Inactivation of Dihydropyrimidine Dehydrogenase by 5-Iodouracil*

6-Iodouracil was a substrate for bovine liver dihy- dropyrimidine dehydrogenase (DHPDHase) and was a potent inactivator of the enzyme. NADPH increased the rate of inactivation and thymine protected against inactivation. These findings suggest that 5-iodouracil was a mechanism-based inactivator. However, dlthio-threitol and excess 5-iodouracil protected the enzyme against inactivation. Thus, a reactive product, presumably 6-iodo-6,6-dihydrouracil generated through the enzymatic reduction of 5-iodouracil, was released from DHPDHase during processing of 5-iodouracil. Since only 18% of [6-3H]5-iodouracil reduced by DHPDHase was covalently bound to the enzyme and radiolabel was not lost to the solvent as tritium, the partition coefficient for inactivation was 4.5. However, the enzymatic activity was completely titrated with 1.7 mol of 5-iodouracil per mol of enzyme-bound flavin. These results indicate that there was 0.31 mol of enzyme- bound inactivator per mol of enzyme flavin. This suggests there were 3.2 flavins per active site, which is consistent with the report of multiple flavins per enzymic subunit (Podschun, B., Wahler, G., and Schnack-erz, oxidation of 1 pmol of NADPH per h. Protein concentration was estimated from absorbance at 274 and was ex- pressed in terms of absorbance units at 274 nm (AZ7,). Specific activities are expressed as activity unit~/A,,~. Enzymatic activity was also measured spectrophotometrically by following the reduction of 5-iodouracil at 283 nm, an isosbestic point between NADPH and NADP'. The Ae2= for the reduction of 5-iodouracil to 5-iodo-5,6- dihydrouracil was 5.9 mM"cm" was an equal mixture of the two isomers, the enantiomer corresponding to enzymatically generated 5-iodo-5,6-di- hydrouracil (1.05 mol (2.1/2)) inactivated 0.19 mol of enzyme (1.05/ 5.5). This meant that 0.81 mol (1.00 - 0.19) of enzyme was inactivated by 1.05 mol (2.1/2) of the opposite enantiomer to yield a stoichiometry of 1.3 (1.05/0.81). D. J. T. Porter,

6-Iodouracil was a substrate for bovine liver dihydropyrimidine dehydrogenase (DHPDHase) and was a potent inactivator of the enzyme. NADPH increased the rate of inactivation and thymine protected against inactivation. These findings suggest that 5-iodouracil was a mechanism-based inactivator. However, dlthiothreitol and excess 5-iodouracil protected the enzyme against inactivation. Thus, a reactive product, presumably 6-iodo-6,6-dihydrouracil generated through the enzymatic reduction of 5-iodouracil, was released from DHPDHase during processing of 5-iodouracil. Since only 18% of [6-3H]5-iodouracil reduced by DHPDHase was covalently bound to the enzyme and radiolabel was not lost to the solvent as tritium, the partition coefficient for inactivation was 4.5. However, the enzymatic activity was completely titrated with 1.7 mol of 5-iodouracil per mol of enzyme-bound flavin. These results indicate that there was 0.31 mol of enzymebound inactivator per mol of enzyme flavin. This suggests there were 3.2 flavins per active site, which is consistent with the report of multiple flavins per enzymic subunit ( DHPDHase was inactivated by 2.1 mol of racemic 5iodo-6,6-dihydrouracil per mol of active sites. The stoichiometry for inactivation of the enzyme by the nonenzymatically generated enantiomer of 5-iodo-5,6-dihydrouracil was calculated to be 1. Two radiolabeled fragments were isolated from a tryptic digest of DHPDHase inactivated with radiolabeled 5-iodouracil. The amino acid sequences of these peptides were Asn-Leu-Ser-X-Pro-His and Asn-Leu-Ser-X-Pro-His-Gly-Met-Gly-Glu-Arg whereXwas the modified amino acid containing radiolabel from [6-3H]5-iodouracil. Fast atom bombardment mass spectral analysis of the smaller peptide yielded a protonated parent ion mass of 782 daltons that was consistent with X being a S-(hexahydro-2,4-dioxo-5-pyrimidinyl)cysteinyl residue.
