Identification of specific carboxyl groups on uracil-DNA glycosylase inhibitor protein that are required for activity.

The bacteriophage PBS2 uracil-DNA glycosylase inhibitor (Ugi) protein inactivates uracil-DNA glycosylase (Ung) by forming an exceptionally stable protein-protein complex in which Ugi mimics electronegative and structural features of duplex DNA (Beger, R. D., Balasubramanian, S., Bennett, S. E., Mosbaugh, D. W., and Bolton, P. H. (1995) J. Biol. Chem. 270, 16840-16847; Mol, C. D., Arvai, A. S., Sanderson, R. J., Slupphaug, G., Kavli, B., Krokan, H. E., Mosbaugh, D. W., and Tainer, J. A. (1995) Cell 82, 701-708). The role of specific carboxylic amino acid residues in forming the Ung·Ugi complex was investigated using selective chemical modification techniques. Ugi treated with carbodiimide and glycine ethyl ester produced five discrete protein species (forms I-V) that were purified and characterized. Analysis by mass spectrometry revealed that Ugi form I escaped protein modification, and forms II-V showed increasing incremental amounts of acyl-glycine ethyl ester adduction. Ugi forms II-V retained their ability to form a Ung·Ugi complex but exhibited a reduced ability to inactivate Escherichia coli Ung, directly reflecting the extent of modification. Competition experiments using modified forms II-V with unmodified Ugi as a competitor protein revealed that unmodified Ugi preferentially formed complex. Furthermore, unmodified Ugi and poly(U) were capable of displacing forms II-V from a preformed Ung·Ugi complex but were unable to displace Ugi form I. The primary sites of acyl-glycine ethyl ester adduction were located in the α2-helix of Ugi at Glu-28 and Glu-31. We infer that these two negatively charged amino acids play an important role in mediating a conformational change in Ugi that precipitates the essentially irreversible Ung/Ugi interaction.

Uracil-DNA glycosylase initiates the uracil excision DNA repair pathway by removing uracil residues that accumulate in cellular DNA following dUMP incorporation or deamination of cytosine (1). The enzyme cleaves the N-glycosylic bond linking uracil to the deoxyribose phosphate backbone of DNA to produce free uracil and an apyrimidinic site (2). Uracil-DNA glycosylase is a ubiquitous and highly conserved enzyme that shares ϳ56% amino acid sequence homology between the proteins isolated from Escherichia coli and humans (3). Molecular modeling studies using the 1.9-Å crystal structure of human uracil-DNA glycosylase revealed that uracil residues in duplex DNA must enter the catalytic pocket of the enzyme through an extrahelical rotation, termed "base flipping" (4,5). Following uracil release, successive uracil residues may be located by a processive search mechanism (6,7). The uracil excision DNA repair pathway of E. coli (8,9), bovines (10), and humans (8) involves the coordinated action of uracil-DNA glycosylase, apurinic/apyrimidinic endonuclease, deoxyribophosphodiesterase, DNA polymerase, and DNA ligase to produce a one-nucleotide repair patch.
Unlike most organisms that exclude uracil residues from DNA, the Bacillus subtilis bacteriophages PBS1 and PBS2 naturally exhibit ϳ72% base composition of U:A base pairs (11). These bacteriophage achieve stable incorporation of dUMP as a consequence of the ugi 1 gene product, and both related phage share an identical ugi nucleotide sequence (12,13). In vivo, the ugi gene is expressed immediately upon phage infection of the host and inactivates uracil-DNA glycosylase within 4 min postinfection, thus blocking the uracil excision DNA repair pathway (14).
The PBS2 Ugi protein has been purified to apparent homogeneity and extensively characterized as a small (9,474-dalton), monomeric, heat-stable, acidic protein of 84 amino acid residues (15,16). The inhibitor protein contains 12 Glu and 6 Asp residues, which help to establish an acidic isoelectric point (pI ϭ 4.2) (16). In vitro studies indicate that Ugi inactivates uracil-DNA glycosylases isolated from a wide variety of biological sources (12,17,18). E. coli Ung forms an extremely stable Ung⅐Ugi complex with 1:1 stoichiometry, which is essentially irreversible under physiological conditions (16,19). Association of Ugi with Ung involves a two-step kinetic mechanism (19). The first step is initiated by an Ung/Ugi interaction to form a precomplex, distinguished by the dissociation constant K d ϭ 1.3 M. The second step results in the formation of the final Ung⅐Ugi complex characterized by the rate constant k ϭ 195 s Ϫ1 . Thus, complex formation involves a preliminary "docking" step followed by a "locking" reaction through which the two proteins achieve optimal alignment and become very tightly bound. Inhibition of Ung occurs, since the Ung⅐Ugi complex fails to recognize the DNA substrate (16,19). Several observations indicate that Ugi binds to Ung at or near the DNA binding site. (i) Photochemical cross-linking of Ung to single-stranded oligonucleotide dT 20 , at the DNA-binding pocket, prevented Ugi binding to Ung (20). (ii) Inclusion of Ugi in UV-induced cross-linking reactions prevented formation of the Ung ϫ dT 20 cross-link (20). (iii) X-ray crystallographic studies of Ugi com-plexed to the human (4,21) and herpes simplex virus type I (22) uracil-DNA glycosylases revealed interactions between Ugi and conserved amino acids near the active site of the enzyme, suggesting protein mimicry of DNA.
