho position have been plotted with open triangles (a), while those substituted with an electron withdrawing group in the meta position have been plotted with open inverted triangles (D). Those compounds which are not included in the above two groups were plotted as filled triangles (A). A regression line has been plotted for the members of this last group found in the range of log P between 1.5 and 3.4. An arrow indicates the log P value of phenyl valerate. The number next to each symbol corresponds to the Inhibitor No. in Table 2.

and phenyl octylthioac~tate (Table 2). The relation- (~~~c~op~~~u ni), acetyl cholinesterase (electric eel), ship between lipophilicity and inhibitory potency is trypsin (bovine) and chymot~psin (bovine} by presented as PI,, vs log P (Fig. 4). OTFP are 2.3 x 10ey M, 2.3 x fO+jM, and The partition coefficient for the distribution >l x lO‘sM, and >l X 10e4M [1.5]. Johnson between an organic and an aqueous phase was reported that the organophosphate sensitivity of measured for OTFP. The octanol/water partition NTE was very similar to that of a-chymotrypsin and coefficient could not be measured directly, due to trypsin, but dissimilar to that of acetyl cholinesterase the formation of a hemiacetal with octanol. There- [ll]. In contrast, these results indicate that tri- fore, the partition coefficient was measured in fluoromethyl ketones are very good inhibitors of cyclohexane and was calculated for octanol, using esterases such as NTE, juvenile hormone esterase the solvent regression equation [30,3lj and acetyl cholinesterase, but are poor inhibitors of peptidases. log K,/, = 0.941 log Kcjw + 0.69 The rapidly reversible inhibition of NTE by these where partition coefficients are K+, for the octanol/ TFKs indicates that they are concentration-depen- water and K+, for the cyclohexane/water systems. dent, competitive inhibitors of NTE. This is similar Log K+ was measured to be 2.21 * 0.11, and log to the kinetics of inhibition of acetyl cholinesterase K,/, was calculated to be 2.77 t 0.09. This value is by these compounds but contrasts with the kinetics 0.93 less than the log P value calculated by the of juvenile hormone esterase inhibition which show FRAGMENT method of Hansch and Leo (Table 2) slow tight binding [32,33]. Juvenile hormone ester- [27,28]. ase requires a period of incubation with inhibitor prior to addition of substrate. No time-dependent DISCUSSION inhibition of NTE by these trifluoroketones was observed. so Is0 titrations were performed by sim- The relationship between the structural and ultaneous addition of sample and substrate to assay physicochemical characteristics of twenty-two ali- tubes containing appropriate concentrations of phatic and aromatic 3-substituted-l,I,l-trifluoro- inhibitor. The failure to observe time-dependent propan-Zones and their inhibitor potencies inhibition of NTE may be partially due to the low toward NTE was investigated. These compounds, occupancy of catalytic sites by phenyl valerate. The which resemble the transition state intermediate of high K,,, (l-10 mM) and low solubility (1.5 mM) of many inhibitors of serine active site hydrolases, were phenyl valerate may result in less than 50% satu- shown to be very effective inhibitors of NTE. The ration of NTE catalytic sites [18, 191. two best inhibitors were identified as 3-octylthio- To characterize more fully the interaction of NTE l,l,l-trifluoropropan-2-one (OTFP) and 3-[4-tert- with the TFKs, the inhibitory potency of each TFK butyl] - phenylthio - 1 ,l ,l - trifluoropropan - 2 - one, against NTE was correlated to several hydrophobic which have Is,, values of 5.88 X 10-‘M and parameters, including the octanol/water partition 4.93 X 10m8M respectively. I3y comparison, the Iso coefficient of the molecules, molar refractivity, and values for inhibition of juvenile hormone esterase n lipophilicity substituent constants of the alkyd or 2594 T. C. THOMAS etal. the aryl substituent group on the sulfur. Since these values for the carbonyl forms. As such, the calculated parameters are highly colinear by nature (cross cor- log P value for OTFP (3.70) and the measured log P relation = 0.87), the one with the highest correlation value for OTFP (2.77) differed by almost one