Dihydropyrimidine dehydrogenase (DHPDHase, EC 1.3.1.2)' catalyzes the reversible reduction of pyrimidines t o 5,6-dihydropyrimidines as the first step in pyrimidine catab-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  olism to p-amino acids (1). The enzyme has been purified to homogeneity from mouse and porcine liver and has been shown to contain multiple flavins and iron sulfur prosthetic groups per subunit (2,3). The purified enzyme rapidly reduces the anticancer drug 5-iodouracil to 5-fluoro-5,6-dihydrouracil (2,3). DHPDHase is estimated to degrade over 80% of an administered dose of 5-fluorouracil (4). The variability of the effectiveness of 5-fluorouracil may be due, in part, to the variability of the cellular levels of DHPDHase (5). The activity of this enzyme not only varies between individuals but also has a temporal variation in an individual with a circadian periodicity of 24 h (6). Consequently, DHPDHase has emerged as a potentially important adjunct target in fluorouracil chemotherapy. The inhibitory properties of a large number of pyrimidine analogs for DHPDHase have recently been summarized (7). Some of these analogs have submicromolar Ki values. Only 5-bromovinyluracil and 5-diazouracil have been reported to be irreversible inactivators of this enzyme (8,9). 5-Bromovinyluracil inhibits the enzyme in uiuo and causes uracil levels in the plasma of mice to increase (10). Furthermore, the half-life of 5-fluorouracil in mice increases when 5fluouracil is coadministered with 5-bromovinyluracil (8,11). Thus, inhibitors of DHPDHase are useful for potentiating the efficacy of 5-fluorouracil (8,(10)(11)(12) and for controlling the variability in 5-fluorouracil availability.
The intravenous infusion of 5-iodo-2'-deoxyuridine into human subjects generates high plasma levels of 5-iodouracil with 50-100-fold increases in plasma levels of thymine and uracil (13). The elevated uracil and thymine levels were suggested to be the result of simple competition between plasma pyrimidines and 5-iodouracil for DHPDHase. Since 5-iodouracil, an efficient substrate for DHPDHase, can compete effectively with the natural substrate thymine (2,14), this hypothesis provides an attractive explanation for the elevated levels of pyrimidines during infusion of 5-iodo-2'-deoxyuridine. Alternatively, the product of DHPDHase-catalyzed reduction of 5-iodouracil, 5-iodo-5,6-dihydrouracil, is expected to be an alkylating agent with properties similar to those of iodoacetamide (Equation 1). Consequently, the elevated plasma levels of pyrimidines in patients treated with 5-iodo-2'-deoxyuridine could be, in part, due to inactivation of DHPDHase by enzymatically generated 5-iodo-5,6-dihydrouracil. We have addressed this possibility by investigating the interaction of 5-iodouracil with purified bovine liver DHPDHase and have found that 5-iodouracil potently inactivated this enzyme.

Assay of DHPDHase
DHPDHase was routinely assayed at 37 "C in buffer A (0.05 M Tris-HC1 at pH 8.0) with 1 mM DTT, 200 p~ NADPH, and 200 p~ thymine. The oxidation of NADPH was followed at 340 nm. 1 unit of enzyme catalyzed the oxidation of 1 pmol of NADPH per h. Protein concentration was estimated from absorbance at 274 and was expressed in terms of absorbance units at 274 nm (AZ7,). Specific activities are expressed as activity unit~/A,,~. Enzymatic activity was also measured spectrophotometrically by following the reduction of 5-iodouracil at 283 nm, an isosbestic point between NADPH and NADP'. The Ae2= for the reduction of 5-iodouracil to 5-iodo-5,6dihydrouracil was 5.9 mM"cm" in buffer A.

Preparation of DHPDHase
Purification of DHPDHase from bovine liver was based on the method of Shiotani and Weber (2). A pivotal step in this purification was the affinity purification of the enzyme with 2',5'-ADP-Sepharose. All steps were at 4 "C.
Step 1. Homogenization-Frozen bovine liver (1 kg) was quickly thawed and homogenized in 2 liters of 0.25 M sucrose in buffer B (35 mM sodium phosphate, 5 mM /3-mercaptoethanol, and 2.5 mM MgCl, at pH 7.4) with a commercial Waring blender at high speed for 5 min.
The homogenate was filtered through cheese cloth and the filtrate was centrifuged for 20 min at 10,000 X g.