Several unique features of Ugi suggest that the negatively charged amino acids of the Ugi protein play an important role in mediating the Ung/Ugi interaction. First, the overall electronegative potential of the Ugi protein, engendered by its high (21%) Glu and Asp content, is quite unusual (12,16). Second, the NMR structure ( Fig. 1) shows that 11 of the 18 Glu and Asp residues terminate the ␤-strands or are located in the more flexible loops, in a nonrandom distribution (23). Third, the tertiary structure brings together seven of these residues (Glu-20, Asp-48, Glu-49, Asp-52, Glu-53, Asp-74, Glu-78) into close proximity on one surface of the Ugi protein, forming a unique structural element (24). The negative electrostatic potential of this region is Ͼ6.6 kcal, similar to that generated by the phosphate backbone of DNA (24). Fourth, both the ␣1 and ␣2 helices contain regions with similar high electronegative potential (24). Although reasons to suspect a functional role for these residues exist, the specific involvement of these negatively charged amino acids in Ung⅐Ugi complex formation remains to be elucidated.
In the present report, we chemically modify the carboxylic acid residues of Ugi to determine their role in forming the Ung⅐Ugi complex. Specifically, the approach involved (i) chemical modification of Ugi with the water-soluble carbodiimide, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and the nucleophile glycine ethyl ester (GEE); (ii) purification of five modified forms of Ugi; (iii) characterization of each form concerning the effect of modification on Ugi specific activity, Ung⅐Ugi complex stability, and reversibility; and (iv) identification of specific sites of Ugi adduction using matrix-assisted laser desorption-ionization (MALDI) mass spectrometry and amino acid sequencing techniques. The findings establish the importance of two Glu residues in the ␣2-helix for achieving stable Ung⅐Ugi complex formation.
following modifications: (i) E. coli JM105 cells transformed with pZWtac1 were grown at 37°C in 9 liters of TYN-ampicillin medium; (ii) following Ugi overproduction, harvested bacteria were resuspended in 360 ml of TED buffer and lysed in a French pressure cell as described above; (iii) DEAE-cellulose chromatography (fraction II) was performed using a 19.6-cm 2 ϫ 10.2-cm column, and the inhibitor was eluted with a 1,100-ml linear gradient from 0 to 650 mM NaCl in TED buffer; (iv) a final concentration step was unnecessary, and Ugi (fraction IV) was stored in buffer EB containing 50 mM Tris-HCl (pH 8.0), 1 mM EDTA, 1 mM dithiothreitol, 10% (w/v) glycerol, and 100 mM NaCl.
EDC/GEE Modification of [ 35 S]Ugi-An EDC/GEE modification reaction mixture (16.9 ml) containing 1,050 nmol of [ 35 S]Ugi (fraction IV), 228 mol of GEE, and 45.6 mol of EDC in 50 mM potassium phosphate buffer (pH 6.0) was incubated at 25°C for 30 min. GEE was introduced into the reaction mixture with [ 35 S]Ugi prior to the addition of EDC. The reaction was terminated by adding 16.9 ml of 2 M sodium acetate (pH 4.75). Excess EDC and GEE were removed from EDC/GEE-modified [ 35 S]Ugi by processing 250-l aliquots through 1.4-ml spun columns containing P-4 resin (Bio-Rad) equilibrated in 50 mM potassium phosphate buffer (pH 6.0) (26). After processing, all aliquots were pooled and dialyzed extensively against buffer A (50 mM Tris-HCl (pH 7.0), 1 mM EDTA, 1 mM dithiothreitol) containing 50 mM NaCl.
Purification of EDC/GEE-modified [ 35  S]Ugi sample was loaded onto a DE-52 cellulose column (0.8 cm 2 ϫ 1.9 cm) to resolve the complex from its component proteins. The column was equilibrated in DAB buffer containing 50 mM NaCl. After sample application, the column was shut off for 5 min and then washed with 30 ml of equilibration buffer. Proteins were eluted with a two-step gradient consisting of 25 and 10 ml of DAB buffer containing 150 and 250 mM NaCl, respectively. Fractions were collected, and samples were analyzed for Ung activity and 35 S radioactivity. Fractions containing Ung⅐[ 35 S]Ugi complex were pooled and concentrated using a Centriplus-10 (Amicon) concentrator.
Enzyme Assays-Catalytic activity of uracil-DNA glycosylase and fluorescein 5-isothiocyanate-labeled Ung were measured under previously described standard reaction conditions (16). One unit of uracil-DNA glycosylase is defined as the amount that releases 1 nmol of uracil/h under standard conditions. Uracil-DNA glycosylase inhibitor activity was measured under previously described conditions (16). One unit of inhibitor inactivates one unit of uracil-DNA glycosylase when 0.05-0.15 units of endogenous enzyme were introduced into the assay. Inhibitor activity was measured in reactions where no more than 80% of the glycosylase activity was inhibited.
Electrophoresis-Nondenaturing polyacrylamide slab gel electrophoresis was performed by a modification of that described by Laemmli (27). Slab gels contained a resolving gel (13 cm) composed of 18% acrylamide, 0.48% N,NЈ-methylenebis(acrylamide), and 375 mM Tris-HCl (pH 8.8) and a stacking gel (1 cm) containing 3% acrylamide, 0.08% N,NЈ-methylenebis(acrylamide), and 125 mM Tris-HCl (pH 6.8). Protein samples (30 -60 l) were adjusted to final concentrations of 50 mM Tris-HCl (pH 6.8), 10% (w/v) glycerol, and 0.01% bromphenol blue before being loaded onto the gel. Electrophoresis was carried out at 4°C and 100 V until the tracking dye migrated through the stacking gel, at which point the potential was increased to 200 V. Electrophoresis continued until the tracking dye migrated within ϳ2 cm from the bottom of the gel. Protein bands were detected by rapid protein staining as described by Reisner (28). Briefly, slab gels were placed in dye solution containing 0.04% (w/v) Coomassie Brilliant Blue G-250 and 3.5% (v/v) HClO 4 immediately after electrophoresis. Complete staining required gentle agitation at 25°C for 12 h. Gels were destained with 5% (v/v) acetic acid, and the gel was photographed with Polaroid-55 film using an orange filter (Eastman Kodak Co.; 23A).