order to the PI,,, values, in this case the octanol/water of magnitude. Since the hydration/dehydration equi- partition coefficient (P), is discussed. librium is unlikely to be the rate-limiting step in the Figure 4 presents the relationship between the PI,, enzyme inhibition 1351, the form which is actually and the log P values of each TFK. In the range of involved in the inhibition of the enzyme should be lower lipophilicities (log P = 1.5 to 3.5), the inhi- used to calculate 1ogP. The transition state mimic bition of NTE appears to be a linearly increasing theory [36, 371 assumes that the geminal diol form, function of lipophilicity on the double-logarithmic which is tetrahedral in geometry, is responsible for plot. A regression line for appropriate compounds inhibition. In this case the difference between the within this range is plotted and defined by the calculated and measured log Pvalues of OTFP (0.93) equation pl,, = 1.014 log P + 3.74 (p.= 0.958). can be used as a general correction factor for all Compounds which appeared to have steric or elec- TFKs. This correction would result in the structure- tronic contributions (discussed later) to their inhibi- activity relationship curve (Fig. 4) being shifted down tory capacity were not included in this regression by about one unit along the log P axis. It should be analysis. In the higher log P zone (log P = 3.5 to noted, however, that due to the electronic par- SS), inhibition does not increase further with log P. ameters of substituents on the sulfur, the equilibrium This saturation at higher lipophilicity is also observed between the carbonyl and geminal diol forms varies with inhibition of juvenile hormone esterase by TFKs among the different compounds 1381. In contrast [15,20]. to the transition state mimic theory, it has been The above results should provide useful infor- suggested that in the lipophilic microenvironment mation for the synthesis of more specific and sensitive of the enzyme surface, the geminal diol form may substrates and inhibitors for NTE. Analysis of the undergo a dehydration [32]. Inhibition would then relationship between PI,,, and log P indicates that occur in an addition reaction between the enzyme the optimal compromise between lipophilicity and and the carbonyl form of the inhibitor. This is con- solubility of NTE substrates and inhibitors is likely sistent with recent evidence indicating that acetyl- to be found in the range of log P between 3.0 and 3.4. cholinesterase is inhibited by the carbonyl form of This is consistent with the fact that phenyl valerate, trifluoromethyl ketones [39]. which is more rapidly hydrolyzed and more selective NTE is one of four classes of phenyl valerate for NTE than phenyl butyrate and phenyl caproate, hydrolases found in brain tissue. These classes are has a calculated log P of 3.11. Phenyl butyrate and distinguished by their differential sensitivities to the phenyl caproate, which differ from phenyl valerate organophosphates mipafox and paraoxon. NTE is by only one methylene group, have 1ogP values identified as the difference in activity between paired outside this range (2.45 and 3.79 respectively). When samples, one of which was incubated with paraoxon phenyl2-(octylthio)acetate, which has a log P value while the other was incubated with paraoxon and of 4.07, was tested as a hydrolase substrate, it too mipafox. However, the presence of paraoxon and was found to be a sensitive and selective substrate mipafox in this differential assay can interfere with of NTE 1121. Our results indicate that phenyl 2- the determination of the sensitivity of NTE to other (heptylthio)acetate, which has a log P value of 3.41, inhibitors. There are now several lines of evidence may be a more selective substrate of NTE than that paraoxon can competitively decrease the rate phenyl 2-(octylthio)acetate. of phosphorylation of NTE and other proteins by Two additional trends appear upon analysis of mipafox [24,40], our unpublished results). It is these TFKs. First, substitution of the phenyl group reasonable to assume that paraoxon can have this at the ortho position reduced the inhibitory potency effect on other inhibitors, and should be removed of the compound. This was probably due