Step 2. Batch DE-52-The pH of the supernatant was adjusted to 8.0 and the enzyme was absorbed onto 2.5 liters of a 60% slurry of DE-52 resin equilibrated to pH 8.0. After the suspension was stirred for 30 min, the resin was recovered by vacuum filtration and was quickly washed with 2 liters of buffer B. After washing the resin with 1500 ml of buffer B with 50 mM KCl, enzymatic activity eluted from the resin in 650 ml of buffer.
Step 3. Ammonium Sulfate Precipitation-The solution was brought to 40% saturated ammonium sulfate by addition of 158 g of solid ammonium sulfate and stirred for 30 min. The precipitate was collected by centrifugation at 10,000 X g for 20 min. The supernatant was brought to 50% saturated ammonium sulfate with an additional 46 g of solid ammonium sulfate. After stirring for 30 min, the enzyme was collected by centrifugation at 10,000 X g for 20 min. The residue was dissolved in 60 ml of buffer B and dialyzed against 2 X 4 liters of buffer B for 20 h.
Step 4. 2',5'-ADP Sepharose Column-The dialysate was clarified by centrifugation at 10,000 X g for 20 min and was applied (2 ml/ min) to a 6-ml column of 2',5'-ADP-Sepharose equilibrated with buffer B. The column was washed with buffer B until the A274 of the eluant was below 0.1. The enzyme was eluted with 0.1 mM NADPH. Fractions with enzymatic activity were pooled and then precipitated with 60% ammonium sulfate.
Step 5. P-200 Column-The precipitate was collected by centrifugation at 10,000 X g for 10 min and was dissolved in 2 ml of buffer B. The enzyme was applied to a 130-X 1.5-cm column (2 ml/h) of Bio-Rad P-200 resin equilibrated in buffer B. The enzyme eluted in fractions (2 ml) 21-23 and was stored at 5 "C as a 60% ammonium sulfate precipitate.

Determination of Enzyme-bound Flavin Concentration
DHPDHase has multiple flavins per enzymic subunit and multiple iron sulfur prosthetic groups (2, 3). The enzyme-bound flavin of bovine liver DHPDHase was released from the protein by heating for 5 min at 100 "C. The free flavin had absorbance maxima at 447,374, and 263 nm. The concentration of flavin released from the enzyme was calculated with an extinction coefficient for FMN at 450 nm of 12.5 mM"cm" (17). On the basis of the flavin released from denatured enzyme, the c~~~ (based on the released flavin) of purified enzyme was calculated to be 31 mM"cm".
2.6 units/ml of DHPDHase corresponded to 1 p~ enzyme-bound flavin. In most cases the concentration of DHPDHase was expressed in terms of units/ml. When necessary, the concentrations of DHPDHase in units/ml was converted to p~ enzyme-bound flavin by dividing units/ml of DHPDHase by 2.6.
The DHPDHase (350 units) in 2.0 ml of buffer A and 2 mM DTT was reacted for 10 min at 37 "C with 380 nmol of [6-3H]5-iodouracil (specific activity, 8 X 10' cpm/nmol). The inactivated enzyme retained less than 5% of its original activity. Free radiolabel was separated from bound radiolabel by size-exclusion chromatography on a 1.5-X 15-column of P-6 resin equilibrated in 0.05 M NH4HC03 and the radiolabeled enzyme (3.5 X lo6 cpm) was lyophilized.

Carboxymethylation and Tryptic Digestion of Radiolabeled Enzyme
The radiolabeled enzyme was dissolved in 2 ml of 0.5 M Tris-HC1 at pH 8.0 with 6 M guanidine hydrochloride and 2.7 mM EDTA (carboxymethylation buffer (18)). 2-Mercaptoethanol (11 pl) was added to the dissolved enzyme and the mixture was flushed with N, for 30 min. The anaerobic solution was boiled for 2 min and allowed to equilibrate to room temperature for 60 min at which time 0.5 ml of 0.5 M iodoacetic acid in the carboxymethylation buffer was added to the reduced protein. This solution was incubated in the dark for 20 min at room temperature. The reaction was then quenched by addition of 50 p1 of 2-mercaptoethanol. The mixture was dialyzed twice against 1 liter of distilled HzO. The protein suspension was lyophilized and then suspended in 4 ml of 0.2 M NH'HCO,. The protein was digested at 37 "C with 1.6 mg of N-tosyl-L-phenylalanine chloromethyl ketone-treated trypsin that was added in five aliquots over 60 h. Trypsin was separated from the digested enzyme by chromatography in 0.2 M NH4HC03 buffer on a 1.5-X 14-cm column of P-6 resin. Fractions with radiolabel were pooled and lyophilized (2.8 X lo6 cpm).