Nondenaturing polyacrylamide tube gel electrophoresis was performed as described by Davis (29) with some modifications. Tube gels (0.6-cm diameter) contained a resolving gel (9 cm) and stacking gel (1 cm). Protein samples were prepared, and electrophoresis was performed at 4°C as described above. To detect [ 35 S]Ugi and Ung⅐[ 35 S]Ugi, gels were sliced horizontally into 3.1-mm sections, placed into scintillation vials (7 ml), and dehydrated overnight at room temperature. Following the addition of 30% H 2 O 2 (500 l), vials were capped and incubated at 55°C for 36 -48 h. Once solubilized, 5 ml of Formula 989 (DuPont) fluor was added, and 35 S radioactivity was measured in a Beckman LS 6800 liquid scintillation counter.
Fluorescein 5-Isothiocyanate Labeling of Ung-Ung (fraction V) was extensively dialyzed against buffer KEG containing 50 mM potassium phosphate (pH 9.5), 1 mM EDTA, and 5% (w/v) glycerol at 4°C. Fluorescein 5-isothiocyanate stock (810 M) was freshly prepared in dimethylformamide, and 1.5 ml was added to 14 ml of Ung (2.9 mg) with thorough mixing. After 2.5 h at 25°C in the dark, the mixture was placed on ice and then loaded onto a Sephadex G-25 column (4.9 cm 2 ϫ 13.5 cm) equilibrated in buffer KEG. Fractions (2 ml) were collected, and samples were monitored at 260, 280, and 496 nm for absorbance. After the FITC-Ung (F-Ung) containing fractions were pooled, the concentration of F-Ung was determined based on absorbance at 280 and 496 nm. The concentration of FITC in the dye-protein conjugate (2.3 fluorophores/enzyme molecule) was determined by the absorbance, where ⑀ 496 nm ϭ 7.8 ϫ 10 4 liters/mol⅐cm (30). It was experimentally determined that 1 M FITC contributed 0.025 A 280 . Thus, the concentration of F-Ung was determined by subtracting the A 280 contributed by FITC from the overall A 280 of the dye-protein conjugate and dividing by the molar extinction coefficient of Ung. No significant alteration of Ung activity was detected following production of F-Ung (19).
Fluorescence Measurements-Steady-state fluorescence measurements were conducted at 25°C using an LS50 luminescence spectrometer (Perkin-Elmer) equipped with a xenon flash tube and a thermostatted minicell (4 mm). Excitation and emission wavelengths were 496 and 520 nm, respectively; both slit widths were set at 5 mm.
Cyanogen After incubation at 25°C for 24 h, samples were evaporated to dryness, resuspended in 65 l of distilled water, and analyzed using MALDI mass spectrometric methods. Samples utilized for peptide C3 purification were resuspended in 600 l of buffer A and applied to individual DE-52 cellulose columns (0.79 cm 2 ϫ 1.3 cm) equilibrated in buffer A at 4°C. After sample loading, the flow rate was stopped for 5 min, and then the column was washed with 2.5 ml of buffer A. Peptide fragments were eluted with a 50-ml linear gradient of 0 -250 mM NaCl in buffer A. Fractions (500 l) were collected in Eppendorf tubes, and samples (1 l) were analyzed for CNBr-cleaved peptide fragments by MALDI mass spectrometry. Fractions containing peptide C3 were pooled and concentrated to ϳ300 l using a Centricon-3 (Amicon) concentrator, buffer was exchanged with distilled water, and fractions were submitted for amino acid sequence analysis.
Mass Spectrometry-MALDI mass spectrometry was performed by the Mass Spectrometry Facilities and Service Core Unit (Environmental Health Science Center, Oregon State University) using a custombuilt time of flight instrument (31). Two different matrices were used in the sample preparation: (i) 10 mg/ml of sinapinic acid dissolved in 33% acetonitrile and 67% trifluoroacetic acid (0.1% solution) was mixed (3:1) with samples (0.5 l) containing 10 M modified or unmodified Ugi protein; and (ii) a saturated solution of ␣-cyano-4-hydroxycinnamic acid in 33% acetonitrile and 67% trifluoroacetic acid (0.1% solution) was mixed (10:1) with samples (0.5 l) containing CNBr-treated Ugi, purified peptide C3, and Asp-N-digested Ugi protein. Each matrix solution was applied to a mass spectrometric probe and allowed to dry as described previously by Bennett et al. (20). A mass spectrum was generated from 30 individual laser pulses (ϩ24 kV), and the summed signals were calibrated using standard ion signals from the matrix.
Amino Acid Sequencing-Samples (30 l in H 2 O) containing purified peptide C3 from Ugi or modified Ugi forms I-IV were applied to a cartridge filter precycled with Biobrene (Applied Biosystems). Amino acid sequencing was conducted using an Applied Biosystems model 475A gas phase protein sequencer by the Center for Gene Research and Biotechnology (Oregon State University).