to steric prior to inhibitor titration of NTE. However, Figs. hindrance. Second, substitution of the phenyl ring at 1 and 2 clearly show that when nonbound paraoxon the meta position with an electron withdrawing group was removed prior to gel filtration, the paraoxon- or enhanced inhibitory potency compared to sub- mipafox-sensitive activity in Peak III was reactiv- stitution at the para position. This may indicate an ated. Experiments with membranes that were treated electronic interaction between the enzyme and sub- simultaneously with paraoxon and mipafox indicate stituents in the meta position. Unfortunately, the that when nonbound paraoxon and mipafox are number of different compounds included in these removed, this esterase may not be fully reactivated. groups was insufficient to determine a quantitative If there is a differential reactivation of the activity in structure-activity relationship for steric and elec- Peak III, then the difference between titrations of tronic parameters. With further study, some of the paraoxon and paraoxon + mipafox-treated mem- trifluoromethyl ketones may be of value as diagnostic branes does not yield a titration curve for NTE inhibitors for NTE activity. They offer the advan- solely. tages of high inhibitory activity and ease of synthesis. We have not provided evidence that the presence In addition, studies in both mice [IS] and chickens or removal of paraoxon and mipafox prior to titration (unpublished} indicate that these compounds have of NTE yields substantially different IsOvalues. How- very low toxicity and are non-neuropathic. ever, the potential drawbacks to these methods bring It has been demonstrated [34] that fluoromethyl into question their suitability for the type of charac- ketones predominantly exist in their hydrated form, terization pursued in this study. To avoid this issue, as geminal diols. However, it was only possible to a method for isolating an essentially pure NTE- use log P fragment constants to calculate the log P type phenyl vaierate hydrolase fraction was sought. Characterization of neuropathy target esterase 2595

Previous chromatographic fractionation of detergent Dudek BR and Richardson RJ, Evidence for the exist- solubilized fractions has indicated that the different ence of neurotoxic esterase in neural and lymphatic types of phenyl valerate hydrolases are separable tissue of the adult hen. Biochem Pharmacol31: 1117- [21,41,42]. Sucrose gradient centrifugation suc- 1121, 1982. Davis CS and Richardson RJ, Neurotoxic esterase: ceeded in separating NTE from the majority of other Characterization of the solubilized enzyme and the phenyl valerate hydrolases [21,24], but yields of conditions for its solubilization from chicken brain active enzyme were very poor and there was a loss microsomal membranes with ionic, zwitterionic, or of specific activity. Gel filtration chromatography has nonionic detergents. Biochem Pharmacol 36: 1393- resulted in better yields and increases in specific 1399, 1987. activity, but the phenyl valerate hydrolase activities Johnson MK, A phosphorylation site in brain and the are poorly resolved ([41], our unpublished results). delayed neurotoxic effect of some organophosphorus In the present study, we have improved the ability compounds. Biochem J 111: 487-495, 1969. of gel filtration to completely resolve NTE by selec- Johnson MK. The delayed neurotoxic effect of some tively reconstituting it into vesicles during frac- organophosphorus compounds. Biochem J 114: 711- 717. 1969. tionation. This was achieved by including asolectin Johnson MK. Organophosphorus and other inhibitors in the elution buffer and by decreasing the con- of brain “neurotoxic esterase” and the development of centration of detergent in the buffer. Without aso- delayed neurotoxicity in hens. Biochem J 120: 523-531, lectin present, NTE migrated as a single peak with 1970. an apparent molecular weight of 850,000. This is Johnson MK, The primary biochemical lesion leading consistent with the previously reported value of to the delayed neurotoxic effects of some organo- 880,000 for 3-[(3-cholamidopropyl)dimethyl- phosphorus esters. J Neurochem 23: 785-789, 1974. ammoniol-l-propanesulfonate (CHAPS) solubilized 8. Cavanagh JB. The toxic effects of tri-orrho-cresyl NTE [41]. Upon inclusion of asolectin, NTE migrated phosphate on the nervous system. J Neural Neurosurg in the void volume of the column (M, > 1.5 X 106). Psychiatr 17: 163-172, 1954. This significantly reduced the overlap between Peak 9 Cavanagh JB, Peripheral nerve changes in ortho-cresyl phosphate poisoning in the cat. J Pathol Bacterial 87: I, containing NTE, and Peak II, containing 365-383, 1964. paraoxon +mipafox-resistant activity (Fig. 1). 