Separation of Tryptic Fragments by HPLC
The tryptic peptides were resolved on a Waters pBondaPak Cls column (0.46 X 25 cm) and a Waters 625 LC system (Milford, MA). The peptides were eluted from the CIS column at a flow rate of 1 ml/ min and detected by absorbance at 215 nm with a Waters 484 tunable absorbance detector. Two solvent systems were used to develop the column. (a) Solvent system 1: the peptides were dissolved in 2 ml of 0.1% trifluoroacetic acid and absorbed onto the Cle column that had been equilibrated in 0.1% trifluoroacetic acid. After washing the column for 10 min, the peptides were eluted with a linear gradient of 0-30% acetonitrile in 0.1% trifluoroacetic acid over 90 min. (b) Solvent system 2: the peptides were dissolved into 2 ml of 20 mM ammonium acetate at pH 5.5 and absorbed onto the C18 column that had also been equilibrated in the ammonium acetate buffer. The column was developed isocratically with 5% acetonitrile in the ammonium acetate buffer over 50 min.
Sequence analysis was performed on a Applied Biosystems 477A protein sequencer equipped with an on-line Applied Biosystems 120A analyzer (San Jose, CA).

Mass Spectrometry
Fast atom bombardment mass spectrometry was used to analyze 2 p1 of peptide diluted with 1 pl of glycerol/thioglycerol (1:l) matrix. A VG70-70SQ high resolution mass spectrometer was used for the analysis and ionization was produced by a cesium ion gun operating at 30 kV. Mass spectral data were collected over the mass range 350-1500 Da in the Multichannel Analysis data aquisition mode.

Data Analysis
The decelerating time courses for oxidation of NADPH or reduction of 5-iodouracil were fitted to Equation 2 as follows.
[Product] = A + B*e-k:mt + C*t (2) where k., (min") was the first order rate constant for loss of activity, B ( p~) was the amplitude of the burst in product formation and t was time (rnin). Since the concentration of product in these experiments was initially zero, the value of A was equal to the negative of the value of B.
The dependence of the rate constant for inactivation of DHPDHase on 5-iodouracil (IU) concentration ( p~) was fitted to Equation 3 as follows.
where A was extrapolated rate of inactivation (pM/min) at zero 5iodouracil concentration and B was the apparent K; (pM) for inhibition of the inactivation process by 5-iodouracil. The constants defined in these equations were determined by an iterative nonlinear least-squares fitting of the data to these equations (19).

RESULTS
Properties of DHPDHase Purified from Bovine Liver-The results from an enzyme purification are summarized in Table  I. The enzymic activity and protein in the initial homogenate was not assayed. The enzyme was purified 290-fold after the DE-52 batch step to a specific activity of 23 units/A~7~ with an 81% yield. The Coomassie Blue protein assay (Pierce Chemical Co.) was used to determine that one kt274 per ml was equivalent to 0.88 mg/ml protein with bovine serum albumin as the reference protein. The enzyme was greater than 90% homogeneous by 8% sodium dodecyl sulfate gel electrophoresis with the major band having a M, of 110,000. The optical spectrum of the enzyme had maxima at 426,376, and 274 nm with ratios of absorbance maxima (A274&7&&27) of 3.5:l.l:l.O. The FMN and FAD released from heat-denatured DHPDHase were quantitated by HPLC and found to be in the ratio of 1.3 to 1. A similar ratio was found for DHPDHase from porcine liver (3).
The K , for thymine with 100 p~ NADPH at pH 8.0 was 0.34 f 0.1 p~. This value was &fold less than the K , values for the rat liver enzyme (2) and porcine liver enzyme (3) determined at pH 7.0. The purified enzyme had an endogenous NADPH oxidizing activity in the absence of thymine that was less than 10% of the maximum activity for thymine reduction. The activity of DHPDHase stored in the presence of thiol was not increased by including DTT in the standard assay. However, when stored overnight at 5 "C in the absence of thiol, the enzyme lost as much as 80% of its activity. The activity was restored when 1 mM DTT was included in the assay buffer. Consequently, 1 mM DTT was routinely included in reaction mixtures to maintain the enzyme in the activated state.