Modification of Carboxylic Acid Residues in Uracil-DNA Glycosylase Inhibitor
Protein-To assess the importance of Glu and Asp residues in mediating the Ung/Ugi interaction, chemical modification of Ugi was conducted, and the effect on inhibitor activity was investigated. The water-soluble carbodiimide, EDC, was used in a two-stage modification reaction with GEE to selectively adduct carboxyl groups (32). Under mild reaction conditions, this carbodiimide preferentially forms an O-acylisourea-activated carboxyl group that subsequently undergoes nucleophilic attack by GEE, forming a terminal acyl-glycine ethyl ester reaction product (32). A limited modification reaction was conducted, and modified forms of [ 35 S]Ugi were resolved by DEAE-cellulose chromatography (Fig. 2). Five distinct forms (I-V) were identified eluting at about 180, 165, 150, 140, and 100 mM NaCl, respectively.
Identification and Purity of Ugi Forms I-V-The five [ 35 S]Ugi peaks were separately pooled, and each Ugi form was analyzed by nondenaturing polyacrylamide gel electrophoresis (Fig. 3, lanes 4 -8). Samples of unmodified [ 35 8). Forms II-V showed progressively reduced electrophoretic mobility that inversely correlated with their order of elution from the DEAE-cellulose column (Fig. 2). Taken together, these results indicate that EDC/GEE-mediated modification of Ugi (forms II-V) caused negative charge neutralization of carboxyl groups.
The purity of [ 35 S]Ugi forms I-V was assessed based on the relative Coomassie Brilliant Blue staining band intensities of individual fractions from across the DEAE-cellulose column (data not shown) and from the radioactivity of [ 35 S]Ugi de-tected in gel slices of forms I-V (Fig. 3). These results indicated that forms I-III are Ն90% pure, whereas forms IV and V appear to contain Ն65% of the corresponding form. From the 35 S radioactivity detected in gel slices of forms I-V (Fig. 3, lane  2), it was determined that forms I-V constituted approximately 14, 23, 26, 18, and 10% of the total [ 35 S]Ugi, respectively.
To determine the extent of EDC/GEE-mediated modification, each Ugi form along with unmodified Ugi was analyzed by MALDI mass spectrometry (Table I). As previously observed (16), two mass peaks were obtained for the unmodified Ugi protein; peak I (9,475 daltons) closely agreed with the predicted mass of the complete amino acid sequence of Ugi, and peak II (9,341 daltons) corresponded with Ugi protein minus the N-  S]Ugi was then applied to a DE-52 cellulose column equilibrated in the same buffer. The column was washed and eluted with a linear gradient of 50 -300 mM NaCl in buffer A, fractions (1 ml) were collected, and samples were monitored for conductivity (f) and 35 S radioactivity (q). Fractions were pooled corresponding to the various forms (I-V), as indicated by brackets, and evaporated to dryness; each pool was resuspended in buffer A (1 ml) and dialyzed against buffer EB.
terminal methionine. Ugi form I also contained two species with masses nearly identical to those of the unmodified Ugi control and apparently escaped modification during the EDC/ GEE reaction. Ugi form II appeared to contain a single acylglycine ethyl ester modification, since this adduct would be expected to add 85 daltons per modified carboxyl group. By dividing the mass increase observed for each modified Ugi form by 85 daltons, it was deduced that forms II-V contained approximately 1.1, 2.0, 2.9, and 3.8 adducts, respectively, per Ugi protein.
Effect of EDC/GEE Modification on the Activity of Ugi Forms I-V-The effect of chemical modification on inhibitor activity of purified Ugi forms I-V was determined using E. coli uracil-DNA glycosylase (Fig. 3). The specific activity of unmodified Ugi was essentially the same as either the mock EDC/GEEmodified Ugi or Ugi form I. The slight increase in specific activity observed for the modified protein sample may have resulted from the removal of inactivated Ugi contained in the original preparation, during DEAE-cellulose chromatography. In contrast, the specific activity of Ugi forms II-V displayed progressively decreased levels (58 -17%) of inhibitor activity coinciding with the increased extent of Ugi modification.
Ability of Ugi Forms I-V to Bind Ung-Ugi forms I-V were incubated with a 3-fold molar excess of Ung under conditions that typically promote Ung⅐Ugi complex formation. The complex was then resolved from its individual components by nondenaturing polyacrylamide gel electrophoresis. As controls, unmodified [ 35 S]Ugi, EDC/GEE-modified [ 35 S]Ugi reaction mixture, and Ung were individually separated by electrophoresis (Fig. 4, lanes 1-3). The addition of unmodified Ugi or EDC/ GEE-modified Ugi reaction mixture to excess Ung resulted in Ung⅐[ 35 S]Ugi complex formation of 99 and 75% of [ 35 S]Ugi, respectively (Fig. 4, lanes 4 and 5). Interestingly, the Ung⅐Ugi complex formed by Ugi from the EDC/GEE reaction mixture resulted in a series of bands with decreased mobility (lane 5). When individual Ugi forms I-V were analyzed, the uncomplexed [ 35 (Fig. 6C). In contrast, free [ 35 S]Ugi was observed for Ugi forms III-V in complex with Ung, as demonstrated by 17, 34, and 30% of the total [ 35 S]Ugi migrating as a smear of uncomplexed inhibitor protein, respectively. A diffuse band of free Ung also appeared in these samples. These results were consistent with an interpretation that more extensive levels of EDC/GEE modification of Ugi cause less stable complexes to dissociate during electrophoresis.