10 Lotto M, Caroldi S, Moretto A, Johnson MK, Fish A final consideration regarding the use of this gel CJ. Gospinath C and Roberts NL, Central-peripheral filtration purified NTE fraction is the possibility that delayed neuropathy caused by diisopropyl phos- detergent solubilization may have altered the charac- phorofluoridate (DFP): Segregation of peripheral teristics of the enzyme [43]. Davis and Richardson nerve and spinal cord effects using biochemical, clini- [3] have reported that solubilization of NTE with cal, and morphological criteria. Toxicol Appl Phar- Triton X-100 does not alter the sensitivity of NTE macol88: X7-96, 1987. for mipafox. In addition, we have determined that 11 Johnson MK, Structure-activity relationships for sub- strates and inhibitors of hen brain neurotoxic esterase. the K,,, of this NTE preparation for phenyl valerate is 5.3 mM. Previous investigations using intact mem- Biochem Pharmacol24: 797-805, 1975. 12. Johnson MK, Sensitivity and selectivity of compounds branes have reported K, values of 1 and 10 mM interacting with neuropathy target esterase. Biochem [18,19]. These results confirm that Triton X-100 Pharmacol37: 4095-4104. 1988. solubilized NTE has the same characteristics as mem- 13. Gelb MH. Svaren JP and Abeles RH, Fluoro ketone brane associated NTE. inhibitors of hydrolytic enzymes. Biochemistry 24: A method has been described for the rapid, high 1813-1817, 1985. yield preparation of an NTE fraction which is suitable 14. Ashour MBA and Hammock BD, Substituted trifluoro- for study without the use of differential inhibition ketones as potent selective inhibitors of mammalian protocols. This eliminates the drawbacks inherent in carboxylesterases. Biochem Pharmacol36: 1869-1879, previously used methods and results in considerable 1987. savings of time and sample. In addition, this fraction 15. Hammock BD, Abdel-Aal YAI, Mullin CA, Hanzlik can be stably stored in liquid nitrogen, so that TN and Roe RM, Substituted thiotrifluoropropanones as potent selective inhibitors of juvenile hormone ester- numerous inhibitors can be tested against the same ase. Pestic Biochem Physiol 22: 209-223, 1984. preparation. 16. Abdel-Aal YAI and Hammock BD, Use of transition state theory in the development of bioactive molecules. In: Bioregulators for Pest Control (Ed. Hedin PA), Acknowledgements-We thank Dr. Marcello Lotti for his ACS Symp. Ser. No. 276. pp. 135-160. American advice and criticism and Dr. Richard Criddle for reviewing Chemical Society. Washington. DC, 1985. the manuscript. This work was supported in part by 16. Abdel-Aal YAI and Hammock BD, Use of transition National Institutes of Health Grant ES00202. T.C.T. was a trainee on N.I.H. Training Grant GM07377. A.S. is a state theory in the development of bioactive molecules. Fulbright Scholar (Fulbright Program No. 33917, Institute In: Bioregulators for Pest Control (Ed. Hedin PA), of International Education) and received funding from ACS Symp. Ser. No. 276, pp. 135-160. American N.I.H. Grant ES02710-10. B.D.H. is a Burroughs Well- Chemical Society, Washington, DC. 1985. come Scholar in Toxicology. 17. Johnson MK, Improved assay of neurotoxic esterase for screening organophosphates for delayed neurotoxicity potential. Arch Toxicol37: 113-115, 1977. 18. Johnson MK, Delayed neurotoxicity induced by organ- REFERENCES ophosphorus compounds-Areas of understanding and ignorance. In: Mechanisms of Toxicity and Hazard 1. Richardson RJ, Davis CS and Johnson MK, Subcellular Evaluation (Eds. Holmstedt B, Lauwerys R, Mercier distribution of marker enzymes and of neurotoxic ester- M and Roberfroid M), pp. 27-38. Elsevier/North Hol- ase in adult hen brain. J Neurochem 32: 607-615, 1979. land, Amsterdam. 1980. 2596 T. C. TH(3M.w et al.