DHPDHase normally catalyzes the reduction of thymine by NADPH (1). DHPDHase also catalyzed the reduction of thymine (measured at 260 nm) by DTT. The K, of DTT for reduction of 100 p~ thymine by DTT in buffer A was 2. Inactivation of DHPDHase by 5-Iodouracil-The rate of enzymic reduction of 5-iodouracil by NADPH decelerated in a first order process as the reaction progressed. For example, in a reaction mixture with 2 mM DTT, 100 pM NADPH, 50 p~ 5-iodouracil, and 0.14 units of DHPDHase, only 4.4 p~ 5iodouracil was reduced before the reaction ceased. A second burst of substrate reduction (3.7 p~) occurred after a second addition of enzyme (Fig. 1). In the absence of a reducing substrate the enzyme lost less than 10% of its activity during a 5-min incubation with 50 p~ 5-iodouracil.
DHPDHase was protected from inactivation by thymine. The rate constant for inactivation of the enzyme by 10 pM 5iodouracil with 100 pM NADPH and 1 mM DTT was 2 min" in the absence of thymine versus 0.4 min" in the presence of 10 p~ thymine. These initial results were consistent with a mechanism-based inactivation of DHPDHase by 5-iodouracil.
If 5-iodouracil were a mechanism-based inactivator of DHPDHase, the ratio of reduced 5-iodouracil to inactivated enzyme (i.e. the stoichiometry for inactivation) should be independent of the reaction conditions. However, the amount of 5-iodouracil reduced by DHPDHase at the end of these reactions was increased 2.4-fold as the concentration of DTT in the reaction mixture was increased from 2 to 20 mM (Fig.  1). Similarly, product formation increased 11-fold when the concentration of 5-iodouracil was increased from 5 to 150 pM (data not shown). These results suggested that the rate of release of the reactive product that inactivated the enzyme, presumably 5-iodo-5,6-dihydrouracil, was faster than the rate of inactivation of the enzyme. Consequently, the free product subsequently competed with 5-iodouracil for binding to the active site. Inactivation of the enzyme most likely occurred by covalent modification of an active site residue.
Since the reduced product should have chemical properties similar to those of iodoacetamide, it should rapidly alkylate thiolates. The reactivity of 5-iodo-5,6-dihydrouracil with DTT or glutathione in buffer A was a model for this reaction. The reaction of these thiols with 5-iodo-5,6-dihydrouracil was followed spectrophotometrically at 225 nm, which monitored the formation of iodide (20). The bimolecular rate constant for reaction of DTT with 5-iodo-5,6-dihydrouracil was 1.07 f

TABLE I Summary of the purification of DHPDHase from bovine liver
Step  The reactions were initiated with enzyme, which was activated by storage in 5 mM DTT, by a 100-fold dilution to give a final enzyme concentration of 0.08 units/ml and a DTT concentration less than 0.05 mM. The reduction of 5-iodouracil was followed at 283 nm. To one reaction mixture 5 mM DTT was added at 8 min (-) followed by 0.08 units/ml DHPDHase at 13 minutes. To the other reaction mixture (. . . . .) 0.08 units/ml DHPDHase was added at 12 min followed by 5 mM DTT at 13 min. At 13 min the total additions, but not the order of additions, to the two reactions were identical. 0.02 mM" rnin". The analogous rate constant for the reaction with glutathione was 0.20 f 0.01 mM" min". 5-Iodo-5,6-dihydrouracil was relatively stable in buffer A in the absence of thiols (data not shown). Consequently, this product was expected to accumulate in the assay medium during the enzymic reduction of 5-iodouracil in the absence of DTT. If 5-iodo-5,6-dihydrouracil was the species that inactivated the enzyme, the rate of inactivation of enzyme added initially should be slower than the rate of inactivation of enzyme added after the concentration of 5-iodo-5,6-dihydrouracil had built up. This possibility was examined by interchanging the order of addition of DTT and DHPDHase to a reaction mixture that contained enzyme that had been inactivated with 5-iodouracil in the absence of DTT (Fig. 2). When 5 mM DTT was added to this reaction mixture prior to the addition of more enzyme (this concentration of DTT would destroy 5-iodo-5,6-dihydrouracil with a tlh of 0.1 min), the newly added enzyme was fully active (Fig. 2). When enzyme was added to the reaction mixture before addition of 5 mM DTT (i.e. 5-iodo-5,6-dihydrouracil was not destroyed prior to addition of the second aliquot of enzyme), the newly added enzyme had greatly reduced activity (Fig. 2). These results demonstrated that a reactive product, presumably 5iodo-5,6-dihydrouracil, was released from the enzyme during the inactivation reaction. The free product either inactivated the enzyme or reacted with an exogenous nucleophile such as DTT.