To determine if modified forms I-V of [ 35 S]Ugi in complex could exchange with free Ugi, each complex preparation was incubated with a 3-and 30-fold molar excess of unmodified Ugi, and gel electrophoresis was performed, as before, to resolve the constitutive components (Fig. 6, D and E). As anticipated for the control Ung⅐[ 35 S]Ugi, no significant release of unmodified [ 35 S]Ugi (Ͻ2.5%) was detected, indicating the irreversibility of the complex. A nearly identical result was obtained for Ung⅐[ 35 S]Ugi form I complex. In contrast, the presence of a 3-fold molar excess of Ugi promoted the release of 27, 50, 63, and 54% of total modified Ugi forms II-V from the complex, respectively (Fig. 6D). Similar increases in the amount of [ 35 S]Ugi released were observed for the exchange reactions containing the 30-fold molar excess of Ugi (Fig. 6E). These results further suggest that charge neutralization due to modification leads to the destabilization of the irreversible Ung⅐Ugi complex.
If modification of Ugi facilitated Ung⅐Ugi dissociation in the presence of exogenous Ugi, then perhaps these complexes also undergo dissociation in the absence of competing Ugi. Free Ung produced from a dissociated complex might be detected by measuring uracil-DNA glycosylase activity. Therefore, the enzymatic activity of Ung in each complex containing unmodified or modified Ugi forms I-V was determined under standard assay conditions. The preformed complex containing unmodified Ugi showed Ͻ0.1% of the total uracil-DNA glycosylase in complex displaying catalytic activity. In contrast, Ung com-plexed with Ugi forms I-V, respectively, possessed 2, 18, 46, 37, and 28% of the uracil-DNA glycosylase activity expected for completely uncomplexed and uninhibited Ung. Taken together, these results confirm that EDC/GEE-modified forms of Ugi are less capable of maintaining an irreversible and catalytically inactive Ung⅐Ugi complex.

Steady-state Fluorescence Measurements of Fluorescein 5-Isothiocyanate-labeled Ung Binding to Ugi Forms I-V-To
further investigate the properties of EDC/GEE-modified Ugi interactions with Ung, the enzyme was labeled with FITC to produce F-Ung. As previously reported (19), F-Ung fluorophores, when quenched, function as reporter groups for both Ugi and nucleic acid binding. To determine whether fluorescence quenching was quantitative between differentially modified forms of Ugi, F-Ung was titrated with Ugi, and the relative fluorescence was monitored under steady-state conditions (Fig. 7). In the control, the addition of unmodified Ugi caused a proportional decrease in fluorescent signal and elicited a 9.4% maximal quench achieved at a Ugi:Ung molar ratio of 0.8:1. Titration of F-Ung with modified Ugi forms I-V also displayed a linear decrease in fluorescence intensity with maximum flu- orescence quench occurring at 9.2, 7.7, 5.4, 4.9, and 5.4%, respectively (Fig. 7). Saturation of F-Ung occurred between Ugi:Ung molar ratios of 0.8:1 and 1.3:1 for each modified Ugi form I-V. These results are consistent with an inhibitor:enzyme stoichiometry of 1:1. We interpret the reduced levels of relative fluorescence quench for Ugi forms II-V to indicate that EDC/GEE modification either alters amino acid residues that directly interact with F-Ung fluorophores or perturbs the local structure and environment around the fluorophores in the final F-Ung⅐Ugi complex.
Effect of Ugi Forms I-V on F-Ung Binding to Poly(U)-The linearity of the titration curves for Ugi forms I-V binding to F-Ung (Fig. 7) indicate that Ugi remains in complex with uracil-DNA glycosylase after complex formation and does not freely dissociate. This observation would appear to contradict findings showing that free Ugi will exchange with modified forms II-V in the preformed complex and that Ung demonstrates catalytic activity in these same complexes. However, an explanation for both results could be that complexes are maintained in solution but that the addition of competitor Ugi or nucleic acid present in the standard reaction mixture promotes dissociation of modified complexes. Thus, the influence of Ugi on F-Ung binding to poly(U) was investigated. When F-Ung (42 pmol) was combined with a saturating amount of poly(U), the average fluorescence intensity was quenched by 7.9 Ϯ 0.8% (Fig. 8A, ⌬RF 1 ). The addition of unmodified [ 35 S]Ugi (63 pmol) resulted in a further fluorescence intensity decrease of 4.5% (Fig. 8A, ⌬RF 2 ), indicating that Ugi preferentially binds to F-Ung in the presence of poly(U) and effectively competes poly(U) out of the complex. Similar experiments were conducted using modified Ugi forms I-V in place of the unmodified Ugi. After the initial fluorescence quench (⌬RF 1 ϳ8%) due to poly(U), the addition of Ugi forms I-V resulted in ⌬RF 2 values of 4.3, 3.1, 1.4, 1.2, and 2.1%, respectively (Fig. 8B). Each of the EDC/GEE-modified Ugi forms, except Ugi form I, showed a significantly reduced ⌬RF 2 value compared with the unmodified Ugi control (⌬RF 2 ϭ 4.5%).