19. Carrington CD and Abou-Donia MB, Kinetics of sub- PP, Philpott M and Matolcsy G, New trifluo- strate hydrolysis and inhibition by mipafox of paraoxon ropropanone sulfides as highly active and selective pre-inhibited hen brain esterask aciivity. Biochem J inhibitors of insect juvenile hormone esterase, Pestic 236: 503-507, 1986. Biochem Phys~o~ 33: 112-124, 1989, 20. S&k&es A, B&d&s B, Matolcsy G and Hammock BD, 33. Abdel-Aal YAI and Hammock BD, Apparent multiple Quantitative structure-activity relationship study of catalytic sites involved in the ester hydrolysis of juvenile aromatic tri~uoromethylketones, in vitro inhibitors of hormones by the hemolymph and by an affinity-purified insect juvenile hormone esterase. In: Probing Bioactioe esterase from Munduca sexta Johannson (Lepidoptera: Mech&isms (Eds. Magee P, Block J and Henry D), Sphingidae). Arch Biochem Biophys 243: 206-219, ACS Symposium Series, Vol. 413, pp. 169-182. Ameri- 1985. can Chemical Society, Washington, DC, 1989. 34. Linderman RL. Leazer J. Roe RM. Venkatesh K. 21. Ishikawa Y, Chow fl, McNamee MG, McChesney M Selinsky BS anb London KE, ‘gF-Nh& spectral evi- and Wilson BW, Separation of paraoxon and mipafox dence that 3-octylthio-l,l,l-trifluoropropan-2-one, a sensitive esterases by sucrose density gradient sedi- potent inhibitor of insect juvenile hormone esterase, mentation. Toxicof Lett 17: 315-320, 1983. functions as a transition state analog inhibitor of acetyl- 22. Johnson MK, Check your paraoxon and parathion for cholinesterase. Pestic Biochem Fhysiof 31: 187-194, neurotoxic impurities. Vet Hum Toxicul24: 220, 1982. 1988. 23. Soliman SA and Curley A, Assay of chicken brain 35 SzCk&cs A, Hammock BD, Abdel-Aal YAI, Philpott neurotoxic esterase activity using leptophosoxon as the M and Matolcsy G, Inhibition of iuvenile hormone selective neurotoxic inhibitor. J Anal Toxicol 5: X3- e&erase by transition state analogs: k tool for enzyme 186, 1981. molecular biology. In: Biotechnology in Crop Pro- 24. Thomas TC, Ishikawa Y, McNamee MG and Wilson tection (Eds. Hedin PA, Menn JJ and Hollingwo~h BW, Correlation of neuropathy target esterase activity RM), ACS Symp. Ser. No. 379, pp. 215-227. American with specific tritiated di-isopropyl phosphorofluoridate- Chemical Societv, Washington. DC. 1988. labelled proteins. Biochem J 257: 109-116, 1989. 36. Pauling L, Chemical achievement and hope for the 25. Lowry OH, Rosebrough NJ, Farr AL and Randall RJ, future. Am Sci 36: 51-58, 1948. Protein measurement with the Folin phenol reagent. I 37. Wolfenden R, Transition state analog inhibitors and Biol Chem 193: 265-275, 1951. enzyme catalysis. Annu Rev Siophys45: 271-306,1976. 26. Wang CS and Smith RL, Lowry determination of pro- 38, Schierling T, Part I. Addition Equilibria of Tri- tein in the presence of Triton X-100. Anal ~iochem 63: ~uoromethyl Ketone Inhibitors of Juvenile Hormone 414-417, 1975. Esterase-A Quantitative Structure-Activity Relation- 27. Hansch C and Leo A, Substituenr Constants for Cor- ship Study. Ph.D. Dissertation, University of Califor- relation Analysis in Chemistry and Biology. John Wiley, nia, Davis, 1988. New York, 1979. 39. Allen KN and Abeles RH, Inhibition kinetics of acetyl- 28. Lyman WJ, Octanol/water partition coefficient. In: cholinesterase with fluoromethyl ketones. geochemistry Hundbook of Chemical Property Estimation Methods 28: 8466-8473, 1989. (Eds. Lyman WJ, Reehl WF’ani Rosenblatt DH), pp. 40. Carrington CD and Abou-Donia MB, Paraoxon revers- 1.1-1.53. McGraw-Hill. New York. 1982. ibly inhibits neurotoxic esterase. ToxicoI Appl Phar- 29. Karickhoff SW and l&own DS, ‘Determination of macol79: 17.5-178, 1985. OctanoljWater Distribution Coeficients, Water Solu- 41. Pope CN and Padilla SS, Chromatographic charac- bilities, and Sediment/Water Partition Coefficients for terization of neurotoxic esterase. Biochem Pharmacol Hydrophobic Organic Pollutants Report No. EPA- 38: 181-188, 1989. 600/4-79-032. US Environmental Protection Agency, 42. Chemnitius JM, Haselmeyer KH and Zech R, Neu- Athens, GA, 1979. rotoxic esterase: Gel filtration and isoelectric focusing 30. Leo A, Hansch C and Elkins D, Partition coefficients of carboxylesterases solubiiized from hen brain. Lifg and their uses. Chem Rev ‘71: 525-621, 1971. Sci 34: 119-125. 1984. 31. Rekker RF, The Hydrophobic Fragment Constant. 43. Johnson MK, S&ubilization procedures cause changes Elsevier, New York, 1977. in the response of brain “neurotoxic esterase” to inhibi- 32. Sz&k&csA, Hammock BD, Abdel-Aal YAI, Halarnkar tors. Biochem J 122: 51P-52P, 1971.