Dependence of the Rate of Inactivation of DHPDHase on the Concentration of 5-Iodouracil-The apparent rate constant for inactivation of DHPDHase by 5-iodouracil and 100 p~ NADPH decreased from 10 min" to less than 0.3 min" as the concentration of 5-iodouracil was increased from 1 to 150 p~ (Fig. 3). The maximum rate of inactivation of the enzyme at low concentration of 5-iodouracil was estimated to be 16 f 3 min" from a fit of the data of Fig. 3 to Equation 3.
In contrast to these results, the rate constant for inactivation of an enzyme by a simple mechanism-based inactivator was expected to increase to a limiting value at high concentrations of inactivator. Since the initial velocity for reduction of 5-iodouracil by 0.14 units/ml DHPDHase decreased slightly from a value of 4.5 wM/min with 1 p M 5-iodouracil to 3.0 pM/min with 150 p M (data not shown), the inhibition of the rate of inactivation at high 5-iodouracil concentration is not likely due to substrate inhibition. These results suggest that 5-iodouracil and the inactivating species, enzymatically generated 5-iodo-5,6-dihydrouracil, were competing for a common site.
DHPDHase was also inactivated by 5-iodouracil with DTT as the reductant. The first order rate constant for inactivation of 1.8 units/ml DHPDHase was measured at five concentrations of 5-iodouracil with 2 mM DTT as the reductant. The rate constant for inactivation decreased as the concentration of 5-iodouracil was increased (data not shown). These results were analogous to those found with NADPH as the reductant. The data were fitted to Equation 3 to yield a maximum rate constant for inactivation of 0.6 f 0.2 min". The maximum rate constant for inactivation of the enzyme by 5-iodouracil in the presence of NADPH and DTT was 27-fold larger than that measured in the presence of DTT alone. reduction of 5-iodouracil did not change over this concentration range of DTT. These results provide further evidence that the reactive product of 5-iodouracil reduction, 5-iodo-5,6-dihydrouracil, partitioned between inactivation of the enzyme and reaction with DTT.
Stoichiometry for Inactivation of DHPDHase by 5-Iodouracil-The ratio of 5-iodouracil reduced to DHPDHase inactivated was approximately 30 nmol/unit (0.14 units/ml of DHPDHase was inactivated by the reduction of 4.4 PM 5iodouracil (Fig. 1)). However, interpretation of this result was complicated by the effects of DTT and 5-iodouracil on the inactivation reaction. For instance, the stoichiometry for inactivation was increased 2.4-fold by increasing the concentration of DTT from 2 to 20 mM (Fig. 1) and it was increased 11-fold by increasing the concentration of 5-iodouracil from 5 to 150 pM. To circumvent these problems, the enzymatic activity was titrated with 5-iodouracil at elevated concentrations of enzyme. In a typical experiment, aliquots of DHPDHase were reacted with different amounts of 5-iOdOuracil and the enzymatic activity was determined at the end of the inactivation reactions. Since the concentration of free 5-iodouracil in these experiments was effectively zero throughout the titration, the effects that elevated concentrations of 5-iodouracil could have on the inactivation process were minimized. Furthermore, at high enzyme concentration, 5-iodo-5,6-dihydrouracil that dissociated from the enzyme was more likely to react with enzyme than the exogenous nucleophile DTT.