As a second approach, an analogous experiment was con-ducted where F-Ung (42 pmol) was combined with either unmodified or modified Ugi forms I-V (63 pmol) prior to poly(U) exposure, and the relative fluorescence intensity was monitored after each addition to determine ⌬RF 1 and ⌬RF 2 (Fig.  8C). The initial addition of unmodified Ugi caused a 8.5% reduction of fluorescence intensity, and no further quenching of the fluorescent signal was observed following the poly(U) addition (⌬RF 2 Ͻ0.17). This observation was consistent with an interpretation that the Ung⅐Ugi complex was refractory to binding nucleic acid (19). Conversely, the addition of poly(U) to each of the EDC/GEE-modified forms of Ugi in complex with FIG. 7. Effect of EDC/GEE-modified Ugi binding to fluoresceinconjugated Ung on fluorescence intensity. Samples (400 l) containing 110 nM F-Ung were equilibrated at 25°C for 5 min and placed into a thermostatted quartz minicell cuvette, and fluorescence was measured (496-nm excitation and 520-nm emission wavelengths) for 1 min (5-s acquisition time) to establish the F-Ung signal corresponding to 100% relative fluorescence as described under "Experimental Procedures." Following this equilibration period, additions (40 l) containing 0, 138, 275, 413, 550, 688, 825, 963, 1,100, 1,238, 1,375, 5,500, and 11,000 nM each of unmodified [ 35 S]Ugi (q), form I (E), II (f), III (Ⅺ), IV (å), and V (Ç) were added to F-Ung, and the fluorescence intensity was monitored for an additional 5.5 min. Measurements taken at 15-s intervals were used to determine the average relative fluorescence for each Ugi concentration. After correcting for dilution effects, the percentage decrease in relative fluorescence intensity was calculated as the average relative fluorescence following the Ugi addition subtracted from the 100% relative fluorescence of F-Ung alone. Each analysis was carried out in duplicate, and the average values were plotted.

FIG. 8. Effect of Ugi forms I-V on F-Ung binding to poly(U).
A set of six samples (400 l) containing 105 nM F-Ung were placed into a thermostatted quartz minicell cuvette and equilibrated at 25°C, and fluorescence intensity measurements were recorded as described in the legend to Fig. 7. A, to one sample, an addition (30 l) containing 4.4 mg/ml poly(U) was made after 1 min (arrow), and a second addition (30 l) containing 2,100 nM of [ 35 S]Ugi (form I) was added after 6 min (arrow). Fluorescence measurements were recorded at 5-s intervals; however, only those taken every 15 s have been plotted for clarity. ⌬RF 1 represents the net decrease in fluorescence intensity caused by the first addition, and ⌬RF 2 is the decrease due to the second addition. All net decreases are plotted as a percentage relative to the F-Ung fluorescence control after correcting for dilution effects. B, poly(U) was added to each of the six F-Ung samples (first addition), and then either 2,100 nM unmodified Ugi or 2,100 nM forms I-V was added (second addition) as indicated above. Quenching of F-Ung fluorescence was measured, and ⌬RF 1 was determined for the poly(U) addition (f) and ⌬RF 2 for the Ugi addition (3). C, unmodified or modified forms (I-V) of Ugi were added to a second set of F-Ung samples (first addition), and then poly(U) was added (second addition). Fluorescence quenching of F-Ung was similarly determined, and ⌬RF 1 for Ugi (f) and ⌬RF 2 for poly(U) (3) were plotted.
F-Ung elicited additional fluorescence quenching (Fig. 8C, ⌬RF 2 ). Collectively, these results demonstrate the reversibility of modified Ugi forms in complex with Ung when nucleic acid is present, a property not exhibited by the unmodified Ugi⅐Ung complex.
Identification of EDC/GEE-modified Amino Acid Residues in Ugi-In order to locate EDC/GEE-modified residues of Ugi forms I-V, the inhibitor protein was chemically cleaved with cyanogen bromide, and the peptide fragments were analyzed by MALDI mass spectrometry (Figs. 9 and 10). Ugi contains four methionine residues capable of producing five peptide fragments following complete CNBr-induced cleavage (Fig. 9). Four peptide fragments were identified for unmodified Ugi; 2 mass values were in excellent agreement with the predicted mass of peptides C1/2, C2, C3, and C4 (Table II). A similar analysis was performed on EDC/GEE-modified Ugi forms I-V ( Fig. 10 and Table II). As anticipated, Ugi form I generated a set of four peptides with nearly the same mass values as the unmodified Ugi control. In contrast, five peptide fragments were identified for CNBr-treated Ugi form II; four corresponded to unmodified peptides C1/2, C2, C3, and C4. The fifth peptide (C3M 1 ), comprising 82% of C3-derived peptide fragments, had a mass of 3,643 daltons, which was 89 daltons larger than that predicted for peptide C3. These findings indicate that the major site of adduction is located on the C3 peptide. Similarly, the vast majority (92%) of Ugi form III adducts were localized on the C3 peptide (Table II). A second modified peptide (C3M 2 ) was identified with a mass of 3,710 daltons, indicative of two adducts, each of ϳ85 daltons. Ugi form III also produced a C3M 1 peptide fragment (3,638 daltons), connoting a single acyl-glycine ethyl adduction (Table II). Mass determination of CNBr-treated Ugi forms IV and V both included peptide fragments C3M 1 and C3M 2 and possessed additional peptides C3M 3 and C3M 4 . The relative percentage of peptides C3M 1-4 detected for each Ugi form is indicated in Table II. In all cases, the primary modification site(s) were specifically localized to peptide C3, which contained 10 Glu and Asp residues.
The location of the adducted site(s) on peptide C3 was fur-ther investigated using the same approach but following endoproteinase Asp-N digestion of Ugi. A comparison of the molecular weights of unmodified and form II Ugi peptides revealed two modified fragments derived from peptides A2 and A6/7 (Table III). The major A2M 1 (ϳ82%) and minor A6/7M 1 (ϳ23%) modified peptide species both showed a mass increase of ϳ85 daltons, denoting the presence of a single acyl-glycine ethyl ester adduct. Peptide A2M 1 was detected in equal abundance with C3M 1 and shared overlapping amino acid sequence from Leu-25 to Ser-39 (Fig. 9). Thus, we deduced that modification by EDC/GEE likely involved Glu-27, Glu-28, Glu-30, Glu-31, and/or Glu-38. The appearance of peptide A6/7M 1 suggested that another site of adduction occurred with reduced frequency at Glu-64, Asp-74, Glu-78, and/or the carboxyl terminus of the Ugi protein. Unfortunately, the latter observation was not confirmed by detecting a modified peptide corresponding to pep-2 CNBr peptide fragments C1 and C5 correspond to the single amino acids, methionine and leucine, that were not detected by MALDI mass spectrometry. [ 35 S]Ugi forms I-V were treated with CNBr, and peptide fragments were produced and analyzed by mass spectrometry as described under "Experimental Procedures." A, mass spectra of Ugi form I peptide fragments showed the relative intensity of singly charged ions corresponding to mass peaks of peptides C1/2, C2, C3, and C4 (Fig. 9). Peptide C1/2 represents the C2 fragment containing an uncleaved amino-terminal methionine residue, and individual fragments C1 and C5 were not detected. Mass spectra on CNBr-generated peptides from Ugi forms II (B), III (C), IV (D), and V (E) are shown, focusing on the mass peaks between 3,500 and 4,000 daltons. This region contained unmodified peptide C3 and four modified peptides, M 1 to M 4 . The other mass peaks corresponding to peptides C1/2, C2, and C4 remained essentially unchanged for forms I-V. tide C4 (Table II). We assume that the low abundance of this modified species and interference by minor mass peaks in the C4 region obscured detection following CNBr cleavage of Ugi form II. Ugi form III digested with endoproteinase Asp-N also produced peptides A2M 1 and A6/7M 1 ; however, an additional modified fragment (A2M 2 ) was detected with a mass increase of 162 daltons (Table III). As observed for peptide C3M 2 , the A2M 2 peptide most likely contained two acyl-glycine ethyl ester adducts among the five glutamic acid residues located between Leu-25 and Ser-39.
In order to identify the precise sites of adduction, modified and unmodified peptide C3 were purified to apparent homogeneity by DEAE-cellulose chromatography (data not shown). Amino acid sequence analysis unambiguously identified peptide C3 from CNBr-treated Ugi forms I-IV. Examination of the amino acid PTH-derivatives detected in sequencing cycles for the C3 peptide of Ugi form II revealed peaks of a unique PTH-derivatized amino acid appearing at cycles 4 and 7 (Fig.  11), thus identifying Glu-28 and Glu-31 as sites of modification. The unique amino acid derivative peaks concordantly appeared, corresponding to decreases in Glu-28 and Glu-31, and similar peaks were not detected at other cycles containing Glu or Asp residues. Almost equal amounts of the novel amino acid derivative were detected at cycles 4 and 7 for both forms II and III. Interestingly, an additional unique PTH-derivatized amino acid was detected for Glu-28 and Glu-31 in Ugi form IV, 3 suggesting two types of adducts (Fig. 9). Taken together, these results identify Glu-28 and Glu-31 as two major sites of EDC/ GEE modification. DISCUSSION We have demonstrated that EDC/GEE modification of Ugi protein resulted in the selective adduction of specific glutamic acid residues that inactivated Ugi activity. Ugi form I represented unmodified inhibitor protein, since it did not exhibit any neutralization of charge, increase in mass, or reduction in inhibitor activity. Ugi form II contained a single adduct per protein, defining the location of the most highly reactive amino acids. We determined that 82% of Ugi form II adducts were located within the peptide sequence Leu-Pro-Glu-Glu-Val-Glu-Glu-Val-Ile-Gly-Asn-Lys-Pro-Glu-Ser (residues [25][26][27][28][29][30][31][32][33][34][35][36][37][38][39]. Of the five glutamic acid residues that reside in this sequence, only Glu-28 and Glu-31 were significantly modified and in roughly equal proportions. The preferential modification of 2 out of 19 carboxyl groups in Ugi may be facilitated by the closely spaced pairing of Glu residues (Glu-27, Glu-28 and Glu-30, Glu-31), the repeating Glu-Glu-Val sequence, and the solvent-accessible ␣2-helix of Ugi (24). Selective carbodiimide-mediated modification is observed when acidic amino acids are clustered within a polypeptide sequence, creating a negatively charged local environment (33)(34)(35). A secondary site of modification was detected on peptide A6/7M 1 (residues 61-84) but represented only ϳ18% of the total Ugi form II adducts. Ugi form III was shown to contain two acyl-glycine ethyl ester adducts, and 65-70% of these protein molecules had adducts at both Glu-28 and Glu-31 sites. The remaining molecules of form III Ugi apparently possessed a single adduct on both peptide A2 and A6/7, since the amount of peptide A2M 1 (25%) nearly equaled that of A6/7M 1 (37%). The type, location, and distribution of each adduct on Ugi forms IV and V became more difficult to exactly define due to the heterogeneity of modification. We suspect that the second PTH-derivative associated with form IV peptide C3 represents an N-acylurea adduct formed by an internal rearrangement of acylisourea (36).