DHPDHase was completely inactivated by 1.7 mol of 5iodouracil per mol of enzyme-bound flavin (Fig. 4). Titration of enzymatic activity with 5-iodouracil decreased linearly as the concentration of 5-iodouracil was increased (Fig. 4). If DTT was competing effectively with the enzyme for the 5iodo-5,6-dihydrouracil released during the course of this titration, significant deviations from linearity should have occurred toward the end of the titration. Consequently, the trapping of 5-iodouracil by DTT was minimal.
When excess DHPDHase (12 p~ enzyme-bound flavin) was reacted with 2 mM DTT and limiting [6-3H]5-iodouracil (0.7 PM), 18% of the radiolabel was bound to the protein. The bound radiolabel was not displaced by incubation with 200 p M unlabeled 5-iodouracil or by heating at 100 "C for 10 min

FIG. 4. Titration of DHPDHase by either 5-iodouracil or 5-
iodo-6,6-dihydrouracil with 2 m M DTT as the reducing substrate. 5-Iodouracil (0) or 5-iodo-5,6-dihydrouracil (0) was incubated with 14 units/ml DHPDHase for 20 min at the indicated ratios of inactivator concentration to enzyme concentration (enzyme-bound flavin). The enzymatic activity remaining was measured in the standard assay and expressed as a percentage of the activity of untreated enzyme. In the absence of 5-iodouracil the enzyme lost no activity during a 20-min incubation at 37 "C. Alternatively, when the enzyme was incubated with 20 p~ 5-iodouracil or 10 p~ 5-iodo-5,6-dihydrouracil in the absence of DTT, the enzyme lost 12 or 23% of its activity. The residual activity, which was the same in the presence and absence of thymine, was due to endogenous NADPH oxidase activity. in 0.1 N HC1. Since tritium was not released from radiolabeled 5-iodouracil during the inactivation process, the bound radiolabel provided a good estimate for the amount of inactivator bound to the enzyme. These results show that the radiolabel was covalently linked to the protein and, assuming one inactivator molecule was covalently bound per active site, the stoichiometry for inactivation (mol of inactivator consumed per mol of enzymic activity site inactivated) was 5.5.
Since the stoichiometry for inactivation was 5.5 and 1.7 mol of inactivator was consumed per mol of enzyme-flavin (Fig. 4), the ratio of enzyme-bound flavin to covalently bound 5-iodouracil was 3.2. This result suggested that either 3.2 flavins were at each catalytic site or not all the flavins were catalytically active.
Sequence were chromatographed on the same column equilibrated in 20 mM ammonium acetate and developed with solvent system 2 (chromatogram not shown). The fraction from each purification with the maximum amount of radiolabel was rechromatographed in this solvent system. Fraction 27 from peptide 1, which was purified further as described in the text, and fraction 33 from peptide 2 were sequenced. enzyme was 3.1. A tryptic digest of inactivated DHPDHase was prepared as described under "Experimental Procedures" and was purified by HPLC on a CIS column with solvent system 1. Over 70% of the radiolabel was eluted in two peaks, fractions 33-36 (49%) and fractions 37-38 (21%), respectively (Fig. 6A). The pooled fractions, peptide 1 and peptide 2, respectively, were chromatographed two times with solvent system 2.' Fig. 6B shows the elution profile of the second chromatographic step. The sequence of peptide 1 was Asn-Leu-Ser-X-Pro-His-where X was the amino acid modified by radiolabeled 5-iodouracil. The sequence for peptide 2 was Asn-Leu-Ser-X-Pro-His-Gly-Met-Gly-Glu-Arg (Table 11).
Fast Atom Bombardment Mass Spectrum of Peptide 1-The FAB mass spectrum of peptide l b shows two weak molecular ion signals. The protonated molecular ion (M + H') was observed at 782 Da together with a sodium adduct ion (M + Na+) at 804 Da which is characteristic of samples containing detectable levels of sodium. The molecular weight of 781 Da was 112 Da greater than the unmodified peptide species and was consistent with the amino acid not identified Peptide 1 could be resolved into two fragments by chromatography on a 2.1-X 100-mm Aquapore RP-300 column (Applied Biosystems, San Jose, CA). The peptides were applied to the column equilibrated in 0.06% trifluoroacetic acid. The column was washed with 2% buffer B (0.056% trifluoroacetic acid in 80% acetonitrile) for 1 min and developed with a linear gradient of 2% B to 12% B over 5 min with a flow rate of 0.3 ml/min. The respective retention times were 7.31 and 7.93 min, respectively. The amino acid composition of these peptides were identical.