The results presented in this investigation provide evidence that Glu-28 and Glu-31, located in the ␣2-helix of Ugi, play an important role in promoting stable Ung⅐Ugi complex formation. Involvement of these two residues was implied, since 82 and 90% of Ugi forms II and III contained at least one adduction in the ␣2-helix sequence. Thus, little room is left to explain the 42 and 67% reduction in Ugi activity by another modified site. However, we cannot discount the possible involvement of minor adducts in influencing the stability of some complexes. We speculate that both Glu-28 and Glu-31 contribute to the formation of a stable Ung⅐Ugi complex. This proposal is reinforced by recent x-ray crystallographic studies of the human and herpes simplex virus type-1 uracil-DNA glycosylase⅐Ugi complexes (4,22). Both structures reveal that residues of the Ugi ␣2-helix and adjoining ␤1 strand dominate the interface with the en-zyme and provide sites of interaction between the two proteins. The hydrophobic face of the ␣2-helix in conjunction with the ␤-sheet forms a hydrophobic pocket that surrounds the conserved Leu-272 active site loop of human uracil-DNA glycosylase (4). This pocket appears to be stabilized by Ugi Glu-28 and Glu-31 contacts with conserved residues that are shared between human and E. coli Ung (3,4). These interactions involve the Ugi Glu-28 carboxylate forming a pair of hydrogen bonds with the Ser-247 (Ser-166 of E. coli Ung) backbone amide and side chain O-␥ (4). In addition, Glu-28 O-⑀1 also forms watermediated hydrogen bonds with the human His-268 backbone amide and Ser-273 O-␥, corresponding to E. coli Ung His-187 and Ser-192, respectively (4). By the same token, Ugi Glu-31 forms a salt bridge with Arg-276 of human uracil-DNA glycosylase (Arg-195 of E. coli Ung). Thus, EDC/GEE modification of either Glu-28 or Glu-31 might be expected to disrupt important protein/protein interactions and destabilize the complex as observed in this investigation.
Several lines of evidence suggest that the interactions between Ugi Glu-28 and Glu-31 with Ung influence the formation of a tight Ung⅐Ugi complex without dramatically impeding the association of the initial precomplex. Steady-state kinetic experiments that monitored the binding of Ugi forms I-V to F-Ung demonstrated that neither modified nor unmodified Ugi freely dissociated from Ung once the final complex was achieved. However, the structure of these complexes differed since binding of Ugi forms II-V resulted in reduced levels of maximal fluorescence quench. We observed from competition binding experiments that unmodified Ugi preferably formed a complex over the modified protein species. Furthermore, unmodified Ugi was observed to replace modified Ugi (forms II-V) from a preformed complex. These properties of Ugi forms II-V were revealing, since unmodified Ugi maintains an irreversible Ung⅐Ugi complex (19). Thus, we conclude that Ugi forms II-V were defective in achieving an irreversible association with Ung. Observations indicating that poly(U) promoted dissociation of modified Ugi from complex and that preformed Ung⅐Ugi (form II-V) complexes displayed uracil-DNA glycosylase activity also support this interpretation.
The results provide new insight concerning the structural and functional relationship of the Ugi ␣2-helix in complex formation. Previously, stopped-flow kinetic experiments indicated that the Ung/Ugi association involved a two-step mechanism composed of a docking and locking reaction. The docking reaction was shown to involve a rapid pre-equilibrium in which Ung and Ugi associated to form a reversible precomplex (19). It is likely that the transition between the docking and locking steps involves an isomerization reaction. This could be achieved by a conformational change in Ung, Ugi, or both proteins. A comparison of the crystal structures of free human and herpes simplex virus type-1 uracil-DNA glycosylase with each respective enzyme in complex with Ugi indicates that only minor changes occur within the enzyme tertiary structure upon complex formation (21,37). A similar comparison between the NMR tertiary structure of unbound Ugi with the crystal structure of Ugi complexed with either the human or herpes simplex virus type-1 enzyme reveals that significant structural changes occur in Ugi (4,22,24). Direct evidence that Ugi undergoes a conformational change upon binding to Ung was provided by NMR results (24). The heteronuclear multiple quantum correlation spectroscopy spectrum of free [ 15 N]Ugi compared with that of the Ung⅐[ 15 N]Ugi complex clearly indicates that many Ugi residues undergo a conformational change as a consequence of forming the locked complex. One of the most dramatic rearrangements appears to involve the location of the Ugi ␣2-helix. In solution, the ␣2-helix is extended away from FIG. 11. Amino acid sequence determination of the Ugi form II C3 peptide. A histogram shows the relative amount of a unique PTHderivative detected during each cycle of Edman degradation of Ugi form II C3 peptide. This novel derivative appeared with a retention time of 15.54 Ϯ 0.05 min between standards identified as Arg (ϳ15.17 min) and Tyr (ϳ16.13 min). The amino acid sequence determined for the C3 peptide that overlaps with the A2 peptide is indicated. the ␤-sheet core of Ugi (24), whereas in complex it is folded across the ␤-sheet, allowing Glu-28 and Glu-31 to interface with the DNA-binding cleft of uracil-DNA glycosylase where Ugi Glu-28 occupies the same spatial coordinates of a DNA phosphate group in Ung⅐DNA molecular models (4,22). We speculate that movement of the ␣2-helix, at least in part, transforms the reversible Ung⅐Ugi precomplex to an essentially irreversible locked state. This interpretation is consistent with the observation that Ugi forms II-V are capable of successfully conducting the docking but not the locking interaction.
The proposed conformational change in the Ugi ␣2-helix provides an attractive explanation that unifies the biochemical, kinetic, and structural data. However, several issues remain to be elucidated regarding the role of Glu-28, Glu-31, and other residues in the Ung/Ugi interaction. First, what is the relative involvement of Glu-28 and Glu-31 in facilitating the locking reaction? Second, does charge neutralization of these carboxyl groups alone bring about inactivation of Ugi, or does EDC/GEE modification sterically hinder ␣2-helix positioning? Third, what other amino acids play a role in the docking and locking interaction? Site-directed mutagenesis and additional protein structural studies should reveal the function of various Ugi amino acids in the Ung⅐Ugi complex.