Amino acid sequnece of the radiolabeled peptides isolated from a tryptic digest of DHPDHase inactivated with [6-3H]5-iodouracil
NADPH as the reductant than with 2 mM DTT. This correlation suggested that reduction of enzyme was the rate-limiting step in the reduction of thymine and 5-iodouracil.
The active site concentration of DHPDHase was initially assumed to be equal to the concentration of flavin that was released by heat denaturation of the enzyme. However, comparison of the titration data for 5-iodouracil with the fraction of 5-iodouracil bound to the enzyme indicated that there were 3.2 mol of flavin per mol of radiolabeled product covalently linked to the enzyme. This implied that there were multiple flavins per active site or some of the enzyme-bound flavin was catalytically inactive. Since the enzyme contained FAD and FMN in nearly equal amounts, it was probable that there were multiple flavins per active site. Porcine liver DHPDHase also contains an iron-sulfur center and multiple flavins per subunit (3).
DHPDHase was completely inactivated by 0.67 mol of racemic 5-iodo-5,6-dihydrouracil per mol of enzyme-bound flavin. This was equivalent to 2.1 (0.67 x 3.2) mol of racemic inactivator per mol of active site. Since DHPDHase probably catalyzed the stereospecific reduction of 5-iodouracil to 5iodo-5,6-dihydrouracil, the finding that the stoichiometry for inactivation of DHPDHase by racemic 5-iodo-5,6-dihydrouracil (2.1) was less than that for the enzymatically generated enantiomer (5.5) suggested that the nonenzymatically generated enantiomer was a more potent inactivator of the enzyme. The stoichiometry for the nonenzymatically generated enantiomer was calculated to be 1.3.3 Since the stoichiometry for inactivation of DHPDHase by the enzymatically generated enantiomer was 5.5, there was partitioning of this enantiomer between inactivation of the enzyme and conversion to a product that did not inactivate the enzyme. Possible mechanisms for deactivation of 5-iodo-5,6-dihydrouracil are: 1) the reductive dehalogenation to 5,6-dihydrouracil that is analogous to the reductive debromination of 6-(bromo-methy1)purine to 6 methylpurine catalyzed by xanthine oxidase (22) or the reductive deiodination of 5-iodouracil catalyzed by thymidylate synthetase (23), and 2) the elimination of HI to form uracil (24). Preliminary studies have shown that uracil and dihydrouracil were products of the reaction of racemic 5-iodo-5,6-dihydrouracil with DHPDHase. 4 UV difference spectra data (Fig. 5) indicate that the enzymatically generated isomer of 5-iodo-5,6-dihydrouracil and the chemically synthesized racemic inactivator modified the same active site residue. This result suggests that the enantiomers either bind as mirror images or they bind similarly and displacement of the iodo group occurred from different sides of the molecule. Since high concentration of DTT or 5iodouracil afforded nearly complete protection against inactivation by 5-iodouracil, it appeared that dissociation of the enzymatically generated enantiomer of 5-iodo-5,6-dihydrouracil might be required for inactivation. If this was the case, rebinding of this isomer to DHPDHase might have correctly positioned the iodo group for nucleophilic attack.
D. J. T. Porter, unpublished observations. resulted in covalent binding of radiolabel to inactivated enzyme. The amino acid sequence of the smaller tryptic fragment was Asn-Leu-Ser-X-Pro-His-in which X was the radiolabeled amino acid. Data from fast atom bombardment mass spectroscopic analysis of this tryptic fragment were consistent with the unknown amino acid residue, X, being S-(hexahydro-2,4-dioxo-5-pyrimidinyl)cysteinyl residue. Since the pyrimidine ring was reduced in this adduct, it was reasonable to suppose that enzymatic reduction of the pyrimidine ring preceded displacement of the iodo group and subsequent inactivation of the enzyme (Equation 1).
In conclusion, 5-iodouracil was an effective inactivator of DHPDHase at low concentrations. Thus, the elevated plasma levels of uracil and thymine that occurred during infusion of 5-iodo-2'-deoxyuridine into human subjects (13) could have resulted, at least in part, from inactivation of the enzyme by enzymatically generated 5-iodo-5,6-dihydrouracil.